The use of AI for clinical care has started to find its feet, with medical imaging and triage being some of the first areas where ordinary people have felt the impact. Various startups have also latched on to the idea of using AI to solve for bottlenecks in heathcare. The hype around these tools has started to become something more than hot air: we are starting to see tangible results.

The optimism is well-deserved: healthcare has, for lack of a better way of putting it, been stuck in the 20th century. Don’t get me wrong, the 20th century was a massive upgrade over the centuries that came before. As legendary as Galen was, I would prefer to be diagnosed by someone with more modern sensibilities and training. However, healthcare has been rate-limited by a very critical node in its chain. The doctor. While doctors have been helped by tools which boost their ability to more granularly understand a sick patient, very few tools have aimed to reduce their cognitive load. Improvements in information technology have led to the elimination of the typewriter, the introduction of the electronic spreadsheet, improvement in calculation, the smartphone, and many other conveniences our ancestors would have been astonished by.

But for doctors, information technology’s impact on their work has been to give them better eyes and ears. A sick person time-travelling from the 1950s would marvel at our diagnostics, but she would fundamentally be cared for in a manner not dissimilar to her own time. By a doctor at her desk.

Until now. The advent of AI does not threaten to do to doctors what the word processor did to the typist. The disruption staring doctors in the face is akin to dictation software vis-à-vis the keyboard. While some people prefer to dictate their thoughts, a large number of people will still type everything out the old fashioned way. Similarly, some doctors will still prefer to read X-rays on their own. But we may be headed for a future where computers help the majority of doctors read X-rays instead.

AI in mental health

The role of AI in mental health, on the other hand, is more nuanced than a cursory glance might suggest. Diagnosis in the case of, say, heart disease is much more algorithmic in comparison to a diagnosis of autism. While the specific parameters may vary, heart disease tends to look quite similar in populations around the world. A doctor who has treated a Caucasian in New York would have little trouble in translating her knowledge to an Inuit patient.

However, mental health poses a whole different challenge. While some diagnoses are objective (the MRI of patients with advanced frontotemporal dementia should have similar tells, for instance), depression may be far harder to get right. A doctor who grew up and received her training in New Hampshire may find it hard to get a grip on the cultural mores of New Delhi, thereby missing signs of depression in her Indian patient.

AI has both natural advantages and disadvantages here. Its natural advantage shines through in places where large datasets make training and analysis easier. On the other hand, large, diverse datasets may lead to tricky situations.

Cognitive assessment and therapy

Cognitive assessment seeks to assess the functioning of the brain in thinking, language, using judgement, and memory. It can be used to diagnose multiple conditions:

• Dementia and other kinds of old age-onset disorders are often diagnosed using cognitive assessments. Their progression is also monitored using cognitive tests.

• Brain injuries can often be diagnosed this way.

• Cognitive impairment caused by malnutrition (such as a B12 deficiency), thyroid issues, depression, etc.

• Learning disabilities and disorders can also be diagnosed and cognitive assessments may be used to create individualised learning plans tailored for each child.

There are many other diagnoses cognitive assessments may help us in. However, a typical problem associated with them is subjectivity in their interpretation. Two doctors treating the same patient may judge a patient’s cognition to be very different from each other, given an identical set of responses.

This may make sense given the very contextual nature of cognition. As the famous dictum goes, you shouldn’t gain the measure of a fish by its ability to climb trees. But the inaccuracy of these processes has been such that the introduction of AI to help in making more objective judgements is being seen as a positive development. From Thakkar et. al. (2024):

Screening of cognitive deficits or impairments and early intervention is currently the most widely accepted strategy to manage a number of psychological disorders. The diagnosis of these is established through thorough assessments, which may also help in understanding cognitive pathophysiology. However, lack of proper standardized screening and guidelines often leads to undiagnosed cognitive impairment which further leads to increased disease progression and cognitive decline.

Automating the assessment and prediction process is the key to timely diagnosis and management. The advent of AI has resulted in automated assessment techniques which improve the accuracy of diagnosis. ML and AI-based approaches like Support Vector Machine (SVM), neural networks and ensemble techniques like Convolutional Neural Network (CNN), AlexNet, GoogLeNet and LeNet5 have yielded some of the best results and accuracies when it comes to the use of AI for the assessment of cognitive mental health disorders.

The evolution and adoption of AI has the potential of leading to early and more accurate diagnosis of cognitive impairment and, perhaps, more effective diagnosis of the underlying cause such as Alzheimer’s Disease and the like. Several startups work in this area:

• Limbic: They offer AI tools like Limbic AccessAI, which claims to speed up assessments by 50%, saving time and improving accuracy for mental health providers. It is used by the NHS, UK.

• EllipsisHealth: This startup uses voice-based AI, called Sage, to check mental health by analyzing speech, helping identify issues like anxiety.

• Kintsugi: They also use voice AI for daily mental health check-ins, helping diagnose conditions like sadness, with over 200,000 users worldwide.

Intellectual and developmental disorders

These include disorders like Down’s Syndrome, AHDH, Autism, etc. In addition to cognitive assessments, AI can be used to analyse patient data to accurately diagnose these disorders. The examination of neuroimaging data, for example, or eye-movement patterns can be used to gauge a number of disorders. Based on this, AI systems have been developed to help physicians, or even ordinary people, in the diagnosis of multiple development disorders.

None of these tools have yet seen widespread adoption. However, that may change with time as AI systems learn and adapt and as news of these products spreads. The primary reason, again, as with cognitive assessments, is to provide more objective and accurate measures of conventional intellect and development.

While AI shows immense promise in this field, it's important to remember that these AI tools are primarily designed to be screening and assistive tools for clinicians, parents, and educators. They are intended to aid in early identification, provide objective data, and streamline workflows, but they do not replace a comprehensive clinical diagnosis by qualified medical and developmental specialists. The ultimate diagnosis still relies on expert human evaluation.

Neurodegenerative disorders

Neurodegenerative disorders are characterised by a gradual reduction in the number of neurons in the brain. The brain may lose neurons from different regions, giving rise to different diseases such as Alzheimer’s Disease, Parkinson’s Disease, ALS, etc.

The oldest form of AI development in this area has been a more traditional and objective approach: interpreting brain scans and EEG data. Various biomarkers can also be identified with AI, which, when combined with other data, can lead to accurate and early diagnosis of various conditions. This includes analyzing cerebrospinal fluid for proteins like tau or beta-amyloid in Alzheimer's, or genetic markers associated with Parkinson's.

However, the holy grail of AI intervention in this area lies in a holistic interpretation of patient biomedical data. AI models analyze historical patient data, genetic information, lifestyle factors, and clinical records to predict disease progression and identify individuals at high risk. This enables proactive interventions, such as starting treatments earlier or planning care strategies. Predictive analytics can estimate how quickly a patient's condition might deteriorate, allowing for timely adjustments.

Impact on emotional and affective aspects

The intersection of AI and humans and their mental conditions has led to the development of emotional AI, where technology has been designed to learn, perceive from and respond to human emotions. The field is quite nascent right now, but has great potential going forward. The technology has many potential uses both in the personal and professional spheres. Emotional sensing, in particular, is an area with many potential benefits, if executed properly. Again, from Thakkar et. al.:

Emotion sensing, a pivotal aspect of Emotional AI, traces its origins to affective computing in the 1990s. Enabled by weak, narrow, and task-based AI, Emotional AI aims to comprehend and interact with emotional states by analyzing a spectrum of data related to words, images, facial expressions, gaze direction, gestures, voices, and physiological signals, such as heart rate, body temperature, respiration, and skin conductivity. The input features for emotion recognition could include facial expressions, voice samples, or biofeedback data, while the output encompasses emotional states used for various purposes. Common machine learning techniques like convolutional neural networks, region proposal networks, and recurrent neural networks are frequently employed for these tasks.

While this may sound like the unlikely beginnings of a 1930s dystopian novel, the technology itself is progressing quite fast and is likely to see mass deployment soon. The advantages of such an approach are many. While areas such as cognitive assessment/therapy and neurodegeneration are seen as medical conditions to be treated, the application of AI techniques for helping out in general mental health may have the potential to be even more transformative. Since the biggest problem with mental health issues is lack of access to care or a wish to avoid social stigma, being able to give people means to access care without being judged might be the push they need to take their mental health seriously.

AI may also be able to help individuals understand their emotional states better. While a certain degree of cultural and situational understanding is required to accurately gauge an individual’s emotional state at any given time, the more rigorously methodological and empirical way in which AI tends to both gather and analyse this data has the potential to give a user a better and more granular understanding of their mental state at any time. Tying this analysis to a service, for example, a guided meditation class based on the data gathered about your emotional state at the time, may help people lead more emotionally fulfilling lives, especially those dealing with emotional dysregulation and/or mood disorders.

Emotional AI may end up with more benefits than merely(!) general mental health. Creative endeavours are very emotion-forward tasks, but ask any artist about inspiration, or the lack thereof, and you will get a cavalcade of thoughts about their recent dry spells. This has inherently been seen as part of the creative process, but artificial intelligence is inherently a different form of intelligence than any other we have seen before. It may have ways to break through creative barriers which ordinary humans may not. The current means of interacting with AI is through chatbots. As AI becomes more integrated with healthcare devices which are always on us (such as smartwatches), the data being collected from these sources may have the potential to inform interventions in this area as well.

However, this area, more than any other, requires deeper thought and engineering. Emotional states are complex topics, and AI-algorithmic biases are more difficult to detect and solve for than many other programming issues. As Thakkar et. al. put it:

Mental health disorders are complex and often involve a combination of subjective symptoms, environmental factors, and personal histories. AI algorithms may struggle to accurately diagnose conditions that require nuanced interpretation of these factors. The human aspects and reading in between the lines of a patient's history may be missed by algorithms. Mental health disorders can manifest differently in different individuals. This variability challenges the development of a one-size-fits-all AI solution and requires algorithms that can adapt to diverse presentations. AI algorithms can produce false positives (diagnosing a disorder that isn't present) and false negatives (failing to diagnose a disorder that is present). AI algorithms may lack the ability to fully understand the context and nuances of a patient's life, emotions, and experiences, which can affect accurate diagnosis and assessment. Many mental health disorders have overlapping symptoms, making accurate diagnosis challenging even for experienced clinicians. AI algorithms may struggle with this complexity as well.

While these are not impossible problems to solve, they require time, effort, and patience. Mental health problems have been medicine’s bugbears for a while now. Hopefully, new technology can lead to new treatments and new gains.

ChatGPT took the world by storm when it came out in 2022. Millions of people thought it was a quantum leap in information processing, generation, and ideation. Hundreds of thousands signed up for a paid ChatGPT account and had a virtual assistant on their fingertips. It was rightly thought of as a Sputnik moment in AI, when the first sparks of consciousness began to slowly light up behind the eyes of formerly inanimate machines.

However, biology and healthcare were a little early to the party. Google released AlphaFold 2 in 2020, which placed first in the Critical Assessment of Structure Prediction (CASP14) that year. The results were seen as ground-breaking and astonishing at the time, even though researchers noted that the accuracy was slightly inaccurate for around a third its predictions. It could also not explain the reasoning behind its structural predictions, which was seen as a major shortcoming.

Fortunately, that was no barrier to its adoption. The AlphaFold paper has been cited more than 32,000 times since it was published in 2021. The medical and biological sciences, while (rightly) careful about accuracy and validity, put great value on utility. This attitude towards AI bodes well for the adoption of new, novel methods of increasing efficiency in both clinical and non-clinical roles, a significant part of which is generative AI.

Generative Adversarial Networks

Generative adversarial networks (GANs) represent a powerful method towards creating new types of generative AI. A GAN comprises of a pair of components which work in tandem to create realistic data and evaluate its authenticity. The two components are a generator network and a discriminator network.

The generator works to create new data samples. It starts off with random noise and trains to create a result which resembles a specific target. In a way, the generator is trying to create a masterpiece by throwing paint at a wall. At first, the generator is very bad at throwing the paint, but with enough practice, a generator can become very good at getting close to the kind of target which we wish to look at.

A discriminator network is the arch-nemesis of the generator. It trains to distinguish between real data and data synthesised by the generator. The discriminator's job is to check the results of the generator and provide feedback, helping it improve over time. Continuing the analogy from before, the discriminator trains to be an art critic on the generator's canvas.

The training process of the GANs starts off with the generator struggling to produce anything realistic and the discriminator being very poor at discriminating. However, repeated iterations lead to the generator improving based on feedback from the discriminator, and the discriminator becoming a better critic of the generator's work. As this back and forth goes on, the networks reach a stage where it becomes difficult for the discriminator to distinguish between real data and data created by the generator. This is called a Nash equilibrium. The generator has learned to generate data which closely matches the target data, and the discriminator is very good at judging between real and fake data.

This process can be difficult: there exists a right balance between generator and discriminator which needs to be hit for the process to make sense of things. The hunt for techniques to improve and streamline this process is a particularly hot topic right now.

Applications of Generative AI in healthcare

This process may seem unimportant for medical technology. After all, what need does the medical field have for generating data? Doesn't medicine have a surplus of data which it cannot handle and/or interpret? Well, yes, and no. While it is true that certain widespread diseases do have more data available than we know what to do with, there are specific diseases called rare diseases where data is scant. GANs have the potential to generate realistic data simulating various rare conditions which can then be used to train diagnostic AI as well as doctors, allowing rarer conditions to more effectively be targeted by practitioners. Similarly, it may be safer to use GANs to simulate virtual patient populations to predict treatment outcomes across diverse social and genetic populations, allowing for a more probabilistic interpretation of the effects of new treatment plans. This is especially important for rare diseases, where insufficient real-world data exists for us to have any idea of the probabilities of various treatment modalities.

Generative AI can also be used to be used to prepare tailored personalised medicine plans. It has shown significant potential in building targeted cognitive behavioural therapy regimens. TP-GAN (treatment planning with GAN) is a framework utilising GANs for automating prostate brachytherapy planning. Generative AI is also used in also assist in diagnosing dental health conditions such as caries and periodontal diseases through image analysis and data interpretation, significantly reducing the cognitive load on dentists and enabling faster onboarding of new staff.

Another way of using generative AI is to use GANs and other technologies to generate synthetic medical data which safely replaces patient information while being able to satisfy laws such as the HIPAA or GDPR. Synthetic data generated this way can be used for training and tool validation. Studies have shown that augmenting real datasets with synthetically created radiographic images using latent diffusion models leads to improved classification performance by diagnostic AI algorithms.

A variety of interesting news stories came out this week. However, building on from last week, it was the Marburg virus and lifestyle disorders which really caught my eye.

Progress on combating the Marburg Virus

The Marburg Virus has been the cause of much concern around the world, with multiple actors working to develop vaccines. However, recent news from Rwanda has been very encouraging. The country has recorded 15 Marburg virus related deaths (out of 66 confirmed cases) with a case fatality rate of 22.7%, which is quite remarkable. Past outbreaks have seen case fatality rates of between 24% - 88%. The past two weeks have seen four new cases, with the trend being that of an outbreak tapering off.

However, it is important to note that this does not mean that the threat is over. The country's health minister has stressed upon remaining on alert and looking out for new cases as and when they come. The country's public health system is in much better shape than its neighbours, allowing Rwanda to hopefully add to the pool of knowledge on combating the virus.

Indeed, the country has built upon its existing strengths to inoculate more than 1,600 people against the virus. The healthcare response has focussed on contacts of known positives and people at high risk. Another 500 contacts of known positive patients are being followed up with to ensure that they receive proper treatment should they show signs of the virus. The country has used a strategy of rapid detection, rapid testing, quick isolation, treatment of positive cases and vaccination, which has proven remarkably effective in slowing further spread.

Rwanda has also demarcated the cave thought to be the most likely origin of the virus and fenced it off to humans. DNA sequencing has revealed that the virus has probably jumped from bats to humans exactly once, with the very first identified patient having a history of being exposed to fruit bats in that particular cave. Rwanda has gone further and marked out a number of caves where human-bat interaction may lead to other such zoonotic events.

Semaglutide for arthritis

A new trial, funded by Novo Nordisk, tries to identify the benefits of semaglutide (sold as either Ozempic or Wegovy depending on whether you take it for weigth loss or diabetes) for people suffering from knee osteoarthritis.

Osteoarthritis is a disorder characterised by the breakdown of cartilage in one's joints over time. It is the most common form of arthritis and tends to affect weight bearing joints like the knees, hips, and the spine. The standard way to treat it is to treat the symptoms. This usually involves lifestyle modifications such as weight loss and increased exercise as well as pain relief medications (opioids in some extreme cases) and at times assistive devices.

However, given the primary observable effect of semaglutide is weight loss, it was thought to have potential to help with alleviating the symptoms of osteoarthritis as well. A double blind randomised placebo-controlled trial was held across 61 sites in 11 countries. Around 400 participants were enrolled in a 2:1 ratio and given either semaglutide or a placebo. All participants were also given counselling and instructions on regular methods of weight loss.

The results were incredibly encouraging for the intervention group. While both groups were able to reduce weight, the target group was able to achieve a much higher reduction in weight as compared to the control group. The target group also reported lesser pain, achieving reductions which were at par with opioids, demonstrating the anti-inflammatory properties of semaglutide. In fact, the results were so effective that many in the target group effectively got treated out of the study.

While these results are incredibly encouraging, it is important to exercise caution when interpreting them. The trial explicitly focussed on those with an extremely high BMI. The mean BMI for study participants was 40.3, which is well into the spectrum of obesity. In addition, it is known that participants tend to regain weight very quickly after being taken off semaglutide, which is a very costly drug to be taken regularly. It remains to be seen whether these benefits transfer to those with lower BMIs and whether the treatment reduces in price enough to become an easy prescription for doctors dealing with these cases.

Involving pregnant individuals in trials relating to rheumatic and musculoskeletal disease

Clinical trials tend to be cautious about recruiting pregnant individuals due to fears about affecting the foetus. This is the ethical thing to do: it would be strange for someone to argue against this in principle. However, the incidence and prevalence of rheumatic and musculoskeletal disorders in younger people (often of childbearing age) is rising. Unfortunately, the mechanics and ethics of clinical trials often work at cross-purposes with efficient diagnosis of these issues during and around pregnancy.

Most studies dealing with rheumatic disorders have historically targeted older populations: these disorders tend to be more common in that age group. Despite the need to test and understand the effect of these drugs during pregnancy, clinical trials in pregnancy are rare, therapeutics are often not validated for pregnant individuals, and pregnant individuals are routinely excluded from premarketing clinical trials. Most data on the effects of new drugs as well as the efficacy of new therapeutics on pregnant individuals come from post-marketing observational trials. Observational studies assessing the bidirectional relationship between rheumatic and musculoskeletal diseases and pregnancy, as well as interventional studies of treatments during pregnancy, are scarce.

The historical bias of medicine against women is well-known. However, new ways to conduct trials for this population which are acceptable to everyone involved, from those conducting the trials to those participating in them. Unless the systems around ethics for clinical trials do not resolve this dilemma and start working towards actively trying to create a set of standards which ensure both mother and foetal needs are prioritised, this will remain a major gap within the healthare system.

Healthcare advances at its own pace, but some stories tend to stand out. This is my list of interesting stories that jumped out at me over the past week.

The race to develop a Marburg virus vaccine

The Marburg virus is a highly pathogenic filovirus, which makes it a relative of the ebolavirus. Similar to the ebolavirus and, more famously, covid-19, the Marburg virus is naturally found in fruit bats. Humans usually get infected by coming in contact with an infected animal (bat, non-human primate, etc.), or coming in direct contact with the body fluids of infected individuals.

While there have been multiple outbreaks of the Marburg virus over the years, the current Rwandan outbreak comes as its public health system already struggles with an mpox epidemic. This is Rwanda's first ever Marburg virus outbreak and there have already been 36 confirmed cases and 11 recorded deaths. While this pales compared to historical cases like Angola's (374 cases and 329 deaths between 2004-05, an 88% fatality rate), the speed with which the outbreak has gained momentum has alarmed scientists around the world. Furthermore, the outbreak has occurred in districts bordering the DRC, Uganda, and Tanzania, which may require a multi-country response.

At least three vaccines are already under development. The most promising one seems to be a candidate led by the Sabine Vaccine Institute in Washington which has been tested in 40 healthy adults in the US and found to safely generate an immune response against the virus. The vaccine is currently being tested in a larger field trial in Uganda and Kenya with a view to being deployed as soon as possible. Another candidate being led by the University of Oxford is currently undergoing trials in the United Kingdom, though it is not yet as far along as the American vaccine. A third vaccine, similar to an approved ebolavirus vaccine, being developed by the International AIDS Vaccine Initiative in New York City is expected to begin production soon.

It is hoped that these vaccines will be used for containing the outbreak in Rwanda. The vaccines are expected to be tested out in the field using a strategy known as ring immunisation, where suspected and expected contacts of the infected individuals are vaccinated. The Rwandan health authority has greater resources than its neighbours, so it is hoped that the outbreak remains contains within Rwandan borders.

Gambling as a public health problem

The realm of public health encompasses practically all individual behaviours. Nearly every behaviour has some health consequence or another. However, gambling is a unique phenomenon which has pernicious effects on individual behaviour which is not fully understood from a public health perspective. A recently published Lancet Public Health Commission makes the case for regulating gambling as a public health policy measure given its similarity to other forms of addiction and its effects not just on the individual but on the family.

Gambling can inflict substantial harm on individuals, families, and communities. Beyond the obvious danger of financial losses and financial ruin, these harms can include loss of employment, broken relationships, health effects, and crime-related impacts. Gambling can heighten the risk of suicidality and domestic violence. Research evidence and first-hand accounts from individuals affected by gambling corroborate the association between gambling and these many and various detrimental effects.

A substantial proportion of harm is suffered by those individuals who fall below the threshold for gambling disorders outlined in the International Classification of Diseases-11 or the American Psychiatric Association's Diagnostic and Statistics Manual-5. Therefore, examining the effect of gambling across the entire spectrum of consumption is crucial. As with other harmful commodities, adverse effects are often felt not just by the person gambling, but also by significant others, families, and friends, and can result in both tangible and intangible costs to communities and societies. Although some harms might be short-lived, others are long lasting and can affect subsequent generations.

The commission conducted a thorough review of published literature and various policy objectives, concluding that around 80 million people around the world may be experiencing some sort of gambling addiction.

Now that number is substantive and difficult to argue against. If that estimate is conservative, we may be looking at the problem too lightly. Given the legality of gambling across countries and the easy availability of online gambling, it may not be long before the problem spirals out of control.

The Lancet recommends restrictions and prohibitions on gambling around the world. The phrasing of the recommendations hearkens back to when bans on smoking were being considered: while smoking has not been banned in most places, restrictions have been growing heavier. A similar approach may lead to generational change when it comes to gambling.

An optimistic view on changing mental health levels

As we gain dominance over diseases that have long been regarded as the scourge of mankind, increasing amounts of ink are being spilled about "diseases of modernity". As Stephen Bush writes in the FT, the human race is suffering from success. We have burned the candle at both ends to take care of diseases once seen as completely intractable and have been remarkably successful at dealing with new problems (COVID-19 being the most prominent example). The result? We're slowly starting to realise that humans aren't exactly built to last. If we don't fall to one illness or another, we tend to more esoteric issues such as diabetes, osteoporosis, or more worryingly, depression.

Bush does not mean to say that mental health issues are any better or worse than other issues, but he does unequivocally make the point that having these problems is better than not having them. Sure, there were fewer cases of depression in the 1800s than there are today, for instance. But fewer people (as a proportion of the population) have known the grief of losing a child, or the acute pain of untreated gangrene. As Bush says,

Just as important, there are significantly fewer causes of what Burton described as melancholia of “disposition”. [...] If, as I think is reasonable, we stick to Burton’s definition that what marks out depression is that it is a characteristic of your resting condition, the average 17th-century person spent an awful lot more time in grief or in pain, and therefore had a lot less time to realise or identify that they were also experiencing Burton’s grievous and common disease.

This does not give us any reason to glorify mental health issues, of course. However, past the darkness of mental health issues lies the distant glimmer of the promised land. Just as diabetes went from something you lived with to something which was solved using insulin and metformin to something being actively prevented by GLP-1 agonists, mental health issues may go from being treated using therapy and prozac to being targeted at the source.

Progressive and good healthcare is all about bringing tools developed in the lab to patients in dire need of good care. However, being able to understand which tools to bring to the bedside, figuring out the challenges, and creating a solution which works for all involved is a daunting project. After all, in theory, theory and practice are the same. In practice, not so much.

A marked challenge faced by tools coming out into the field is being able to recreate the setup in which the tool was designed to be used. A successful example of surmounting this is the ultraportable chest X-ray machine. The concept is simple. Miniaturise an X-ray machine so that it can be carried out into the field and be used for X-raying a patient at their doorstep. Simple and easy. The advantages are obvious. If the patient is unable to come to the X-ray machine, the X-ray machine can go to the patient.

Unfortunately, X-ray machines are, by law, required to be installed in AERB1-approved rooms because of the emitted radiation. The solution? Create an X-ray machine which emits less radiation. Ultraportable machines emit a lot less radiation than conventional X-ray machines.

Once you have an X-ray machine out in the field, the next challenge is interpreting the X-ray. Typically, once an X-ray is printed, it is taken to a doctor, typically a radiologist, for their opinion. The radiologist's interpretation of an X-ray is then looked at by another specialist or a general physician.

As it turns out, good radiologists are not cheap! Radiologists are some of the best earning doctors in the country, and probably around the world. The solution? Get an AI to interpret the results. X-rays performed in community settings will, ideally, skew towards being normal and not show disease progression. If the AI flags an X-ray it can either be reviewed further or the patient can be asked to go to a facility for a full check-up.

Cough Sound AI

Another area in which people have been trying to get technology to the patient has been in the realm of cough sound AI. The concept, again, is fairly simple. If a patient comes to you with a cough, use a trained AI to match the sound of the patient's cough with a probable disease.

There has been remarkable progress in creating these AI models. Reported sensitivities and specificities of around 90% are not uncommon. Unfortunately, field trials of these AIs have not yet been as successful as those of ultraportable chest X-rays.

Why is it that these models are unable to recreate their success in the lab out in the field? Is there anything special about hospitals which makes them hostile to cough sound AI? Turns out there is. The real world is a noisy place. Getting someone to cough in a way which the AI can capture, remove the noise, and give an accurate response turns out to be a major problem.

This would probably not come as a surprise to a seasoned hospital worker. Patients are remarkably difficult to coax into doing the most straightforward of things. Getting them to cough in a way that an AI trained in the lab is able to accurately capture the sound is probably very difficult.

Unfortunately, it's not just the patient who is at fault here. The levels of background noise varies from region to region. Regardless of whether the area of operation is a health facility, a community-based setup, or a door-to-door setup, the programmer who created the AI would not have thought of every situation in which that AI would end up.

This is a place where advanced noise reduction algorithms would need to be created. While these have been created for transcribing, say, phone calls or audio messages, the challenge of accurately removing background noise from a patient's cough is infinitely greater. Not only is the sound of coughs uniform (you or I probably have very different styles of coughing), the sound of the same person coughing during a bout of pneumonia or active tuberculosis may be very different. That is, after all, what the entire idea of a cough sound AI is predicated upon.

Radiologists in the building - or not?

A cough sound AI demonstrates the failure of AI to perform out of the lab. However, there are times when the competence of AI echoes in its absence.

X-rays are a source of information about patients of all stripes, and are often read by all kinds of doctors, not just radiologists. Some of those doctors are still learning to read X-rays. I am talking, of course, of doctors early on in their careers, not those who have had some experience. It is usually these junior doctors who read the largest volume of X-rays in a hospital.

The rate at which AI is progressing may make it a better steward of X-rays than most non-radiologists. While it is too early to call AI a replacement for radiologists, it may serve as a useful aid to harried junior residents concentrating on ensuring proper diagnosis and treatment for their patients. While purists may bristle at letting AI or computers handle something clearly within a doctor's domain, it may end up being better in the long run for doctors to concentrate on tasks where machines are absolutely unable to match up to them.

The rising number of elderly in the world brings forth a set of problems which were unanticipated ere the dawn of the modern era. Bismarck, when he created the forerunner to the modern welfare state, did not imagine it needing to take care of a large number of people in their 70s and 80s. Yet here we are, living in a world in which people routinely live to their mid to late 80s.

This would traditionally not be seen as a problem. Having ancestors live long has historically had multiple advantages. However, with decreasing birth rates and increasingly high standards of living, some parts of the world have been contending with the surprising fact that taking care of elders is not exactly an easy thing to do.

The Netherlands was the first country to introduce mandatory universal social insurance scheme which provided long term care services in a variety of settings. Germany was the first country to implement long term care insurance (LTCI) in the form of social legislation in 1995. It has gone on to become the fifth pillar insurance after endowment insurance, medical insurance, accidental injury insurance, and unemployment insurance. The United States has also gone on to formally enact public LTCI.

Japan's long term care insurance

Japan was the first country in Asia to go down this route. With a birth rate below replacement for the longest time, the country has been dealing with epidemics of loneliness and single elders having no one to care about them. Japan's post service actually provides a service in which children living away from their parents can get a postman to visit and take stock of the situation at home.

The solution was a national policy response to this problem since the year 2000. Japan has public LTCI designed to provide long-term care services to residents based on their individual care needs.

The Long-Term Care Insurance Act in Japan categorizes long-term care services into three main types:

1. In-home care services: These include various services such as visiting, commuting, short-stay, and others that are provided at the individual's home.

2. Nursing-home care services: These are provided by welfare, health, and medical facilities for the elderly who require more intensive care.

3. Community-based care services: This category includes visiting, commuting, nursing-home, and composite services that are designed to support the elderly within their communities.

The primary candidates for LTCI are residents aged 65 years and above. Additionally, residents aged 40 to 64 years who require care due to specific conditions are also eligible. The requirements are divided into seven care levels (support levels 1 & 2, which do not qualify for nursing home service, and care-need levels 1 to 5, which qualify for nursing home service).

The classification is initially done based on the number of care minutes required per day to care for the person. The final care-need level is determined by the Nursing Care Needs Certification Board, which consists of physicians, nurses, and other experts in health and social services appointed by the mayor. The board takes into consideration the notes provided by the assessing officer and the statement from the primary care physician when making their decision. Once the care-need level is assigned, it remains valid for a period ranging from six months to two years. Typically, the care-need level is re-evaluated once or twice annually to ensure that the individual receives the appropriate level of care based on their current needs.

The LTCI law in Japan makes sure that the insured only pays 10% of the costs out of pocket. Out of the remainder, half is borne by the government's insurance policy into which the insured pays a premium, and the other half is paid for by the government.

South Korea's LTCI programme

South Korea's population is also aging rapidly, with the proportion of people aged 65 or older expected to reach one-third of the population by 2050. Alongside this, social changes such as increased female participation in the labour market, a rise in the number of elderly living alone, and increased public demand for governmental responsibility in caring for the elderly have begun exerting their demands on Korean society.

In response, the Korean government introduced the public LTCI program in July 2008, aiming to provide in-kind benefits for daily and social life activities for the elderly, either at home or in LTC institutions. Benefits are made available to individuals aged 65 and older, as well as those under 65 with debilitating conditions, subject to an eligibility test conducted through a national care need-assessment system. The system is very similar to Japan's.

Research has found that the most likely users of this programme were traditionally underserved people such as the elderly, women, people who live alone, and people living in rural areas. The programme is also effective: people in the programme have lower out of pocket costs across the board as compared to those who do not and most of them come close to maxing out their monthly benefit limit.

China's public LTCI programme

Another country to launch an LTCI programme was China, where the one-child policy has led to a fairly large skew in the demographics towards the elderly. While the primary caregivers in Chinese society have always been immediate family, with each married couple often needing to care for two sets of parents and four sets of grandparents, institutionalised policy needed to be made to plug this gap.

China launched its first pilot LTCI program in 2016, covering 15 cities. In 2020, the central government expanded the pilot projects to an additional 34 cities, bringing the total number of pilot cities to 49. Despite these efforts, challenges remain, such as the lack of community and home-based care services, the need to expand insurance coverage, and the importance of diversifying funding sources.

However, there was no central authority governing these plans in China. Guidance provided by Beijing was only that: guidance. Every province had autonomy in creating and enforcing its own LTCI programme. So for example, while the guidance recommended a reimbursement rate of 70% for LTCI, this varied wildly across the various provinces where the programme was operationalised. Programmes in most provinces are designed to alleviate both financial and caregiving burdens faced by families, but are varyingly successful in either dimension.

Is public LTCI the answer to taking care of the elderly?

There are significant upsides to having public LTCI. The physical health of those covered by good public LTCI has been seen to show a marked improvement over comparable cohorts elsewhere, especially if they have received good home care. There is also a significant increase in financial security amongst those taking advantage of these programmes. A small positive effect on the labour market has also been noticed, likely due to caregivers having more time to spend in economically productive endeavours.

However, there some significant downsides too. The presence of LTCI in a country increases the use of institutional healthcare and the intensity of LTC sought. In other words, if you give people a benefit, people will use it regardless of whether they require it or not. The Korean system, for example, has no gatekeeping or care management system. As long as the beneficiaries can afford the (relatively small) co-payment, they can choose whether they would prefer institutional or home-based care leading to increasing strain on the health system.

China is also facing a similar problem. Although its LTCI programmes have led to a reduction in "social hospitalisation", which is a trend of seeking in-patient care for minor ailments to substitute for long-term care, the growth of institutional care over home-based or community-based care has been seen in all provinces. The programmes also do not address the needs, preferences and values of older people and also have spotty coverage in rural areas.

Implementation of such programmes in low-income countries such as India and Thailand would require careful thought about mitigating these issues. While the benefits are many, the major challenges are resource utilisation and sustainable funding. These countries have few public resources to spare, and legislating something like a LTCI would require major restructuring of the healthcare system. Financing such an endeavour is something with which countries such as Germany and Japan have grappled for years, with both countries increasing contribution rates and creating a public "population reserve fund". It would be difficult to say what such a system would look like in, say, India.

Ensuring coordination between the LTCI sector and tertiary healthcare would be another stumbling block. While India in particular has a good track record in certain health verticals (the national TB elimination mission, the Pulse Polio programme etc.), their integration with the wider healthcare system is spotty at best. These systems also clash with traditional ways of thinking about the elderly in these countries: Korea has faced heavy challenges with deeply embedded familialism in the society. There is no reason to think the same would not be true of countries such as Thailand, Indonesia, India, etc.

However, the biggest challenge would be countering fraud in such a system. Public systems have a way of turning into rent-seeking structures in many low and middle income countries (LMICs). The cannibalisation of limited resources by those who may not need them purely for the motive of seeking rent is quite common across LMICs: a public LTCI would be very vulnerable to it.

A solution might be to utilise digital identities to minimise theft and rent-seeking. It may not be 100% there, but it may go a certain distance in reducing fraud and increasing efficiency, especially in India, where the India Stack has the potential to allow for more transparent and quicker checking of identities, and combined with the one's abha number, may allow for quicker and more direct referrals to LTC institutions and/or for the provision of home care. A focus on home care over institutional care might be another way for LMICs to go forward.

Sources

• https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2024.1252817/

• https://www.sciencedirect.com/science/article/abs/pii/S0168851013001061

• https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913-020-05878-z

• https://www.jstage.jst.go.jp/article/ace/6/1/6_24001/_html/-char/en

My latest project has me working with Tertiary Care Facilities in New Delhi. Which is, not to be unfair to them or my work, very interesting, but at the same time, it’s incredibly frustrating. They are places doing some of the most thankless work for those most in need while being perennially understaffed. That deserves a lot of credit. At the same time, that perennial understaffing leads to lots of problems in operations and management.

The data from these locations is amazing

Public tertiary care facilities tend to have large numbers of patients. It is something people in India experience quite closely if they’ve been to a public hospital, but the sheer scale can be quite mind-blowing. The AIIMS nephrology OPD saw nearly 100,000 visits in 2021. Extrapolating that number to the entire OPD tells us that AIIMS treated close to a million people in its OPDs alone. Sure, some of them are repeat visits, some of them are now inpatients, but still, the number ought to give anyone pause.

And this is, of course, before looking at other public hospitals such as Safdarjung Hospital, Lok Nayak Hospital, or Dr RML Hospital. Not only do they provide tertiary care to patients but all these hospitals are teaching hospitals as well. People from all over India flock to these places to get tested, diagnosed and hopefully cured, giving these hospitals an unparalleled look at the disease profile of a modern Indian.

The systems are battle tested

The systems set in place in these locations are all battle-tested. They have to be, given the number of patients who go through them. That’s not to say they don’t have bits that can be improved: being battle-tested does not make them immune to criticism. However, I do not have to wonder what’ll happen in case a public hospital in Delhi gets overwhelmed. The systems tend to be overwhelmed all day. That is their default mode of functioning.

The hospitals are overflowing with people of all ages and social backgrounds. And when I say overflowing, I mean overflowing. They’re literally sitting in the hallways and spilling out onto the sidewalks.

The same cannot be said about the major private hospitals. Their prices tend to make them prohibitive for the common man to walk in through the door and demand treatment, unless of course he or she has an insurance plan in place. However, they too have many many patients queuing outside the doctor’s door waiting for a consultation.

Prioritisation is always a problem

However, this set of characteristics make these systems fairly brittle. While overwhelming them is difficult, changing them also, by necessity, is difficult. They have been hyper-optimised to function in a particular manner. They function so as to ensure that their existing mechanisms do not break down under patient load. This often leads to these hospitals reaching short-term optima at the cost of long-term optimisation. Asking a large public hospital to introduce a new set of tests for a population will always remain a herculean task because the sheer quantity of resources they would demand for the same.

This problem of prioritisation extends to resources beyond the physical. A doctor’s time tends to be one of the scarcest commodities in the country. India has one of the lowest number of physicians per capita, with 6 physicians per 10,000 people as per 2023 numbers. Asking a physician to allot a minute more than what they currently do becomes a major ask, with most hospitals asking you to justify the need.

The problem of a minimum viable solution

This style of functioning forces you into thinking about minimum viable solutions. Can you replace a CAT scam with an X-ray? Can you put in some test which can be performed by an orderly instead of requiring a doctor? Can you bring in AI? Will your doctor trust the results the AI gives him or her?

And more importantly, it forces you into rethinking your problem set. Is my problem set actually worth it? At whom should I target this? Does the average person visiting a public hospital in Delhi really require the solution I’m peddling? Does my solution solve a problem which this population actually cares about?

Answering these questions in the backdrop of some of the world’s busiest hospitals is a challenging task. But onwards and upwards! These questions are worth answering, so the only way is forward.

Following the path of many developing nations, India's population of overweight and obese people has been increasing steadily. It is estimated that 14% of the world's overweight people live in India, which is beginning to catch up to the fraction of the world's population living inside its borders (17.5%).

The Indian state and healthcare system has historically been focused on a very different set of diseases. Communicable diseases such as polio, malaria and measles have been on top of its agenda. At the time the Indian state came into being (and for many years afterwards), this focus was warranted. Communicable diseases have been a bugbear of tropical and semi-tropical regions for a long time. Any investment into this area would have had the best cost/benefit ratio in the short to medium term. To make matters more complicated, this had to be done while simultaneously creating a public health system which benefitted all. That the country made any progress at all is a minor miracle.

This led to the neglect of other aspects of healthcare, especially diseases of the non-communicable variety. This was not a major factor in the case of obesity, because obesity and overweight were not major problems in the country for a long time. Unfortunately that is now starting to change. Obesity has doubled in the country between 2005 and 2015, and the problem is not limited to urban and educated areas of the country. Rural areas have been seeing similar gains to urban areas.

The transition to a better life

To understand the reasons behind what is going on in the country, Aiyar, Dhingra and Pingali (2021) looked at data from the Indian National Family Health Surveys held in 2005-06 and 2015-16. They modelled various variables in order to extract a causal chain for what is going on.

Some baseline figures before we delve into obesity. Fertility rates are close to replacement levels across the country. In other words, India's population has begun to plateau. The consumption of alcohol and tobacco has both decreased, as has the frequency of physical exercise such as walking. The average Indian watches more television than she ever did, and more people own cars and bikes than ever before. People are also eating a more diverse diet across the country. This ties in with the greater socioeconomic prosperity observed across the country. The average Indian is also older and greyer than she used to be.

Somewhat unsurprisingly, the authors find that obesity incidence is linked to the state of the nutritional transition a family is in. What exactly does one mean by nutritional transition? A nutritional transition occurs when the kind of nutrition a family gets changes due to a rise in socioeconomic stature. So the transition from subsistence agriculture to working in a garment factory may be the start of a family's nutritional transition because the family switches from eating food it produces to processed goods bought in a market. In our example, this process reaches its end around the time when a child completes university and joins an IT company. The family might make multiple switches in food sources during this process, all of which fall under the term "nutritional transition."

What was somewhat surprising to me, though, is that the incidence of obesity follows a bell curve. The beginning of a family's nutritional ascendance is associated with an increase in obesity, which begins trending downward again during the later stages.

As it turns out, there is literature on this topic which gets into detail about this phenomena. It is mainstream thought that during a country's nutritional transition, obesity risks increase amongst poorer populations and decrease in richer populations. This is caused by a transition from labour-intensive work in agriculture to more sedentary work.

Lest one believes that this is limited to men, the number of women working in these sedentary jobs also increases. The opportunity cost of women not participating in the workforce tends to be too high for them to not shift to doing some work, be it sitting in shops, manufacturing something in factories, or more service-oriented work in general. This gets combined with relatively cheaper high-energy density processed foods, leading to an increase in overweight incidence.

The rich tend to avoid this through greater dietary diversity, more leisure time, and increased physical exercise. To understand this phenomenon, a conceptual model was developed which formalised the factors leading to an "obesogenic" environment. This environment is created due to certain biological, socioeconomic, cultural and transportation factors.

The biology of obesity

Aiyar, Dhingra and Pingali add to that model by adding risks arising from reproductive stress and the age at which the socioeconomic transition occur. Reproductive stress can be thought of as the strain placed on the body due to childbirth. Quite obviously, this only applies to women. The more the number of children one gives birth to in a given time frame, the greater the reproductive stress on the body. The greater the reproductive stress, the lower the chance of obesity.

The other biological factor is age. BMI is positively correlated with age. The older you are, the higher your BMI is, generally. This can be due to either no change or an increase in nutritional intake coupled with a decrease in physical activity.

Both these factors couple together as well. Typically, younger women tend to have greater reproductive stress during the early parts of a socioeconomic transition. However, as the socioeconomic transition progresses, the average age at which women have children increases. Fertility also decreases. Women begin to space their children out more. Access to contraceptives leads to fewer accidental pregnancies. This leads to a decrease in reproductive stress, causing a corresponding increase in BMI. As one can expect, being unmarried is correlated with being overweight.

The role of technology

Several technologies contribute to obesity with the motor vehicle and the television being the most prominent examples. The impact of these technologies is very evident during the early part of this transition. This is because people tend to have limited free time at this point and they prefer to use it for relaxation. At this time, the short-term effects of getting some relaxation time trump optimisation for long-term health benefits. Sitting in front of the TV and relaxing is one of the first things a family starts to do as their nutritional transition begins. Since this is a nutritional transition, it is accompanied by a shift in diet to high energy density foods.

This is also the time at which appliances like gas stoves, washing machines and microwaves start popping up in these homes. The influence of these appliances is typically felt amongst the female members of the household, leading to a greater incidence of overweight amongst women as compared to men.

On the other hand, the acquisition of motorised transport tends to play a greater role for men, at least initially.

Behaviours that generate obesity

Coupled with technology are behaviours the authors classify as obesogenic (generating obesity). These behaviours included watching television, binge eating and binge drinking. Smoking is an anti-obesogenic behaviour. At the beginning of a nutritional transition, both men and women are more likely to smoke, less likely to binge drink, and less likely to binge on fatty foods.

This starts changing as the nutritional transition chugs along. People start smoking less, eating more fatty foods, and binge drinking. As the transition reaches its tail end, one sees this behaviour start trailing off as individuals begin prioritising long-term health goals over short-term fulfillment.

Interestingly, one finds dietary diversity weakly correlated with a reduction in overweight amongst rural women.

The role of socioeconomics and health environment

During the earlier stages of a nutritional transition, overweight incidence is associated with higher socioeconomic status. This is especially pronounced in Indian households, because women tend to face higher barriers to food access than men. When socioeconomic conditions ease up, they see a greater increase in obesity.

As the nutritional transition continues, following the bell curve, we see individuals start to make the shift to the prioritisation of long-term goals over immediate benefits.

However, overweight indicators tend to be highly correlated with development indicators. Low-productivity agricultural states are at the earliest state of the nutritional transition. States with high agricultural productivity are somewhere in the middle. Rapidly transforming areas are at the last stages of their transition and tend to be rapidly urbanising. In India, because most states are agricultural, education tends to be positively associated with obesity.

It is not just development indicators which affect one's weight. Women from minorities tend to have greater incidence of overweight, and women from scheduled castes and scheduled tribes (SC/ST) tend to have lower rates of overweight incidence. While the authors do not speculate overmuch about this, it may be inferred that minorities may be further along the road of nutritional transition than majority families, and SC/ST families tend to not be as far along this road.

All these factors combine with genetics, over which one has little control, and a person's intrinsic and mental health to produce a colourful picture.

The mosaic of obesity in India

One of the most interesting points highlighted by the authors is the difference between rural and urban citizens, and between men and women. Rural women have seen the greatest increase in overweight incidence between the two rounds of the NHFS analysed in this study. However, urban men have seen a greater increase  in overweight and obesity incidence compared to rural men.

Why does this discrepancy exist between overweight incidence in men and women? Part of the reason is that obesogenic factors contribute more than socioeconomic stressors for men, while it is the other way around for women. The authors do not speculate overmuch about why this might be so, but certain dots have been provided which can be connected. Women tend to be more marginalised than men in terms of access to resources and food.

The discrepancies between urban and rural women are largely explained by the reduction of significant reproductive stress amongst rural women. Urban women already had low reproductive stress, for fertility rates were already near replacement rates in urban areas.

The one factor which remains a significant factor for all groups is community incidence. If you live in a community with high overweight incidence, you are more likely to be overweight yourself. This cuts across all lines, whether they be of gender, caste, or the urban-rural divide.

A recent paper by von Haaren and Klonner compares two different conditional cash transfer (CCT) schemes being offered by the Indian government for the purposes of increasing maternal and infant welfare. The Janani Suraksha Yojana (JSY) was started in 2005 and focuses on increasing institutional delivery, and the second scheme, the Pradhan Mantri Matru Vandana Yojana (PMMVY) which incentivises a much broader set of behaviours, was started in 2011.

The authors find that the PMMVY was able to incentivise a number of actions, such as the increase of complete immunisations, the increase in the delay between subsequent births, increased contact with health workers, and an increase in other immunisations like the one for measles. Unfortunately, its coverage was less comprehensive than the JSY's. At the same time, it did not have any adverse effects on fertility rates, which increased under the JSY.

The JSY was designed as a one-time payment contingent on either institutional delivery or by a certified professional at home. This caused an uptick in institutional deliveries, as was expected. It also resulted in an increase in the number of babies being breastfed and the number of families in contact with healthcare professionals. However, what was not expected was the increase in fertility rates (though it has been seen in numerous conditional cash transfer schemes across the world), and the lack of any reduction in either maternal or neonatal mortality. It also resulted in a substitution away from private healthcare providers.

The PMMVY is a second generation benefit scheme which has been implemented after learning from the first generation. It incentivises a larger number of behaviours and features a number of training programmes. It covers a longer period of around 9 months around delivery and additional supply-side financing. During its pilot phase, eligibility for its cash transfers were given for the first and second child, but post-2017 when the programme rolled out to everyone in India the benefits were confined to the first child.

"Consistent with IGMSY's incentives, we find that polio, DPT and BCG vaccinations increase. As an indirect effect, measles immunizations, which are administered well beyond the period covered by the scheme, also increase. As a consequence, complete infant immunizations increase by 9%. We also document two positive side-effects: mothers of once eligible children report 14% more contacts with the government health system 3 to 4 years later. Moreover, there are no adverse effects on fertility, and birth intervals increase by 11% on average and by 17% between the first two parities, which are covered by the program. On the other hand, similar to JSY, we find no robust evidence of increased breastfeeding or gains in health outcomes, albeit some of our results suggest improvements in breastfeeding duration, child mortality and weight-related outcomes for both children and mothers." - von Haaren and Klonner

At this moment, this is one of the first papers to compare different Indian conditional cash transfer schemes. The authors point out that it seems that the PMMVY was a better designed scheme as compared to the JSY, for it had fewer side effects. For example, as mentioned earlier, fertility rates did not go up after its introduction. It also prioritised training new mothers about breastfeeding and nutrition leading to improvements in child mortality, neonatal weight, and weight-related outcomes for mothers. However, the authors state that the effects were extremely small, which tells us that more work needs to be done to design these programmes properly for Indian conditions.

Some misc. interesting papers

The effects of pollution on neonatal health: It is common wisdom that excessive pollution in developing countries leads to excess mortality. However, a group of researchers decided to look at whether the relationship really holds or not. They wished to see whether it was the pollution in developing countries which led to babies dying or whether it was the limited ability in populations to manage the effects of pollution which caused excess neonatal mortality. The study looks at the case of Hong Kong, which is a high-income city with a high concentration of particulate matter (PM). The effects of PM concentration on birthweight, the number of low birthweight babies born, and neonatal mortality were observed across 2001 - 2019.

It was found that while seasonal variations, even marginal ones, in PM concentration caused measurable changes in birthweight and in the number of low birthweight babies, it had no measurable effect on neonatal mortality. The authors also performed a comparative analysis to check whether their results hold up or not. In the authors' own words, "In light of our conceptual framework, these comparative results suggest that vulnerability to particulate matter exposure may be more important than particulate matter exposure itself in explaining differences in marginal mortality damages across countries. To be clear, we do not assert that baseline particulate matter exposure does not matter but rather that marginal mortality damages may be linear in exposure."

Identifying high-cost users of CKD: An exercise was described by Sowa et. al. in which they attempted to identify differences between two different populations suffering from Chronic Kidney Disease (CKD). The 90th percentile in terms of cost of treatment (High Cost Users, or HCU of in-patient care) and everyone else. The study specifically focuses on the first year of patients coming into specialist care in Australia. They found there to be no difference between the populations when it came to gender or ethnicity (indigenous vs White Australian). However, the median age of the latter population was higher than non-HCU patients. They were also, unsurprisingly, more prone to being at stages 4 and 5. Diabetic neuropathy was also more common in the HCU population, as was the presence of co-morbidities such as diabetes, cardiovascular disease, and hypertension. It was also seen that HCU patients were far more likely to have been admitted after an episode change. They were also much more likely to be readmitted within a 30-day period.

In addition, "HCUs were at an increased risk of admissions due to issues of the nervous system (RR: 1.94; 95% CI 1.74 to 2.15), factors influencing health status (FIHS) (1.92; 1.74 to 2.09), circulatory (1.24; 1.14 to 1.34) and respiratory system (1.2; 1.03 to 1.37). HCUs were at a lower risk of admissions caused by digestive system issues (0.71; 0.56 to 0.87) or other MDCs (0.73; 0.66 to 0.81). Key FIHS that distinguished HCU from non-HCU involved the use of rehabilitation procedures (Z50, 47.4% vs 20.8%, respectively) and problems related to medical facilities and other healthcare (Z75, 13.5% vs 1.5%, respectively)."

Cost Benefit analyses of returning incidental findings in genomic research: An interesting question posed to advocates of whole genome sequencing and exome sequencing is, "Is it worthwhile reporting incidental findings (IFs) back to patients?" In other words, should doctors tell patients about findings of dubious clinical significance, or findings which may have clinical significance but were not the ones for which the doctors were actually screening? There is a rich debate in the community about this question, but there is no definite answer. Marx et. al. sidestep ethical questions and focus on performing a systematic review of economic analyses surrounding this work.

The results they get are inconclusive. This is not a question pursued with much rigor in the literature. The authors wished to focus on Africa, and thus found only seven studies which satisfied all their criteria. The authors state that while five of those studies actually provided economic evaluation results, they do not give enough evidence to judge the practice of sharing IFs. Marx et. al. aren't the only ones to come to such conclusions. An older scoping review found that the costs and benefits of sharing IFs aren't being considered in most studies dealing with whole genome sequencing.

However, one finding which the authors highlight is that the costs of sharing IFs probably vary with the primary health condition affecting a particular patient.

"For instance, Christensen et al. found that the lifetime average health care cost of returning IFs for a cardiomyopathy patient is around $90, compared with $325 for a colorectal cancer patient, and $440 for each healthy individual. Christensen et al. results show that omitting or limiting the types of IFs results that patients in cardiology and primary care receive can save on average $69 and 182, respectively. Hart et al. add to the discussion that returning IFs to patients will increase healthcare resource utilization and the cost on average is $421 (range $141–1,114) up to 1-year post-result return of the IFs." - Marx et. al.

The authors go on to state that the community seems to be settling on the answer it might not be worth sharing IFs with patients, especially if there is no diagnostic support available to them. There are often mixed psychological reactions to sharing genetic data ranging from absolutely ignoring results to being very scared of them. Sharing incidental findings might further complicate such reactions.

Including survivor costs in economic evaluations: The economic effects of the saving of a life are typically not taken into account by studies conducting economic analyses from the societal perspective. However, recent American guidelines have recommended taking these costs into account. Kellerborg et. al. say that the inclusion of these costs has an impact on ICERs (incremental cost-effectiveness ratios) as well. They state that the inclusion of these costs leads to an increase in the numerator of the ICER, making it higher. This increase is the lowest for people from a lower socioeconomic status which makes interventions targeting them look more economically favourable. At the same time, people from a higher socioeconomic status both lived longer and had a higher quality of life at all ages, therefore their quality adjusted life expectancy was higher than that of people from low socioeconomic status. While the effect of including these consequences is substantial, the effect across educational groups is much smaller. They relate this to two issues:

• People with a lower educational background enjoy a lower quality of life than those with a higher educational background

• People with fewer years of education tend to form single-person households in their later years, implying higher consumption costs compared to people in a multi-person household

The major finding from this study, however, is that the differences in consumption seem to be mitigated by the concomitant differences in quality of life and household size.

This blogpost is the first in a series exploring a framework created by John Kingdon which deals with policymaking in a public setting. The book was first published in 1984, though we will be using the second edition which was reissued in 2010. It is a veritable mammoth and seeks to deal with all the participants, stakeholders and processes which go into making public policy in the United States.

Although the framework itself is extremely America-centric, it does bring out certain features of the public policy-making apparatus which are common across jurisdictions. In elucidating the elements of the Kingdon framework, we will be trying to tease out these common features and seeing how they apply to health systems.

The first part of the framework focuses on the components of the policymaking apparatus. Specifically, this post will focus on "the players in the game", as Kingdon calls them. These are the people inside the government who have a very direct effect on creating policy. Some of these should be obvious, such as the President and Congress, but we will also be looking into the "deep state": career bureaucrats and political appointees. Some of these components comprise what is known as the administration, and so we start with...

The President

The President of the United States occupies pole position in the hierarchy of policymakers. If the President proposes a bill, it goes right up there in the front of the queue. The President has the prerogative, which (s)he often exercises thoroughly, of setting the agenda not merely for the executive branch, but also for the Congress as a whole. This has been seen in several administrations, such as the Carter Administration's priority of reducing hospital costs, President Reagan's "Morning in America" and Reaganomics, the Clinton Administration's welfare reforms and focus on reducing the budget deficit, the Bush Administration's focus on what became known as the "Bush Tax Cuts", and so on.

This, of course, does not mean that the President manages to get his or her way in all cases. External events often stymie the goals of any President (the COVID-19 pandemic being an epic case in point) and more than anything else, the President is merely one person. The alternatives to the President's proposed policy agenda are usually not crafted by him or her personally nor is the President able to guarantee the final outcome. The example given by Kingdon is that of President Carter's healthcare policies:

The Carter initiative set the agenda for the health-related congressional committees, whose members spent substantial portions of their hearing and markup time over three years on this subject. But the administration proposal, which provided for caps on inflation rates and limits on capital expenditures, was only one of several options considered. People on the Hill also considered defeating all proposals and doing nothing, providing for a voluntary effort by the hospitals with the imposition of government controls if the effort failed, and a longer-range strategy including reimbursement reforms. One Hill staffer said, "This particular piece of legislation is not near and dear to the heart of anyone on Capitol Hill. The ones that favor it, favor it only out of a sense of obligation and duty, out of a sense that they must do something about cost inflation, and out of loyalty to the White House." This sense of obligation and loyalty could carry the president's initiative as far as agenda status, but could not restrict the range of alternatives to his proposal that were seriously considered.

We see similar issues with President Biden's agenda about pumping money into infrastructure. Had the Democrats got their way, the bill being passed through Congress would have been bigger and would have been targeted at what is called "soft infrastructure" as well. America is in dire need of investment into infrastructure beyond roads and power grids. Unfortunately, the President isn't guaranteed to get their way all the time.

Why does the President occupy such a paramount position in the policymaking process? The most important reasons for this are institutional. The President has the power to veto bills which he or she does not want passed. The only way to override a Presidential veto is by getting two-thirds of the Congress to agree to override the veto. But as one of Kingdon's interviewees put it, "You couldn't get two thirds of Congress to pass the ten commandments." The President also has the prerogative to hire and fire people. If you aren't aligned with the President's agenda and work for him, there's a good chance you will be fired from your job and replaced with someone more aligned to the President's way of thinking.

Another institutional reason is that the Executive is a much more unified decision-making body as compared to the Legislature. Once the President makes clear what his or her agenda is, that agenda is likely to carry the day. In contrast, the Congress has 535 different members with 535 different agendas. While some of these agendas might be aligned with their fellows, there is no overarching mechanism to compel Congress to act together. It becomes much easier for Congressmen and Congresswomen to define their agendas as being for certain parts for the President's agenda or against certain parts of it.

A third, non-institutional reason is the President's command over public discourse. As Theodore Roosevelt put it, the Presidency is a bully pulpit. It's much easier to get your point of view out there and convince the public to rally behind it as President. This is, of course, not guaranteed to work, but when leveraged correctly can be a powerful mechanism to pressure the government into getting something done. A Congressional staffer pointed this out to Kingdon in an interview by talking about the vast flood of emails Congressmen and women receive when the President goes on the offensive in a news conference.

The final way in which the President can direct policy agendas is through his or her involvement in its setting. Policies which the President does not care about and does not personally get involved with tend to not get as much attention from the executive branch as those where the President makes it their mission to work on. News conferences and public opinion is one thing, but repeated phone calls, letters, and direct reporting tend to matter a lot more.

There is a huge caveat to this, though. The President's bully pulpit works in their favour when the President is popular. A less popular president would be unable to use their office the same way as a popular one. It has been seen that members of the Congress often try to set their own agendas against that of an unpopular President and align their agendas with a popular one. This tells us that while a President might be very powerful, they aren't all-powerful.

The Presidential Staff

The Presidential Staff is of vital importance to the functioning of the Executive. However, unlike the President himself (or herself), the staff is less involved in setting the agenda and more in setting up the alternatives. It has been noticed that the President and the Cabinet members are generally the ones who set the tone of the agenda, and the staffers are those who have to set up the alternatives.

Once the agenda and alternatives are set, the Presidential Staff engages in talking to the relevant stakeholders and establishing a detailed negotiating position. They flesh out the bones of administrative proposals and engage in discussions with everyone else who needs to be consulted. In other words, while they don't really set the agenda itself, they are the ones who actually do the work.

Political Appointees

This group comprises of cabinet secretaries, undersecretaries, heads of bureaus, administrations, and other such agencies. Traditionally speaking, these people are thought to have a lot of clout in the day-to-day running of their agencies the setting of their agendas. However, modern, conspiratorial thinking tends to portray them as being subsumed by the desires of the agency they are supposed to be running. In other words, they are thought to be prisoners to the "deep state" itself.

Interviews conducted by Kingdon seem to suggest otherwise. The interviewees tended to mention political appointees as some of the most powerful actors in the policymaking system. They are mentioned more often than the President or their staff when talking about powerful actors in the policymaking apparatus. Their business is not to generate new issues or ideas. Their work is to "elevate" ideas which are already there. The ideas they need to work on already exist and are floating around in the executive branch. All they need to do is to take an interest in them. A neglected idea can often gain a lot of traction in case a political appointee takes an interest in it.

This is not to say that they have the power to trump the President's wishes. The President will always win in a difference of opinions between the President and a political appointee. The political appointee may try to convince the President of his or her point of view, but the President will always have the last word. A public disagreement between the President's agenda and the wishes of a political appointee have the potential of completely derailing a political appointee's career. The reverse can also be true: the President might find it embarrassing to be seen to be in conflict with a political appointee. This is the reason why such spats are typically not publicised.

Finally, these political appointees tend to have shorter tenures than the President's two terms. This leads to them trying to get things done as soon as possible. In the words of one interviewee about a particular Cabinet Officer, "He wants to be the firstest with the mostest, and it doesn't matter if it's the bestest."

Civil Servants

By contrast, while there is a lot of speculation and conspiracy about civil servants due to their relative permanence and expertise, Kingdon's interviews did not find them to be a major force in policymaking. As a civil servant said of his appointed superior, "You go in and tell him X and he wants to hear Y. You go back again and you tell him X and he says he wants to hear Y. The third time, you finally conclude that you'd better say Y."

This might be surprising, because it has often been argued that it is not possible for a political appointee to control their subordinates. The counter-argument to that is that Kingdon was trying to understand the power civil servants hold in the process of setting an agenda. The development of alternatives and the actual implementation of policies is where civil servants have a lot of influence. Political Appointees generally tend to define the agenda for their sphere of influence, but they leave the drafting of the exact policy and its alternatives to civil servants. In the words of a congressional staffer, "Bureaucrats are not so important with respect to the generation of ideas, but they're critical with respect to their professional advice and consultation in pursuing approaches which we have generated. For example, with manpower, if we want the definition of underserved areas to include Pittsburgh or other urban areas, technically how can we do that? The bureaucrats are in a position to tell us how."

Civil servants also work on proposals which they agree with for extended periods, waiting for a sympathetic superior to pitch it to. This showcases their relative powerlessness in the practice of actually setting the agenda, for which they are critically dependent on their political superiors. However, their strength comes from their longevity. Since they tend to have worked on specific problems for a longer time than their politically appointed superiors, they are the ones often turned to for advice on how to go forward. Kingdon's interviews highlight the fact that this does not mean their advice is always followed. Kingdon does not go into detail about the frequency of political appointees "turning native", as it were, but it is noted that it goes both ways. Political appointees often start looking at things the way the bureaucracy wants, and they often don't. After all, civil servants aren't the only sources of expertise for political appointees. They often have their own networks which they turn t0 for wider ranging advice.

Capitol Hill

Congress is seen as the location of the people's representatives, and thus definitely does have an extremely important part to play in the policymaking process. Kingdon's interviews showed that interviewees held the Congress as being more important than any single part of the Executive, though less than powerful than the Executive in aggregate. However, since there are 535 different members of Congress, it might be difficult to say that Congress has a unified agenda.

That isn't to say that individual members of Congress do not have any power to set the agenda. Senators, especially those in prominent positions, can often leverage public position and their own standing to set the agenda in ways which is usually only possible for the President. The example given by Kingdon is that of Senator Long, who, as chairman of the Senate Finance Committee had scheduled mark-up sessions for national health insurance which prominently featured catastrophic medical bills. This caused it to rise up in the policy agenda of the day. In the words of an administrative source:

He says that it was when he scheduled his markups that the administration was really galvanized into action. And I think there probably is something to that. You know, in one of the markup sessions he told the story about this, and he said, "It's a little like the game you played when you were kids when you count to ten and then say, 'Ready or not, here I come. ", He says that's what he did to the administration; he counted to ten and he said, "Ready or not, I'm going to mark up a bill."

Senator Long's interest did not mean that anything would pass, but it was able to move the policy agenda forward in many important ways. And his example is not an isolated one. Senator Bennett pushing a proposal of Professional Standards Review Organizations onto the agenda, Senator Magnuson's interest in health manpower maldistribution causing the Health Service Corps to come into being, and Senator Ted Kennedy's forays into a number of policy areas are merely a few other examples of a fairly widespread occurrence.

Senators and Representatives are discussed very often by Kingdon's interviewees because of their ability to define both the agenda as well as the alternatives. While most people in the administration have the power to set either agendas or alternatives, few have the power to influence both in the way members of Congress do. The President cannot pass a law if there is no support from Congress. Policy proposals tend to undergo major changes in order to pass Congress.

Why is this so? First and foremost, the Congress derives its power from its legal authority. Major changes in domestic policy require new legislation, which requires Congressional approval. Another reason is that Congress tends to be a breeding ground for future Presidential candidates. No one wishes to go afoul of a Senator who may run for President in the future. The third is that Congressional information tends to be a blend of all types. In Kingdon's own words:

Congressional information is not the type of expertise that comes from undertaking a detailed study or being on an operational firing line. Rather, its major characteristic is a blend of the substantive and the political, the academic and the pressure group information, the bureaucracy and the constituency. Members and staffers are exposed to an impressive variety of information-studies, administration arguments, leaks, interest group pressures, complaints from the districts, concerns of constituents-and the combination of these various communications is different from the perspectives of others in the system.

This information is also generally very freeform and informal. Unlike the Executive, where everything happens in defined channels, information in Congress tends to go about in very informal channels and one gets access to it through the strength of one's relationships. And finally, members of Congress often have more longevity than the President does, and this extends to their staffers as well.

Kingdon also identifies three goals which motivate Senators:

1. Satisfying their constituents to get re-elected. This is a major consideration, and getting things done is seen as better for this than doing nothing. This often extends to make themselves known to a greater number of potential voters in order to prepare for a future Presidential run. However, since this is a fairly sensitive thing, Congressmen and women tend to be fairly conservative in their choices, ducking hard choices and backing safe proposals

2. Enhancing their intra-Washington reputation. It is in the interest of a Senator to carve out their own niche and be seen as a heavyweight in their area, so as to become a go-to person for that topic and to perhaps lead committees for the same

3. Achieving their own conception of a good set of public policies. In other words, Senators tend to try to follow their consciences and ideologies

Kingdon has a subsection on Congressional staffers as well. However, it is sufficient to say that they have a role similar to that of political appointees and civil servants in the Executive. They manage the details, they often come up with ideas, alternatives, and specific provisions of legislation. However, everything they do has to be approved by their political masters.

In the next post, we will look at the role of non-governmental organisations, such as interest groups and the media.

The global HIV community, in 2014, adopted a target principle called 90-90-90: the diagnosis of 90% of people living with HIV (PLHIV), antiretroviral therapy (ART) for 90% of those diagnosed HIV–positive, and viral suppression in 90% of those receiving ART by 2020. These goals were revised up to 95-95-95 by 2030. However, we remain quite a distance from where we ought to be at this time. We are currently at 81-82-88, which gives us an overall viral suppression rate of only 59%, a long way from where we ought to be.

Why is this so? What is the point at which people stop being part of the system? Ehrenkranz et. al. (2021) point out that most people who get diagnosed with HIV do not tend to drop out before the start of ART because of a push to begin ART on the same day as diagnosis. Thus it becomes important to understand where people actually decide to disengage and what the best method for getting these people back into the system would be.

Typically, most models which aim to show the system for dealing with HIV are linear with a person entering through one end and exiting at the other due to death or loss of follow up. However, that is not how these systems work: people can leave at any point. People tend to enter and exit the system at various points. As the authors point out:

Routinely collected medical records in most settings do not adequately document this phenomenon. In one notable exception, South Africa’s Western Cape Province undertook a pilot that used unique patient identifiers and digitized routinely collected point-of-care HIV test results to assess testing and restarts. Within their intervention site, 51% of people who tested HIV–positive had previously been diagnosed, and 71% of these had previously started ART. In other words, more than one-third of HIV testers had previously been on ART. This information was used to highlight the need to redirect resources from expansion of HIV testing to improved focus on continuity of care.

Any model which aims to properly model an HIV intervention ought to take this into account.

In order to do that, the authors propose the creation of a cyclical cascade model.

This model has four stages with the addition of a disengagement step as well. The primary advancement in using such a model, simple as it may sound, is that one can now model disengagement at any step. This can make it easier (institutionally) to understand the stages at which patients tend to have the greatest tendency to drop out as well as the frequency of a patient's drop-out at each stage. Further it also allows one to understand the effect of each stage on on return (when it happens) and the implications of loss or reentry at each stage on future stages. The authors do note:

Clinical outcomes, such as viral load suppression, are not explicitly included in the cascade as long-term retention in care, the fourth stage in our cascade, is highly correlated with suppressed viral load, and our intention is to better understand PLHIV behavior as it relates to patterns of engagement, not the biological results of treatment.

Policy Implications

This cascade may be used to distribute resources more efficiently. Different kinds of interventions may need to be targeted at people who disengage at different stages, something this model explicitly enables. It will make it possible to figure out populations most likely to disengage and help their retention in care, figure out people who have disengaged and work towards re-engaging them with the process, and most importantly, identifying sub-populations which are susceptible to dropping out at any given stage. Targeting entire populations without tailoring interventions towards those who explicitly need them is a waste of resources because majority of people who are diagnosed and start ART do not leave it.

Of course, putting this cascade in action will require specialised data from clinical interventions, surveys, and population surveillance. This data will differ from country-to-country and can be used to tailor the cascade according to the needs of the area and the population. Specifically, the integration of electronic medical records can be used to track people who fall through the system (for example people who were previously diagnosed but decided to stop taking ART need not be diagnosed again).

This model may evolve differently within countries. For example, if transportation to the clinic for conducting ART is a major bottleneck that may be added to the model. If it has been found that reintegrating people who were previously in the system is best done by diagnosing them again it would be best if that relationship is added to the cascade. These models will thus change from area to area, country to country and will, after a while, not be applicable to other places without major adaptations. However, they will be extremely valuable locally and have the potential to increasingly streamline existing systems and processes.

Cost-effectiveness of gene therapy for Sickle Cell Disease

Sickle Cell Disease, often confused with Sickle Cell Anaemia, is a group of genetic disorders associated with lifelong complications, a reduced life span, and increased usage of healthcare systems. The clinical manifestation is characterised by damaged and abnormally shaped red blood cells which result in their rupture, the blockage and obstruction of small blood vessels, and other complications resulting from these issues. Severe pain episodes tend to be a common result of sickle cell disease.

Currently, treatment varies according to the age group but typically includes both preventive and palliative care. The most common forms of treatment include vaccination, hydroxyurea, blood transfusion, penicillin, and opioids. Hydroxyurea has been found to be tolerated fairly well in patients of all ages: however, it has many side effects and patients and providers both often skip taking it due to concerns about long-term risks.

On the other hand, there are a few other therapies available. Haematopoietic Stem Cell Transplant is a potentially curative treatment but is typically reserved for those with extremely severe forms of the disease. It's availability tends to be scarce and it remains associated with many ifs and buts, like the possibility of transplant rejection, long-term issues, and various long-term adverse outcomes.

Salcedo, Bulovic, and Young (2021) focus on a new gene therapy targeted at these patients and check whether it falls within current thresholds of cost-effectiveness or not. The treatment involves removing haematopoietic stem cells from a patient, modifying the gene responsible for sickle cell disease to produce an anti-sickling haemoglobin B, and then reintroducing these cells into the patient. Initial trials of this therapy showed substantial improvements within 3 months of it being introduced.

The authors utilised a Markov Model to estimate the incremental cost-effectiveness ratio (ICER) of the treatment. Based on their analysis, they showed that an average patient with sickle cell disease stood to gain 8.5 quality-adjusted life years (QALYs) and 3.7 life years (LYs) over their lifetimes at an incremental cost of $1,196,917 over a lifetime. This assumes a $2.1 million cost for the treatment. Accounting for gender, it was found that this translated to around $146,511 per QALY and $135,574 per QALY for females and males respectively. However, this presumed that a single administration of the therapy would work for all patients. In case 50% of the patients were to relapse in 20 years, the ICER of this treatment goes up to $410,607 per QALY. If the relapse rate were to be 50% in 10 years instead, this increases to $740,058 per QALY.

It was also found that if one were willing to pay around $250,000/QALY for this treatment, then the chance of this treatment not being cost-effective decreases dramatically. The authors further found out that the treatment would be cost-effective 32.7%, 66.0%, and 92.6% of the time at values of $100,000, $150,000, and $200,000 per QALY, respectively.

There are a few limitations of this study:

• The authors assume that the therapy is provided at birth, which might not happen for a variety of reasons even in developed countries

• Their most optimistic assumption is that the therapy is 100% effective. However, until we have more data, this remains an extraordinarily optimistic scenario

• The authors do not consider a societal perspective here and look only at direct medical and healthcare costs. The costs borne by society (caregivers, workplace, etc.) are not considered

• This study was conducted looking at a single specific treatment called Zolgensma

• The price of this therapy may be determined by market prices. This price differential might be enough to tip this therapy into being cost-ineffective

• The acceptance of gene therapy amongst Americans is also unclear. It would be wise to address mistrust in the health system and suspicion towards genetic therapies in general while introducing this treatment option

Using Behavioural Economics to promote advance directives for EOL care

Over the course of a lifetime, individuals make many different choices about their healthcare. One of the most important choices they have to make nearing the end of their lives is that about provision of care towards the end. While an extremely important task, surveys of nearly 800,000 American adults in aggregate showed that only around 30% of them had taken part in this activity. This proportion has changed very little across the first half of the 2010s and numerous legislative and research initiatives have done little to change this.

Spivey, Brown and Courtney (2021) apply the logic of Behavioural Economics to see what kind of messaging makes the idea of advance directives (ADs) stick in the heads of older people. Prior research has found that the most important reason to go for ADs at EOL is to reduce the burden on family members. This trumps financial autonomy as well. It has also been found that an individual with defined ADs is likely to incur lower costs during their EOL times and have a lower chance of dying in a hospital rather than at home. The primary reason why people reported as not having made an AD is because they believed that their loved ones already knew their wishes.

In order to check what may motivate someone to create ADs, the authors look at the idea of framing.

Framing, a behavioural economics technique, posits that the way in which information is delivered matters to how individuals make choices. For example, the likelihood of an individual choosing to have a surgical procedure will differ depending on whether they are told the procedure has a 5% mortality rate (negative or loss-framed) or that it has a 95% survival rate (positive or gain-framed). Prior studies of framing related to health have shown that positive frames are more effective for prevention, while negative frames are more persuasive for detection.

In order to test the effects of framing on people, the authors recruited older people who had not yet defined their ADs. After an initial survey, they were shown an informational video about AD. A survey was conducted after this, and then they were divided into groups and shown one of three videos:

1. There is a 1 min and 28 s negatively framed video that features a brother, sister and sister’s husband in a hospital waiting room trying to make an end-of-life care decision for their father who is incapacitated and who did not have an advance directive. This family is in grief and worries over the decisions they should make given they don’t know their father’s wishes (we will refer to this video as negative framed video)

2. Another video is a 1 min and 7 s positively framed video that features the same family in the previous video. This time their father had an advance directive, so while in grief, they are relieved their father’s wishes are carried out (we will refer to this video as the positive framed video)

3. The final 1 min and 37 s video uses a social norm to frame the decision to make an advance directive. This video features two friends discussing a third friend’s experience when her husband was incapacitated but had no advance directive. The friends conclude an advance directive is a good thing and after doing research have decided to get one, too

The participants were asked survey questions after watching the video. The participants were then shown the other two videos (in different orders for different groups) and asked survey questions after each one.

The authors find, not surprisingly, that negative framing of this message is correlated with efforts to get more information about ADs. For people who had not sought information about ADs prior to the study, positive framing was seen to have a small positive impact on the approximate change in creating ADs. The authors seem to conclude that negatively framed videos might be better at getting people to get more information about ADs, and positive messaging may be better at getting them to actually create ADs. Social norm messaging was also seen to be positively correlated with the creation of ADs.

The authors also found that positive messaging around ADs was seen to be more convincing the older a participant was, while negative messaging was more convincing to younger participants. On the other hand, people with a chronic disease were best targeted using social norm messaging.

Policy Implications

It is important for messaging to be tailored to the audience: this is true for COVID-19 vaccination messaging as well as for creating advance directives for EOL care. It would be a good idea to incorporate training in behavioural economics for both doctors and nurses in order to get them to understand how to convince or nudge people towards any specific health policy. This information also ought to be considered when creating government campaigns about specific diseases.

Customer trust in public and private organisations in storing COVID-19 data

A huge problem in combating the COVID-19 pandemic today is consumer trust. Consumer distrust manifests itself in a lot of ways, be it vaccine hesitancy, refusal to trust authority figures on the actual reach and impact of the pandemic, or a refusal to share data about one's own health.

Grande et. al. (2021) take a look at the last issue. They administered a survey to a group of US respondents to see their levels of trust in various organisations when it came to keeping track of COVID-19 data. The respondents were asked to rate nine types of organisations:

• University Researchers

• State Health Department

• Local health department run by a your county/city

• Health Agency at the federal government

• Health Insurance Company

• A company that makes digital thermometers

• Apple

• Google

• Facebook

It was found that respondents actually had the strongest confidence in university researchers followed by State or local health departments. Corporations were uniformly ranked as the least trustworthy. The authors also state:

In multivariable models estimating the proportion of consumers that were at least somewhat confident in public health agencies, there were large differences by political ideology of the respondent—71% of self-identified liberals expressed confidence in state health departments to protect digital health information compared to 57% of moderates (p<0.001) and 45% of conservatives (p<0.001). These differences were similar for local public health departments and federal health agencies. We did not observe greater confidence among those with a family history of COVID-19 or those living in a higher incidence county. We did not find significant differences between Black vs. White respondents. Hispanic respondents reported higher confidence than non-Hispanic respondents (61% vs. 56%, p=0.04). Differences by income, geography, and age were small and generally not statistically significant.

The authors also report that the only variable they find really affecting these responses are the political beliefs of the respondents: no other variable seemed to make a difference.

Policy Implications

While the polarisation of society and its effect on vaccine skepticism and mask-wearing is well-known, it seems these effects extend to cover the stewardship of data as well. The agencies responsible for collecting data and analysing it would do well to alter their messaging and communicate properly with all sections of society. Prior research has shown that consumers are very particular about data protections and privacy: it might be a good idea to be very clear in the communication of the same to the general public. In addition, the usage of services by private companies may have the effect of pushing people away from sharing their data: it would be a good idea for governments to either build this data collection and analysis capability in-house, or bring it within the ambit of the HIPAA or related laws. That would go a long way towards reassuring people about the safety of their data.

The effect of Customer Directed Care (CDC) on Mental Health in Australia

Australia has been promoting the use of a model called Customer Directed Care. In the words of Ausmed:

Consumer-directed care (or ‘CDC’), refers to a ‘self-directed’ healthcare model in which the client is afforded the right to full autonomy in all decision-making related to that care. Historically, these care decisions would be made by the care team, however CDC empowers the care recipient by aiming to improve their health literacy, so that they can play an active role in their care pathway (Ansara 2014). All home care packages in Australia are currently mandated to be based on a consumer-directed care model, fulfilling Standard 1 of the Aged Care Quality Standards: Consumer Dignity and Choice (My Aged Care 2015).

Specifically, this model provides greater choice to the customer: while it remains the care team's responsibility to give recommendations to the customer and enable them to make choices about their own healthcare, the final responsibility lies with the customer himself or herself. The care recipient has the right to delegate as much or as little of the decision-making to care providers as they wish.

Specifically, Australia follows this model in all home care packages (HCPs) as well. Home Care Packages are meant for older people who wish to receive healthcare without being shifted out of their homes into an external care home. Unfortunately, the mental health effects of the model in this context have not been studied quantitatively before. Tran and Gannon (2021) set out to fill that hole in public policy literature. They analysed data on 1780 individuals aged 65 years and older across 11 iterations of the Household, Income and Labour Dynamics in Australia (HILDA) Survey. To measure mental health, they used the MHI-5 scale.

The MHI-5 scale is constructed from the following items in which respondents answers to how often in the past 4 weeks they have (1) been nervous, (2) felt so down in the dumps that nothing could cheer them up, (3) felt calm and peaceful, (4) felt down, and (5) been happy. Answers to each item were coded on 1 to 6 scales where 1 represents “all the time” and 6 represents “none of the time”. All scores across each item, are summed up, then subtracted by 5 and divided by 25 to derive a 0-to-100 mental health scale.

Specifically, they find that older adults tend to have significantly worse mental health outcomes after the introduction of the CDC model (which was introduced in 2013). To make sure that this actually corresponded with the uptake of Home Care Packages, they looked at mental health scores across regions and found that regions with higher uptake of HCPs tended to have worse mental health outcomes.

In addition, the authors find that this tends to vary across educational levels as well, which they posit might be because education corresponds to financial literacy, which may cause less educated people to take on greater stress when it comes to managing their care. Similarly, access to care might be harder in non-urban areas causing more stress to rural residents. And finally, the gender disparity seen in stress levels (men are less likely to report good mental health than women) was posited by the authors to be because men tend to live in more rural areas than women in their specific sample.

The principle of attractive mates developed because humans wish to give good genes to their offspring. However, there are important differences between the sexes when it comes to recognising good partners. In the words of Whyte et. al. (2021):

Females are more selective, not only because their maximum fecundity is time limited but because choosing poorly increases the long-term opportunity costs of reproduction (internal gestation, ongoing lactation, and disproportionate maternal investment) and reduces the probability of offspring. Their mate preferences should thus reflect characteristics or traits that can compensate for disproportionate maternal investment and ensure offspring survival and reproductive success, especially with respect to economic proxies for resources and/or increased paternal investment such as educational level, intelligence, and income. In fact, research has shown females demonstrate far more stringent preferences than males for mates with good earning potential or higher education, particularly during the years of peak fertility. Males, in contrast, need only invest the time taken to copulate, which paucity of paternal investment implies the favouring of mates whose genetic fitness guarantees a maximum chance of offspring survival and reproduction.

The assumption humans go with is that certain characteristic of age, attractiveness, symmetry, etc. seem to imply a lower likelihood of disease or illness. However, this manifests itself differently in different ways in males than in females. It has been seen that males seem to value facial cues more in long-term mating contexts over short-term ones, in which they seem to value bodily cues more. This seems to be because females have oestrogen-dependent facial features (lips, cheeks, jaw line) as well as bodily features (waist-to-hip ratio and accentuated gait).

Aesthetics, Resources and Personality

• Aesthetics is an extremely important part of the way humans judge each other in different social contexts. Since females have a shorter period in which they can have children, it has been found that males place more of a premium on aesthetics than females

• Resources tend to be an interesting part of partner selection. Since females have to invest resources into childbirth and the associated activities, they tend to prioritise males who can make up for this shortfall. The Australian Bureau of Statistics has found that earnings tend to peak for women in their mid-30s, but they tend to peak for men in their mid-40s or later. Since earnings are associated with greater age in males, that leads to women going for older men. This also works in step with the ability of men to father children remaining fairly constant with age

• Personality traits have started to become more important in mate selection as well. As the authors put it, "In addition to increasing pair bond strength through parental investment, such positive externalities in mate choice may also reinforce reciprocally altruistic behaviour between mates, increase complementary production in the household, promote kin selection towards genetic relatives, and increase the chances of long-term mate retention." The importance of personality traits seems to be increasing in developed countries where the differences between males and females when it comes to access to resources has been narrowed through legislation. In this study, the authors aim to see how the relative importance of difference personality traits changes with age

Results

Women rate the following around 9 - 14 points higher than men do:

• Age

• Education

• Intelligence

• Trust

• Emotional Connection

It was found that women rated all nine factors a few points higher than men did. Some other interesting results were:

• Males regard attractiveness and physical build as the most important factors, while women regard age as being more important

• There is little difference between the ratings given by men and women to income: it's not considered a very important point in choosing a partner

• While women on average rate openness higher than men, men rate it relatively higher. That is to say that men do not value it as much as women on average, but they value it relatively more than they value most other characteristics

There is considerable variation with age, as this series of graphs shows. Typically, preferences for attractiveness and age decrease over time, but the male preference for physical build remains constant and that for females actually increases with age. On the other hand the 60+ female cohort shows a strongly increased preference for education. But the preference for intelligence decreases over time across the sexes. The relative importance of emotional connection seems to remain constant across genders, and the importance of income, emotional connection and trust seem to increase with age (young people don't care about them as much).

Conclusions

On the surface, one may make the observation that for the population sampled, and compared with males, females care more about a greater number of characteristics when considering attractiveness in a potential mate. Such findings lend confirmatory weight to previous research findings and broader historical evolutionary theory that predicts that females tend to be choosier than men.

The authors also report that sex differences (across ages) are the least for those who do not rate aesthetics highly.

Here, we find a consistent statistical sex difference (males relative to females) that decreases linearly with age for aesthetics. The opposite is true for resources and personality, with females exhibiting a stronger relative preference, particularly in the younger cohort of our sample.

More highly educated females express a higher relative preference for aesthetics, and more attractive females exhibit a higher relative preference for personality. We also find absolute differences for females with offspring, who place more emphasis on personality, whereas males with offspring report this trait as less important.

The ethics behind resource allocation in a pandemic

When it comes to the allocation of scarce resources among people in the context of absolute scarcity, there have been many principles proposed:

• Maximising Individual benefits

• Treating people equally

• Maximising social benefits (instrumental value)

• Priority given to the sickest

• Lifecycle principle

Of course, it is well-recognised that ethical values are not islands: there is no single value which can define the outcome of a resource allocation problem. One needs to use a combination of them in order to be as fair as possible to as many as possible. Currently, the model being used in Portugal uses the following criteria in the order given:

1. “Maximising health benefits” (efficiency consideration – Priorities should be set according to patient’s survival/prognosis)”

2. “Severity of health condition” (used whenever patients have similar prognosis)

3. “Instrumental value” (frontline health professionals deserve priority whenever patients present similar prognosis and severity of the health condition)

4. “Random selection” (used whenever patients have similar prognosis and similar health condition severity)

Pinho (2021) set out to validate whether the Portuguese public also believes in these ideas as well as ethicists do. In general, it was found that respondents gave the most importance to the principle of maximising benefits, as do ethicists. However, it was found that younger generations cared less maximal benefits and were more concerned with responsible stewardship of resources. The second most important criteria identified matched up with what most ethicists believe to be important: the severity of the health condition. The third criteria, surprisingly, was age. People preferred younger patients getting access to more resources. Instrumental value ended up receiving the least support as a criterion for resource distribution.

Prognosis was seen to be the most overriding concern in the minds of respondents. It was seen that if confronted with a situation in which one of two patients had to be chosen, the first with a good chance of survival and a severe impact on the rest of their life if not given treatment, and the second with a bad chance of survival yet minimal impact upon their lives if they survive, people would choose the former. They emphasised prognosis, not the probability of survival. Saving the sickest does not seem to be the intuition of most people. However, the one exception seemed to be people with a prior experience of COVID-19. They were more likely to give more importance to the severity of sickness.

It was also seen that participants in the survey, regardless of age, preferred to allocate resources to younger people. This tendency was more pronounced among younger respondents, reflecting rational choice behaviour theory. This is an interesting result, because age is typically not seen as a valid strategy for the allocation of resources. It also contradicts the supposition that older generations would put more stock in the "fairness" principle. One possible explanation could be that COVID-19 has been extremely lethal for older people and has generated specific fear for their lives in them. However, regarding the principle of instrumental value:

Our findings indicate that Portuguese participants moderately agree with the idea of giving frontline healthcare professionals priority in access to critical medical resources. Younger respondents, health professionals and respondents who tested positive to coronavirus showed greater preference than older respondents, respondents from other professional occupations and those that never had the disease, for the instrumental value principle. Again, an adherence to the rational choice theory seems evident among health professional’s participants. Empirical evidence on population support for the instrumental value principle is lacking but one study (Krutli et al., 2016) reached results contrary to ours. They concluded that older respondents consider more than younger respondents this principle as fair. Additionally, the fact that respondents previously infected support more the instrumental value principle may be because they felt the sacrifice and the risk that frontline health professionals are experiencing and exposed to.

The lottery principle, on the other hand, was seen to be unfair by almost all the respondents.

Policy Implications

It is important for politicians to make sure that they consider the population's idea of what constitutes ethical resource distribution before attempting to impose a framework on the population. While the Portuguese people, in general, seem to be happy about their national policy, it might be a good idea for the leaders of other countries to attempt to understand these principles. In certain cases, they may conflict with utilitarian principles and those of efficiency, but getting the population on board with such strategies is pivotal to actually getting a generalised national strategy for pandemics in place.

Improving Public Quality Metrics in Hospitals

In the USA,  providing high-quality care is a central element in the mission of a hospital. Further, quality of care contributes to a hospital’s reputation and affects its financial performance. Wernz, Song and Hughes (2021) develop a model to analyse how hospital interventions may affect physicians' CT scan decisions. This particular area was chosen because medical imaging is one place where patients tend to pay a lot of attention to quality. According to the authors:

The hospital interventions that we consider in our model are (1) incentivization, (2) training, and (3) nudging. The model captures the physicians’ CT scan decision process, clinical uncertainties, and physicians’ varying diagnostic abilities.

One important thing to note is that while imaging is important and often necessary to get imaging done to diagnose certain diseases, overuse of imaging as a diagnostic tool is not a good idea. It is costly and increases the chance of misdiagnosis.

Certain points which they make:

• Models using partially observable Markov decision processes (POMDPs) for optimal breast cancer screening account for uncertain patient and cancer states

• Zhang et al. in 2018 as well as 2016 analyzed CT scan related decisions of payers, hospitals, physicians, and radiologists, and the multi-level effects of payment innovations using a multiscale decision theory (MSDT) model. They found that payment innovations, while effective, can have unintended consequences, especially across multiple system levels, and their design needs to be data-informed and the program parameters carefully chosen

• While a few models have accounted for some aspects of physicians’ CT scan decision-making, none have incorporated physicians’ varying ability levels in diagnosing a patient

CDRs (Clinical Decision Rules) have seen to be an effective tool against over-prescribing imaging, specifically same day brain and CT scans. The authors use the #Model developed by Zhang et. al. through the inclusion of CDRs to actually build their own model.

Model

It is assumed that the physicans wish to maximise each patient's health utility, which is denoted by \(\mu_{S,A}\). The \(S \in {b,s}\) state variable describes whether the patient suffers from a brain ailment or a sinus ailment. The \(A \in {B, B + S }\) action variable shows whether the doctor prescribes a brain or a CT scan.

Thus, there are 4 prospects a patient can be have: \((b, B), (b, B + S), (s, B), (s, B + S)\). Since \((b, B)\) is always preferred over \((b, B + S)\) because it avoids unnecessary radiation, thus one can sat that \(\mu_{(b, B)} > \mu_{(b, B + S)}\). Similarly, one can also state that \(\mu_{(s, B + S)} > \mu_{(s, B)}\) because it is always better to get a brain + sinus CT scan in case one has a sinus infection. Of course, the point is that the first inequality always holds, whereas the second is mostly about avoiding treatment delay.

Next in the decision process, the physician assigns the patient a rank as the basis for their CT scan modality decision. As stated earlier, the more convinced the physician is that the patient suffers from a brain ailment, the lower the rank. Conversely, the physician assigns high ranks to patients who appear to be suffering from sinus ailments. To then decide whether a patient should receive a brain CT or a brain+sinus CT, the physician compares the patient rank to a chosen threshold \(\theta\). For \(x < \theta\) the patient receives a brain CT, and for \(x \geq \theta\), the patient receives a brain+sinus CT. The patient then experiences the corresponding health utility, whose expected value the physician seeks to maximize.

Discussion

The authors actually look at 3 different ways in which this can happen: incentivisation, nudging, and training. The authors did this through the use of a decision-theoretic model.

The authors state that a targeted approach for giving incentives might be used (in a world without ethics) by hospitals for maximising their own metrics (OP-14). However, targeting has the problem that it would incentivise the physicians best at doing their jobs the least. Training costs are the highest among the interventions assessed, even though it might be the best. Nudging is seen as, perhaps, the most cost-effective way of achieving what the hospital wants. However, it will probably not be enough by itself and will need to be supplemented with training.

Crowdfunding Pandemic Communication

Most countries have been very bad at co-ordinating their pandemic responses. While the list of issues is too long to list in full, the most egregious issue among these was the delay in compiling a good list of symptoms of COVID-19. It took quite a while before it was properly understood what symptoms even constituted an infection by the novel coronavirus.

Zolbanin, Zadeh and Davazdahemami (2021) point out that the CDC referred to the symptoms of MERS as the basis of its list of symptoms for Covid. The lack of haste in compiling symptoms may have lead to its alarming spread in the US. As the authors say:

First, the cause of the disease was a novel coronavirus, and as a result, the information about the possible signs and symptoms of the disease was accumulated as the virus infected more and more people. Second, the variation in geographical presentation of diseases and the ease of traveling from one corner of the world to another created challenges for the identification of a complete list of symptoms. Third, a lack of international cooperation hindered the dissemination of experiences and sharing of knowledge among the global and national agencies. Fourth, the global and national agencies did not fully utilize the power of social media platforms in obtaining and spreading information about the symptoms of the new disease.

The obvious question, of course, is whether it was even possible for countries to reduce the amount of time required for this exercise. The authors use Twitter to create a symptom surveillance system (SSS) which has the advantage of being able to avoid the question of national pride and its like. This is because people tend to use social media for sharing health related issues long before they actually become a blip on the radar for healthcare authorities. As the authors state:

The utility of these platforms, however, extends beyond states of emergency to allow for public health surveillance and the exchange of health information, including information about illnesses and associated treatments.

An example provided for the issue of communication was:

While there was anecdotal evidence on the effectiveness of hydroxychloroquine in treating some patients, WHO hastily halted trials of the drug based on the results of a single study but soon after restarted the process when it was known that the study had validity problems. This and other examples, such as recommending to not wear masks at the end of March and then advising to wear them in public areas, suggest there is a lot of room for improvement in future pandemics.

The authors were able to demonstrate the face that social media is perfectly able to tell us about these issues as and when they crop up. They did so by performing a network analysis and a time series analysis: the first to see which symptoms were being reported the most, and the second to see the chronology of these reports.

Conclusions

The authors state that there was very good correlation between the actual progress of symptoms as observed and those reported on Twitter. They also state that these symptoms and their progression became evident on English-speaking twitter in during February itself and continued on till mid-March. They posit that checking Chinese and Italian social media might have led to those dates shifting even earlier.

The relationship between indebtedness and subjective well-being

Over-indebtedness is truly a malaise on society: it tends to affect people extremely negatively. Researchers have converged on four features of over-indebtedness in a bid to characterise it:

• Making high repayments relative to income (e.g., households spending more than 30% of their gross monthly income on unsecured repayments)

• Having a high number of credit commitments (e.g., four or more credit loans)

• Being in arrears

• The subjective perception of debt as a burden

Ferreira et. al. (2001) begin by stating some objections with this characterisation. The first two metrics are fairly inflexible. The third does not really try to understand the seriousness of the arrears themselves (which depends on the financial state of the household), and the fourth, of course, is very subjective. The authors thus choose a very distinctive measure for understanding over-indebtedness: people who chose to go to a debt advise expert.

Looking at the literature, one sees that being over-indebted has many specific negative effects:

• Increased risk of suicide and depression

• Poorer subjective health and increased physical illness

• Low sleep quality

• Increased risk of chronic diseases

However, there is little work performed on the effect of over-indebtedness on subjective well-being (SWB). Previous work done on this subject has found a tenuous link between SWB and indebtedness. However, most such studies do not distinguish between indebtedness and over-indebtedness. As the authors point out, there is a correlation between over-indebtedness and both careless consumer behaviour and financial imprudence, often leading to social stigmatisation of the people under these circumstances.

Two different reasons for over-indebted individuals having lower SWB are considered in this paper. The first is that since financial well-being is literally the feeling of being satisfied with one's financial status, SWB is directly correlated to it. The second is the fact that being overly indebted tends to greatly reduce a person's chances of actually achieving their goals. Thus, over-indebtedness might not just cause financial anxiety but also reduce the perception of control a person has over their lives.

The authors utilised surveys in Portugal to get the data they required.

Results

It was seen that over-indebted people displayed lower life satisfaction than non-over-indebted people. Surprisingly, despite their lower satisfaction with their lives, it was seen that over-indebted people seemed to be more optimistic about their futures as they tended to anticipate a steeper increase in their life satisfaction very soon.

The results for emotional well-being also matched up to the ones found for life satisfaction. It was also seen that emotional well-being seemed to deteriorate going from morning to evening. It also seemed to affect patterns of sleep, with over-indebted people sleeping less than non-over-indebted people. A similar result was found for self-reported health.

All of this data pretty much agrees with the data found in literature. The subjects in this study were mostly medium-to-long term sufferers of over-indebtedness. This lends one to believe that unlike many other life-changing circumstances, over-indebtedness is not a temporary phenomenon. Other circumstances which are routinely compared to being over-indebted are long-term unemployment and chronic back pain.

Out of the three variables that constitute SWB, it was found that perceived control (or a lack of it) was found to explain a great deal about the relationship between over-indebtedness and all dependent variables (life satisfaction, emotional well-being, health, and sleep). In the authors own words:

A lack of control over one’s life not only contributes to fully explaining the relationship between over-indebtedness and emotional well-being, but also partially explained the relationship between indebtedness status and life satisfaction. Furthermore, perceived control was also found to partially explain sleep quality and to fully explain reported overall health.

Policy implications

Efforts to combat over-indebtedness exist in many countries. However, most of them primarily focus on reducing the financial issues faced by people. This study posits that not only should policymakers focus on that metric, they should look at factors which can influence an individual's perception of control over their own lives better. A mixed approach consisting of both primary and secondary approaches might better serve the population being targeted. Better mental health may be a driving factor in getting people out of debt and getting them to feel better about their situations.

The LSE–Lancet Commission commission came out with a report on the state of the NHS earlier this week. In the light of the pandemic, the authors argue, the United Kingdom faces a once in a lifetime opportunity to restructure the NHS and reorient it towards facing issues more relevant to the zeitgeist.

There is much to criticise and praise the NHS about when it comes to their handling of the pandemic. The flexibility and sense of duty demonstrated by its staff in working long hours to keep health services running cannot be overstated. Unlike places like India, where government hospitals struggled and people had to pay exorbitant amounts for private care, the presence of the NHS made sure that most people in the UK received quality care without having to pay a lot. The equality of resource access and allocation also made a big difference in the outcomes of patients with chronic diseases (such as chronic kidney disease, diabetes, etc.) who were fairly well-taken care of. And most importantly, the clinical trials of many vaccines in production today (most notably the Vaccitech-AstraZeneca vaccine) were first conducted through the NHS. There is a lot to be thankful about.

But there are many places where the NHS can still use improvement. The high number of deaths per capita represents a deep failure of the health system. The inability of the NHS to increase testing, a lack of hospital capacity, the lack of personal protective equipment (PPE), and a failure to test and trace properly all led to the pandemic getting out of hand very fast. Local branches were unable to get the equipment they needed, labs ran out of testing kits, many testing sites were unable to use the National Pathology Exchange properly, and procurement was not done in a standardised manner. This was patched up very quickly, but the mistakes committed in the early days of the pandemic snowballed as thousands of untested covid positive people were released back into the country.

Of course, most of this can be traced back to inadequate support from Westminster and the devolved national legislatures at Edinburgh, Cardiff, and Belfast. The NHS's funding has been dropping for years and to expect the NHS to be able to perform perfectly after having emerged from a period of long austerity after the 2008 financial crisis would be stretching the credulity of any sane person. Unfortunately, the effects of this reduction in funding can be felt in many other metrics of healthcare measurement as well. The UK has seen a reduction in the rate of improvement of life expectancy as compared to its peers in the G7 and the EU and has seen inequalities grow between richer, more urban regions and rural, more deprived regions. The number of doctors and nurses employed by the NHS is now below average when it comes to high-income countries. The NHS has also not been able to fundamentally change the way it interacts with patients: patient engagement is also firmly stuck in the twentieth century.

The authors' responses to these problems are fairly straightforward at first glance. Increase NHS funding, spend it wisely. The direction of that funding, however, is somewhat important. The authors stress on increasing social funding and integrating it with the NHS. Social care has been a neglected part of the UK budget for a long time: it has been proposed to tie this in more closely with healthcare. Another recommendation talks about increasing access to diagnosis and improving the availability of cheap diagnostic tests. The NHS has already done this for COVID by supplying OTC home diagnostic kits across the country for individuals to monitor themselves. The kits remain a means of screening: if one gets a positive result, a confirmatory test is carried out through PCR-based tests. The authors also recommend tightening up the NHS's workforce strategy in order to retain more workers and train doctors and nurses to fill in existing gaps in the system.

The state of Health Information Technology (HIT) systems also remains a reason for concern. There is widespread agreement that the current HIT systems seem to hinder rather than facilitate care. A good HIT ought to ease data entry operations, allow working with big data to strategically improve healthcare, aggregate data from multiple endpoints, and be easy to use. Current NHS workers seem to be inadequately trained to use available HIT systems, leading to frustration and a desire to not use the ones in place.

Finally, the authors also talk about integrating the different parts of the NHS more closely together in order to make the patient experience more seamless and wholesome. The integration of health and care, as is being done in Northern Ireland and Scotland might be a good place to start.

Voluntary Healthcare Insurance

Xu and Yang write about the pitfalls of implementing voluntary healthcare insurance schemes in a country. In a nutshell, Voluntary Healthcare Insurance Schemes (VHIS) require a population to pay a flat rate regardless of their age, existing preconditions, or any other reason in order to cover everyone else. However, as the name suggests, VHIS are voluntary, so one can drop out of actually being part of them. This causes the structure to become untenable, because there isn't enough money to actually cover everyone.

The authors focus on China and the effect of VHIS in the country. Their findings line up well with prior studies reporting young and healthy individuals tend to shun VHIS because the costs are too high for their risk profiles. The authors also report that the most socio-economically vulnerable parts of a population also tend to shun VHIS (because the cost is too high for them to afford or due to insufficient knowledge or understanding of insurance), which, again, lines up well with previous studies. And finally, people with worse health indicators tend to remain signed up to VHIS in both China as well as elsewhere.

The authors also report that dropping out of VHIS tends to affect the usage of secondary and tertiary healthcare services, in line with what other studies report. This is probably because primary services are much cheaper than secondary and tertiary services. One group which bucks this trend is rich people in rich provinces in China: the authors speculate this may be due to the more developed commercial healthcare systems in these places.

Policy Implications

A careful reading of the paper and the authors' comments recommends that VHIS, if implemented, needs more thought and some state support. It has been suggested previously that it might be a good idea to introduce well-thought out exemptions for the poor (not in the Chinese context, but in an African one), and Xu and Yang come to the same conclusion. Related work by Hall, Hamacher and Johnson in Michigan also suggests that a social safety net, if properly constructed, is better than local private insurers as well as Medicaid. In the context of China, the authors also suggest improving primary care and improving the distribution of healthcare services across China to assure uniform access for all.

However, the greater implication seems to be the introduction of a Mandatory Health Insurance Scheme (MHIS) for poorer and more rural areas. The major issue with a VHIS based on a fixed rate is that people with better healthcare indicators find it harder to justify paying for it given their considerably better risk profile. However, lack of insurance has been associated with decreased healthcare utilisation among people with similar healthcare indicators, which indicates that people fear incurring healthcare costs when not insured. From the point of view of a society, it might be a better idea to invest more in health insurance in order to increase health service utilisation and utility.

Factors behind COVID-19 Vaccine Hesitancy

COVID-19 has to feature extremely prominently in any newsletter talking about health issues. Serda and García sought to understand the reasons behind vaccine hesitancy amongst the population of Chile. This is an important step towards learning how to tailor one's approach towards pro-vaccination messaging. Vaccination campaigns need to target people's existing preconceptions, fears, and the real barriers they face when going to get a vaccine if they are to be successful.

The authors utilise a Health Belief Model (HBM) and apply logistic regression to understand the reasons behind vaccine hesitancy. To quote the authors:

In terms of public policy, the HBM reveals that the variables to be considered relate to perceived barriers, benefits, susceptibility, severity, and cues of actions, among others; in this vein, scarce literature exists regarding the COVID-19 vaccine.

The primary reasons which people cited fro not getting vaccinated were a lack of knowledge of side effects and the extent of risk, a lack of knowledge about the vaccines themselves, and a preference to see other people get vaccinated first. Interestingly, educated people tended to be more prone to rejecting getting a vaccine dose due to lack of knowledge about the vaccine itself or its side effects than less educated people. On the other hand, the reasons which motivated individuals to actually go and get vaccinated were the perceived benefits of protecting oneself and one's family, positive cues from family members, fear of the severity of complications from catching COVID-19, and an understanding that the vaccine would reduce chances of catching COVID and inducing immunity against the disease.

A more interesting thing to come out of this study was the fact that people were more likely to care about potential side-effects than effectiveness. In other words, people are more concerned about the safety profile of the vaccine than how effective it was. A very large number of people expressed his preference, which has major implications for targeting communication policy. Being convinced about the efficacy of the vaccine was also important for most people. The presence of an effective vaccine in the country made a significant difference to a lot of people. Another major factor was encouragement from their social network, or at least no negative pressure. This encouragement did not have to be direct. If a person's social network indicated that the severity of the symptoms of COVID-19 was high, or that the side effects were negligible, they were more likely to get vaccinated than not. Thus people whose family members had already suffered from COVID-19 were extremely likely to get vaccinated.

These results make a lot of sense. Behavioural economics has taught us that humans are not rational actors. We also have a tendency to see small negative probabilities as being bigger than they are, and small positive probabilities as being smaller than they actually are. These factors have to be kept in mind by policymakers when they create vaccination communication campaigns. Policymakers ought to focus on convincing people about the transient nature of side-effects and their lack of strength over talking about the efficacy and the effectiveness of vaccines. Another angle to look at is convincing people and teaching them about the short-term effects of COVID-19 as well as cautioning them about its unknown long-term effects.

Another major reason for vaccine hesitancy, not really discussed in-depth by the authors, was the issue of price. A previous paper by the same authors talks about the willingness of a person to pay for the vaccine, and they found that people in Chile were willing to pay a mean of around $232 for getting vaccinated. That study has been criticised for omitting some nuances: especially that the majority of the population was not very willing to pay the mean price if willingness to pay was analysed slightly differently. Both the paper and the comment are fairly interesting reads if one wishes to understand a couple of different perspectives on the same data. A small quote from that paper, however, caught my eye:

The main reasons for respondents refusing to pay for the vaccine are as follows: the government should pay for the vaccine (44%), the vaccine is not important (16%), I do not have enough money (11%), those who caused the virus must pay for it (10%), it is immoral to pay for a vaccine (10%), and society has bigger problems/I do not want to pay (8%). These results show that almost 90% of the refusal responses are protest responses.

This seems to strengthen the argument that good messaging can lead to a lot of change in vaccination uptake.

No link between the socio-economic conditions of parents and the cardiovascular health of their children

There is some evidence from animal studies that a child's health is often invariably linked to the conditions endured by the parent as a child. Some evidence exists in humans as well, but there is not a lot of literature that explicitly looks at conditions apart from obesity, and very little literature from South Asia on this phenomenon. Mallinson et. al. explore the effect of the socioeconomic status of parents and the incidence of cardiovascular disease in their children. Prior literature looking at obesity has focussed on rich countries such as the United States and Sweden, where are being rich is associated with being thin. However, poorer countries in South Asia tend to have a positive relationship between obesity and socio-economic status, which is seen in the results. The higher the standard of living of the parents during childhood, the higher the wast circumference and the BMI of the offspring.

Of course, this study was conducted in rural South India and its results may not be applicable elsewhere. Nonetheless this study does put something in perspective. In the West, the richer you are, the less your chances of actually getting heart disease. But in India, the richer you are, the fatter you are likely to be. This is a well-known and understood part of the country. But the more interesting part was that parents' childhood circumstances had little bearing on the risk of heart disease for a child. The science of epigenetics is not very advanced yet, so there may be scope for increasing our understanding of these links with future studies.

Managing Gastric Cancer

Gastric cancer is an extremely significant cause of death worldwide. More than 1 million people get diagnosed every year, making it the fourth most common sort of cancer. Among cancers, it accounts for the third highest number of deaths worldwide.And more unfortunately, it seems as if the incidence of this cancer worldwide is rising, not falling. There were an estimated 356,000 more cases and 96,000 more deaths from gastric cancer in 2017 compared to 1990.

An interesting thing to note about gastric cancer is that it is similar to cervical cancer in one sense: one sees both their numbers rise with certain infections. Human papillomavirus is strongly linked to an increase in chances of cervical cancer, and Helicobacter pylori (H. pylori) infections are considered a strong biological risk for developing gastric cancer. In fact, H. pylori has been designated as a class I carcinogen by the International Agency for Cancer Research (IARC).

Fortunately, H. pylori infections can be controlled using antibiotics. While many Western countries have a low prevalence of gastric cancer, it has been found that treatment for H. pylori infections can still lead to significant reductions in deaths caused by gastric cancer. Regions with high gastric cancer prevalence such as China, Japan and South Korea have begun testing for H. pylori infections through endoscopies and shown a reduction in gastric cancer mortality by up to 40%. Lansdorp-Vogelaar et. al. have reviewed publications looking at the cost-effectiveness of such interventions in Western countries to see what the overall picture looks like. Nine studies looked at the long-term costs and the quality adjusted life years (QUALYs) gained by H. pylori testing and treatment in Western countries for the overall population.

All studies evaluated once-only serology testing for H. pylori, with one study comparing this strategy with faecal antigen testing and C-urea breath test (C-UBT) screening. Assumed test characteristics were high with sensitivity estimates exceeding 85% and specificity estimates of around 80–90%. Test costs mostly varied between US$10–30, with the exception of New Zealand where inclusion of costs for the invitation and promotion campaign resulted in costs exceeding US$70. Eradication was generally assumed to be successful 80–90% of the time. Two studies assumed lower eradication rates of 50% and 64% respectively. Costs for eradication therapy differed significantly between the studies, from as low as US$ 20 to US$ 125. None of the studies considered the potential adverse effects of widespread antibiotics use.

Three studies looked at the cost-effectiveness for screening for pre-malignant lesions through serum pepsinogen testing and upper endoscopies. Three studies looked at differences between the sexes when it came to the effectiveness of gastric cancer screening, and four looked at the effect of race. One study also compared the cost-effectiveness of screening between smokers and non-smokers (smoking is a risk for gastric cancer).

A surprisingly interesting conclusion to come out is that screening is cost-effective in Western countries. The average cost of screening was $35,000 per QALY gained, which is less than the typical threshold of $50,000. Testing for pre-malignant lesions through pepsinogen tests or endoscopy was seen to not be cost-effective in people with an average probability of developing gastric cancer. However, as the authors point out, there is more research to be done in this arena. One, most studies do not consider the effect of H. pylori eradication after gastric ulcers have started to develop (the Correa cascade). Second, many studies do not look at the other effects of H. pylori eradication, such as the effect on peptic ulcers and dyspepsia, nor have they considered the effects of increased antibiotic prescriptions. The authors also suggest combining endoscopies with colonoscopies for screening, but posit that this might not be a good idea in Europe, where stool samples are more commonly used for screening.

The idea of a universal right to health has formally been around in international circles since the publication of the World Health Organisation's (WHO) Constitution in 1946. The "enjoyment of the highest attainable standard of health" was declared a "fundamental right" in that document and has since also been recognised in various international health treaties. Modern Liberal thinkers also contend that the right to health is a prerequisite for enjoying all other rights (perhaps apart from the right to life itself). Most OECD countries have state-sponsored healthcare or insurance schemes which cover the entire population. The developing world is also starting to move in that direction.

This begs the question: does having the right to health spelled out in the constitution make any difference to the provision of healthcare on the ground? After all, the mere presence of a law on the books does not mean that it is being followed in spirit.

The COVID-19 pandemic provides a fairly natural experiment to test this out. In order to check whether the Right to Health makes any difference to the number of infections per million and the number of deaths per million, I graphed the average number of infections per million in countries with an explicit right to health enshrined in the constitution and those without.

The result is definitely surprising. Far from doing better than countries without the right to health, the average country did worse! Does this get better when we look at deaths per million?

It really doesn't. Countries with an explicit right to health were worse when it comes to preventing both infections and deaths.

But that's a counterintuitive result. There is some evidence to show that countries with a guaranteed right to health tend to perform better than countries without one (which I explore further in this post), but if it didn't matter during an actual emergency, there's definitely something wrong. But what?

Defining the Right to Health

Before we go in that direction, it might be a good idea to define the right to health exactly. The "right to health" is somewhat strange terminology. Superficially, it seems to mean that countries and governments ought to work towards preserving and enhancing a person's state of well-being. But not only is this definition too broad, it is difficult to interpret in a legal context. While many such phrases tend to be used in international law, it requires time for their meanings to become properly defined. For example, the right to property, if implemented, does not mean that a person has the right to seize any property they desire. What it does mean is that they cannot be deprived of property they already own. The term's meaning has developed through long usage and application in legal systems.

One interpretation of the right to health is that it guarantees a citizen the right to healthcare. However, there is a marked difference between the right to health and the right to healthcare. One implies the right to a certain sense of well-being, while the other implies the right to access a certain standard of care in case one's sense of well-being were to diminish for any reason. While narrower, even the second definition has incurred disapproval in certain sections of the intelligentsia. Philosophers and policymakers have expressed concern about the coercive nature of redistribution which is implied by the right to healthcare.

Why might this be? One reason may be because a typical definition of the right to healthcare often includes things which go beyond medical care. The rights to "protective environmental services, prevention and health promotion and therapeutic services as well as related actions in sanitation, environmental engineering, housing and social welfare" are also typically included in the right to healthcare by many philosophers and public health officials. In other words, the right to health as understood by many professionals may be better defined as the right to health protection and the right to healthy conditions.

However, despite these problems, almost all countries have ratified at least one international treaty which includes the right to health. Many countries also explicitly include a right to health in their constitutions. Among the international treaties and declarations, the most well-known which contain a RTH include the Charter of Fundamental Rights of the European Union (Art.35, 2016), the International Covenant on Economic, Social, and Cultural Rights (Art.12, ICESCR, 1966), the Convention on the Rights of the Child (Art.24(1),1989), the Convention on the Elimination of All Forms of Discrimination against Women (UN, Art.12, 1979), the International Convention on the Elimination of All Forms of Racial Discrimination (UN, Art.5, 1966)), and the WHO Constitution.

Does the right to health actually have any effect on health outcomes?

International law recognises the right to health must be implemented progressively. A country must gradually move towards offering better healthcare to its citizens, and in the absence of such movement, it is expected that such a country will attempt to justify the reasons for which it has sacrificed its pursuit of better healthcare. The presence of greater resources in developed countries means they are further along the path towards the implementation of a comprehensive right to health. However, certain basic steps have to be taken by every country for its implementation. One of those steps is the right to non-discrimination, and another is the presence of a basic national plan towards realising comprehensive national healthcare.

The Alma-Ata declaration of 1978 gave us a clear idea of how a country can apply the right to health. The principle themes identified were:

• The importance of equity

• The need for community participation

• The need for a multi-sectoral approach to health problems

• The need for effective planning

• The importance of integrated referral systems

• An emphasis on health-promotional activities

• The crucial role of suitably trained human resources

• The importance of international cooperation

Some essential health interventions proposed in that document were:

• Education concerning prevailing health problems

• Promotion of food supply and proper nutrition

• Adequate supply of safe water and basic sanitation

• Maternal and child health care, including family planning

• Immunisation against major infectious diseases

• Prevention and control of locally endemic diseases

• Appropriate treatment of common diseases and injuries

• Provision of essential drugs

A number of indicators can be checked to see whether the right to health has informed a country's healthcare plan or not. Backman et. al. (2008) identify 72 indicators for the same.

But one of the best, most unbiased methods to understand the effect of a country's healthcare policies on its populace is to look at under-five mortality. The under-five mortality rate is, as the name suggests, the risk of a child dying before it reaches five years of age. Another, similar metric which may be used is infant mortality. However, under-five mortality is recognised as a superior measure of a health system because it measures five years of potential intervention. The illnesses which cause this are usually preventable, treatable, or both, and most common in low-income settings. Since the right to health encompasses both the right to medical care as well as the right to a healthy setting, the measure which we have chosen ought to reflect the effect of both. The most common cause of under-five mortality tends to be malnutrition, which is easily preventable by providing an adequate food supply to the entire population. Other common causes include acute respiratory infections, diarrhoea, malaria, and birth complications. If a child dies due to any of these conditions, it reflects systemic failures in the health system in both prevention and cure. Since children under five are young enough to not be affected by other factors (such as unemployment, or alcoholism, or anything else affecting an adult's well-being) their deaths can be blamed more squarely on health institutions than others.

Unfortunately it is clear that being a signatory of an international treaty which emphasises the right to health is not associated with improved social or health outcomes. Encouragingly, though, multiple studies have noted that the presence of an explicit right to health in the constitution of a country has been associated with a significant decrease in under-five mortality. As Kavanagh (2017) points out, Botswana has many ingredients for "better health than its neighbor: higher GDP, stronger economic growth, lower economic and gender inequality, fewer ethnic divisions and electoral democracy that is both longer-standing and by several measures, as strong, if not stronger." Yet, South Africa's under-five mortality rate is 10% lower.

This does not imply that an explicit guarantee to health in the constitution can compensate for other issues within a country. Typically, a wealthy country will always have better healthcare than a poorer country. Countries with greater state capacity will almost always have the advantage in deploying healthcare resources. Other factors which affect mortality, and under-five mortality in particular, include female education, inequality, ethnic fractionalisation, urbanisation, political conflicts and violence, and population density.

However, the effect of having such a constitutional right is not small. It is the equivalent of going from a polity like Iran or Belarus to one like Switzerland or Costa Rica, or moving from the ethnic fractionalisation of Malawi to that of the Netherlands. This relationship seems to point towards an institutional effect.

In fact the effect is not just limited to under five mortality. It has also been seen that countries with a constitutional right to health have low inequities between boys and girls. In other words, a constitutional right to health significantly reduces the disparity between male and female under-five mortality rates in a country. The effect is roughly similar to the difference between Zimbabwe and Zambia, or that between the United States and an average high income country. Another interesting observation is that countries with a constitutional right to health do not invest more in healthcare, they just invest better. They tend to have more targeted programmes and more focused healthcare systems. It seems that the right to health manifests itself in a firmer social commitment towards providing better healthcare, possibly in the arena of governance, rather than increased funding.

What happened during COVID-19?

Ideally, if the right to health has actually made a measurable difference in the institutions of the country, then the effect ought to be felt in its response to the coronavirus. But as we've already seen, that's not the case.

Nonetheless, let have a look at some more data. Worldometers has a very nice table which lets us see the number of infections, deaths etc. per country. Sorting them by number of deaths, we see that the first ten countries are evenly split between those which have the right to health in their constitutions and those which do not. The United States, India, France, Germany and the United Kingdom do not have the right to health, and Brazil, Mexico, Italy, Russia and Spain do. Sorting by number of cases reported changes the rankings slightly and  Mexico gets replaced with Turkey. Turkey does not have the right to health either.

Sorting by the number of deaths per million yields a radically different set of countries. Yet even in this bunch, we see that Hungary, Czechia, North Macedonia, Montenegro and Belgium have constitutionally defined rights to health, and Gibraltar, Bosnia and Herzegovina, Bulgaria, San Marino and Slovakia do not.

Sorting by the number of cases per million gives us a slightly different set of countries. Among these, Andorra, Montenegro, Czechia and Slovenia have the right to health, while San Marino, Gibraltar, Bahrain, Luxembourg, Sweden and the United States do not.

This, being an admittedly poor way of looking at these results, tells us very little. Most countries do not have a constitutionally defined right to health, so there being approximately equal numbers in the top 10 of these tables may mean nothing.

But we've already had a look at the average country with and without the right to health in its constitution. Just to give some more context, here is a graph of the standard deviation for each day of daily infections per million which measures intra-group differences. A high standard deviation points to there being radical differences in the number of infected persons per million within a group of countries (in this analysis those with the right to health and those without). Conversely, a low standard deviation means that the number of infected people per million were similar within a group.

Similarly, here is a graph of the standard deviations of the number of deaths per million.

This naive look at standard deviations tells us as compared to countries with a constitutional right to health, countries which do not have a right to health saw more uniformity in the number of people getting infected (even as the average number of people per million getting infected in these countries stayed lower than those with a right to health), and both groups showed similar uniformity in the number of deaths per million (though the average country without the right to health managed it better than the average country with the right to health).

At first blush, it seems that having a constitutionally defined right to health may be detrimental to actually being healthy. If this analysis is correct, that would mean that the areas of the health system which these countries focus on might not be the ones which actually matter in an emergency. It could be that the social obligation towards providing healthcare to the entire population leads to excessive focus on governance aspects, while neglecting other parts of the system, such as healthcare workers, funding, or supply chains.

However, a more mundane reason might be that countries without the right to health include many developed countries with very good health systems. Japan, South Korea, Singapore, Australia and New Zealand all have no provision for such a right in their constitutions. Most Western European countries also recognise no such legal right. While I have not done this analysis, it would not surprise me to know that most countries with a guaranteed right to health tend to have weak health systems and low budgets for healthcare. These countries also tend to be in historically poorer parts of the world: there are very few countries with the right to health in Western Europe and North America. Surprisingly, even China does not have the right to health enshrined in its constitution (or did not in 2008 which is the year my dataset on the right to health dates back to).

Another reason may be that countries with a clearly defined right to health are more transparent about the number of infections and deaths within their borders. This presupposes a clearly malicious intent on part of the political executive in countries without the right to health to make sure they are not caught with their pants down.

At the end of the day, more nuanced research needs to be performed on this topic. It might be interesting to see whether these results change by region, or when controlled for GDP. Maybe we will see that the legal right to health is very good for normal, mundane health issues, but it isn't very good when applied to epidemics or emergencies of this type. Or maybe this demonstrates the need for more inclusive metrics when analysing health systems.

References

• Leary, Virginia A. "The Right to Health in International Human Rights Law." Health and Human Rights 1, no. 1 (1994): 24-56. Accessed May 01, 2021. doi:10.2307/4065261.

• Roemer, R. “The right to health care--gains and gaps.” American journal of public health vol. 78,3 (1988): 241-7. doi:10.2105/ajph.78.3.241

• https://www.orfonline.org/expert-speak/declaring-the-right-to-health-a-fundamental-right/

• Hunt, Paul, Gunilla Backman, J. Bueno de Mesquita, Louise Finer, Rajat Khosla, Dragana Korljan, and Lisa Oldring. "The right to the highest attainable standard of health." Oxford textbook of public health (2009): 335-350.

• Backman, Gunilla, Paul Hunt, Rajat Khosla, Camila Jaramillo-Strouss, Belachew Mekuria Fikre, Caroline Rumble, David Pevalin et al. "Health systems and the right to health: an assessment of 194 countries." The Lancet 372, no. 9655 (2008): 2047-2085.

• https://www.exemplars.health/topics/under-five-mortality/what-is-under-five-mortality

• Palmer, Alexis, Jocelyn Tomkinson, Charlene Phung, Nathan Ford, Michel Joffres, Kimberly A. Fernandes, Leilei Zeng et al. "Does ratification of human-rights treaties have effects on population health?." The Lancet 373, no. 9679 (2009): 1987-1992.

• Kavanagh, Matthew M. "Constitutionalizing Health: Rights, Democracy And The Political Economy Of Health Policy." (2017).

• https://www.worldometers.info/coronavirus/

1971 saw Tudor Hart write an enormously influential paper in which he talked about the reasons people face inequalities in gaining access to healthcare. He named this phenomenon the Inverse Care Law (ICL). In his own words:

The availability of good medical care tends to vary inversely with the need for it in the population served. This ICL operates more completely where health care is more exposed to market forces, and less so where such exposure is reduced.

It is a fairly memorable phrase, borrowing terminology from various inverse-square laws in the sciences. However, unlike those inverse-square laws, this law was not based on empirical evidence or rigorously tested experiments. This statement came about through an examination of the common experiences of frontline doctors in the British National Health Service (NHS). Doctors had been talking about these issues for quite a few years at this point. Richard Titmuss (1968) notes that it had been seen that higher income patients "know how to make better use of the service; they tend to receive more specialist attention; occupy more of the beds in better equipped and staffed hospitals; receive more elective surgery, have better maternal care, and are more likely to get psychiatric help and psychotherapy than low-income groups – particularly the unskilled."

Hart argues that this is not an issue one would be able to capture through plain statistics. After all, hospitals make it a point to not note down characteristics such as social class and neighbourhood, and even if they do, getting access to such data tends to be an exercise in frustration. At best, the data one can obtain access to tends to be use-rates across different social groups, that is to say the rates at which different social groups access healthcare services.

However, use-rates can be interpreted in different ways. One of Hart's contemporaries, Rein, writing in 1969, says that the conclusion drawn by Titmuss is abjectly wrong. Since I was unable to access the original paper, I do not have any particular understanding of his argument save what Hart mentions. Rein's argument in the context of Hart's paper is that there are two types of diseases:

• Ones in which low consultation rates are associated with high mortality, which is to say that patients tend die in case they do not get access to a practitioner

• Ones in which high consultation rates are associated with high mortality, which is to say that going to a doctor does not really make any difference

Rein states that these associations remain the same for different diseases regardless of your social class and goes on to say that one cannot make any sense of the argument given by Titmuss. Hart, of course, disagrees. He states that "the more one examines this argument the less it means," and that it can only be used to justify the raison d'être of a universal healthcare system. If there's a disease whose mortality reduces with increased consultation rates then it makes sense to increase consultation rates, and one way of doing that is by increasing everyone's access to doctors.

A modern approach to the ICL

2021 marks the fortieth anniversary of Hart's paper. It has been cited more than 3500 times since its publication and remains a classic in healthcare studies. However, a lot of work has been done since in understanding causes, effects, and factors. The Inverse Care Law has been examined, re-examined, and broken down into its various factors. Most of the data in this section comes from Cookson et al (2021).

Fundamentally, the Inverse Care Law remains the same as when Hart first introduced it. The ICL remains concerned with differences in healthcare which stem from social disadvantage. It doesn't care whether those disadvantages arise from race, class, or ethnicity. If there are two people with the same need for healthcare, the ICL tells us that the more socially advantaged person is more likely to receive it. The ICL cannot tell us anything about the case when two equally socially disadvantaged people have differing needs for healthcare (questions like "will a socially disadvantaged person be more likely to get treatment in case they have cancer or the flu").

However, modern approaches to the ICL differ from Hart's time in terms of certain details. They tend to distinguish between the quantity of healthcare resources (workforce, expenditure per capita, utilisation of services) and inequality in in the quality of care (clinical process and risk-based outcomes). Unlike Hart's paper, most social scientists would define the ICL without alluding to market forces or other causal mechanism. To quote Cookerson et. al.

This definition facilitates a dispassionate scientific approach to investigating different causal mechanisms and how they operate in different social, institutional, and regulatory environments, including the market mechanisms emphasised by Tudor Hart (such as financial barriers for patients and labour market choices by doctors) but also mechanisms that can arise in both market and non­market [sic] settings (such as dysfunctional government, non­financial [sic] barriers, and unequal costs and benefits of care).

And finally, there are two different types of ICLs identified, both shown in the graph above. The first is the traditional inverse care law, in which increasing social disadvantage leads to dropping standards of healthcare. It can be defined in two parts:

1. More socially disadvantaged people tend to have worse health than less socially disadvantaged people

2. More socially disadvantaged people tend to receive lesser healthcare and worse healthcare service quality

The second is the disproportionate care law, in which increasing social disadvantage does lead to better healthcare, but not the extent required, causing there to be an increased relative disadvantage compared to the better off. It can also be defined in two parts:

1. More socially disadvantaged people tend to have worse health than less socially disadvantaged people

2. More socially disadvantaged people tend to receive more healthcare than less socially disadvantaged people, but less as a proportion of need and of generally worse quality

Understanding the concept of "need" and where our processes fall short

It tends to be far easier to measure outcomes and availability as compared to need. Defining need conceptually, harmonising various value judgements, getting detailed data on resource usage, morbidity and other need variables tends to be a herculean task. In general, different variables need to be considered for one to make a good, detailed model of healthcare need. Some of those variables are health behaviours, family support networks, living conditions, travel distance to healthcare facilities, availability of medicines, local labour market conditions, etc. Even when there is agreement about what constitutes a need, the question of whether a need outweighs another whenever there is a resource shortfall tends to be an extremely prickly one. This is often due to clashing and divergent ethical backgrounds of the debaters, but also due to some very fundamental questions about how need ought to be defined: through the lens of cost-effectiveness, through absolute need (as defined by experts), through individual perceptions (the need felt by the population) or through resource use in comparable populations (comparative need).

Keeping this in mind, there are many sources of error. It has been seen that low and middle income countries (LMICs) show a disparity in data availability when it comes to lower and middle income groups. In general, the more socially disadvantaged the population, the more serious tends to be the underreporting about their morbidities.

The experiences of different countries

Various studies have been done to understand the effect of the Inverse Care Law in different countries so as to understand cause and effect better. As a result, it has been found that a complete ICL is seen to operate in most LMICs, where one sees high private expenditure and highly fragmented systems of public funding with extremely high urban-rural divides, and an incomplete ICL (the Disproportionate Care Law) is found to operate in high income countries.

To get a better perspective on this, let us take a look at the experiences of a few different countries.

Brazil

Brazil is an upper-middle income country with a population of around 220 million people. It is an interesting case to study because of its extremely high GINI coefficient (53.8 in 2018, up from around 51.9 in 2015) and the introduction of the National Programme for Improving Primary Care Access and Quality (PMAQ) in 2011. The data used in this section is taken from Kovacs et. al. (2021).

The PMAQ programme is a pay for performance (P4P) programme launched in order to improve primary care delivery through better allocation of funds and improved organisational arrangements. P4P programmes provide an interesting study in contrasts because of the fundamental question one encounters: will healthcare teams serving richer areas be able to take more advantage of the financial incentives being offered in P4P programmes over teams working in poorer areas?

The data from Brazil are fairly unequivocal about this. The measure used by Kovacs et. al. is a PMAQ score, which takes into account many many variables. It was found that while there was a tendency for teams serving richer areas to have better PMAQ scores, that tendency approached zero as time passed. Income-related inequality of healthcare had been observed in Brazil before the advent of the PMAQ, so to see this inequality reducing seems to indicate that embracing universal healthcare and better, more targeted programmes in which performance is rewarded might be a good way of battling the ICL. One factor in this outcome might have been the fact that teams in poorer areas were given higher performance bonuses as compared to those working in richer areas.

Thailand

This section uses data from the paper Tangcharoensathien et al (2019). Thailand is considered a fairly typical success story for universal healthcare, with favourable outcomes, including better access to healthcare services, low levels of unmet needs, and low probability of catastrophic health expenditure. Thailand has a predominantly public healthcare system, with nearly 80% of all beds being in government facilities.

The most important part of making a public health programme a success is to make sure that it's planned properly from the outset, because changing direction once it has been set is extremely difficult. Thailand made a good choice in prioritising more than just service delivery: the government decided to recruit more medical students from rural areas, provide hometown placement for doctors, mandate a three-year service in district hospitals for all public school graduates, as well as impose penalties for non-adherence.

In addition, Thailand managed to unify a series of patchwork insurance schemes dating back to the 1970s into a set of three nationwide schemes in 2002. This managed to increase financial efficiency as well as reduce administrative overhead. While integrating these three schemes into one would have been even better, this structure was adopted in order to appeal to the political divisions of the time. The government coupled this with budget reforms which saw the inclusion of more stakeholders and increased transparency in accounting.

Thailand's experiences have seen them flatten the ICL graph. While it is not completely flat, by all accounts, Thailand's health system is remarkably equitable and covers everyone fairly equally.

United States of America

The USA is an outlier among developed countries in that it does not have a universal healthcare system, nor a universal government insurance scheme. Thus, one sees greater differences here as compared to countries with universal healthcare. The data for this section was obtained from the National Healthcare Quality and Disparities Report 2019.

In general it has been reported that more than half the measures used for measuring access showed improvement from 2000 - 2019. Specifically, there was an improvement in people getting access to insurance coverage. Unfortunately, 25% of the measures reported upon showed no improvement and 20% showed worsening. It was also noted that significant disparities persist and some of them have worsened, especially for poor and uninsured populations. Specifically, it was seen that socially disadvantaged minorities received worse quality of care as compared to Whites according to 40% of the quality measures recorded. In contrast, while Asians received worse care going by 30% of the recorded quality metrics, another 30% showed them getting better care than Whites.

It was also noted that America has a fairly large urban-rural divide when it comes to quality of healthcare. A quarter of quality measures showed there being a significant urban/rural divide.

America, being a rich, free country with comparatively few regulations is worse than Thailand in terms of healthcare inequality. The complete ICL is not seen here, but one does observe the effects of a disproportionate care law.

Switzerland

Switzerland is a high-income country with a relatively strange healthcare system as compared to most other countries in its income bracket. Despite the fact that nearly 65% of healthcare is privately funded, healthcare is fairly equitable and affordable to all. The data in this section is primarily taken from the Commonwealth Fund website.

The Swiss government introduced a Health Insurance Law in 1994 which went into effect in 1996. The 1994 law had three basic aims:

1. Strengthening equality through universal coverage and subsidies for low-income groups

2. Expanding the benefit basket in order to cover more and more conditions and ensure a high standard of healthcare

3. Controlling the growing costs of healthcare

In order to do this, citizens are legally required to buy health insurance. The government's role in this system may be summarised as follows:

Duties and responsibilities in the Swiss health care system are divided among the federal, cantonal, and municipal governments. Each of the 26 cantons has its own constitution and is responsible for licensing providers, coordinating hospital services, promoting health through disease prevention, and subsidising institutions and individual premiums. The federal government regulates system financing, ensures the quality and safety of pharmaceuticals and medical devices, oversees public health initiatives, and promotes research and training. The municipalities are responsible mainly for organising and providing long-term care (nursing home care and home care services) and other social support services for vulnerable groups.

Mandatory health insurance is provided by 56 different insurers, all of which are legally required to be non-profit entities. Individuals select an insurer and pay their premiums directly through the companies. The money is then redistributed through a central fund back to the insurers in accordance with a "risk-equalisation scheme that is adjusted for canton, age, gender, and major expenditures in the previous year, such as hospital or nursing home stays and pharmaceutical costs."

Switzerland's unique system differs from what most developed countries offer. However, it has been remarkably successful at flattening the ICL curve.

India

India represents another extreme in this scheme. It is a middle-income country with almost no universal health coverage to speak of. This led to nearly 85% of all payments to be out of pocket. The data in this section are mainly sourced from Dwiwedi and Pradhan (2017) and Bowser et. al. (2019).

It has been found that the more the per capita out of pocket healthcare spending in a state, the greater the ratio between the spending of the rich and the poor. While this is a fairly obvious result, it was found that higher healthcare spending is concentrated among people residing in urban areas, non scheduled caste and scheduled tribe people, as well as non-Muslims. It was also found that healthcare spending is generally low in certain geographically isolated states: the north eastern states as well as Orissa and Chhattisgarh.

Looking at government-provided services, it was seen that their utilisation tends to be pro-poor. Unfortunately, when net benefits are taken into account, services tend to become more equal and less pro-poor. While the process of reducing inequities in outpatient services has seen progress, in-patient services, which account for a majority of out of pocket spending, have not seen anything resembling advancement. However, there is considerable variation across Indian states which national results tend to hide.

Regardless, the introduction of the National Health Mission and associated programmes has led to significant changes in pro-poor utilisation of services. Unfortunately, India remains a textbook example of the Inverse Care Law. To quote Bowser et. al.:

Although we see a more pro-poor trend in utilisation for deliveries at the national level, when net benefits are included in the analysis, it becomes pro-rich. A potential cause of this trend is that, relative to the unit cost of services, poorer women are paying more OOP [Out of Pocket] for the location where they decide to deliver. This is counterintuitive, as poorer women should theoretically qualify for the different incentive and reimbursement programs that are part of the NHM [National Health Mission]. Randive et al. postulate that these incentives are either insufficient, or that there are other factors accounting for some of this inequality, such as the higher male illiteracy rates or low-quality public health facilities in poorer areas. A qualitative study by Vellakkal et al. also highlights several impediments to institutional delivery. They note that, due to other associated costs (e.g. informal payments), the cash incentive component of JSY [Janani Suraksha Yojana, a component of the National Health Mission] is not an enabling factor for institutional delivery in health facilities, and may actually cause poorer groups to opt out of utilising such initiatives.

Causes of the ICL and healthcare inequality

Judging from these five examples, it is difficult to find causal factors for anything relating to inequalities in healthcare. However, one can hypothesise about a few proximate and distal causes.

1. The first, and perhaps most important cause tends to be financial barriers: if one cannot afford it, one cannot get it. This tends to mostly be a problem in countries where a majority of healthcare spending is out-of-pocket. India is a textbook example. A country which bucks this trend is Switzerland, where, despite most healthcare spending being private, benefits are still distributed very equitably

2. A second important cause tends to be the fragmentation of insurance due to differential eligibility. This difference in eligibility generally occurs because service providers in more socially advantaged communities have greater access to both political and economic power. The United States provides an obvious example of this issue. Another country highlighted by Cookerson et. al. falls on the opposite end of the spectrum: Mozambique. Most healthcare spending is financed by external donors, which tends to be concentrated in easier to reach urban areas. On the other hand, Brazil has managed to reduce this disparity through its P4P healthcare programme

3. A third issue is non-financial barriers. These may be education or literacy, fewer disabilities, better social support systems from family and friends, cultural homogeneity with the doctor and other treating staff, etc. India and the United States provide some examples of this issue. In the United States, Blacks, Native Americans and Hispanics tend to have worse outcomes as compared to Whites and Asians. India sees forward castes and non-Muslims receive better care than other social groups

4. The fourth issue is an offshoot of the third issue. Socially advantaged people are better placed to follow the recommendations of doctors in terms of taking time off for rest and recovery, following certain diets, changing one's surroundings etc. Socially disadvantaged people generally tend not to have those options

5. The fifth problem is the costs and benefits of healthcare delivery. Socially disadvantaged groups tend to have more co-morbidities, more health issues, and more social and psychological problems which leads to more stress on healthcare systems when extended to cover these populations. At this stage, the cost-benefit ratio might not work in favour of whatever entity is providing healthcare coverage

There are other, more systemic factors which affect the provision of healthcare which are endemic to health systems and human nature:

• Doctors wish to work in more affluent areas because there are better facilities available

• The doctors who wish to work in less affluent areas usually wish to do so because of personal or family ties to the area

• There is an over-representation of professional families in the ranks of doctors (who are usually socially advantaged and urban)

• There is a lot of work pressure in socially disadvantaged areas because socially disadvantaged populations tend to have greater burden of disease

• Healthcare facilities (labs, equipment, etc.) tend to be easily available and more plentiful in socially advantaged areas because it is more cost-effective to have them there

And finally, there are more distal factors which affect inequality of healthcare provision among different social groups, leading to a stronger ICL for that country:

• Poor governance

• Low healthcare spending

• Increased wealth inequality

• Being a low or middle income country

• Political power being unevenly distributed

What can be done about it?

There isn't a straightforward answer to this question. The unique combination of factors present in every country guarantees that a single method or framework will not work for everyone. Brazil went for a P4P scheme, Thailand consolidated its various government insurance programmes into three broad insurers, Switzerland regulated insurance providers and made it illegal for a citizen to not have insurance, the United Kingdom has the NHS, and many other countries have other, different mechanisms to chase the same eventual goal.

Unfortunately, despite all the issues which these systems have managed to solve, there remains the problem of social inequality (different social groups not having the same amount of social capital). Implementing a policy which deals with this problem is fraught with challenges. In fact, trying to solve for this problem might even increase intervention-generated inequalities.

Universal healthcare was seen as the first step towards this goal, and it remains the only step which has some degree of consensus. However, there are many countries which do not wish to go down this route (most notably, the United States) because of the substantial public costs involved, and some countries which have managed to achieve it without increasing public costs at all (Switzerland). Unfortunately, even after having achieved universal healthcare, there remain substantial political and human factors which are difficult to work around. It would be considered ethically wrong to create a policy which restricts the right of movement of a doctor in order to achieve significant health equality. It would also be extremely difficult to convince existing actors in the health system of a country to shelve their interests and work towards a common good.

Despite all these factors it is well-understood that putting money into solving this problem is one of the most important things a society can do in order to reduce wider inequalities in health. A reduction in these inequalities can lead to a positive cascade – investments in primary care, community care, preventive care, and basic surgery tend to have some remarkably far-reaching effects.

The first step to be taken is increasing transparency and creating actionable information where there is none. As Cookson et. al. state very elegantly, "Statistics are the eyes of the state, but the state has a blinkered view when it comes to health inequality impacts. Public decision making still prioritises effectiveness and efficiency over equity, relying on analytical approaches that measure averages rather than social distributions." It is important for healthcare analysis to be very close to the ground: to create forecasts based on costs which actually correspond to ground realities. It is imperative that value-for-money calculations be done without inflating actual spending thresholds in real households.

The second step needs to be the inclusion of data scientists and technology in the arena of healthcare planning. The Apple Watch is the first platform seeing widespread adoption where people are volunteering health data for large-scale studies. These studies need to be extended, made more inclusive (which might be an easier problem than increasing the inclusiveness of healthcare itself) and used to seriously design healthcare programmes and policies. Increasing healthcare costs are inevitable, but gathering this kind of data might be the best way of limiting it as far as possible by eliminating waste and increasing targeting efficiencies.

I wish I could think of a third step. The truth is that COVID-19 has shown us the power of the global healthcare system. When pointed at a problem, it definitely does have the capacity to solve it. Advances in technology might lead to a reduction in healthcare costs which has the potential to reduce financial barriers to healthcare access. Or maybe advances in artificial intelligence will make all these questions and calculations moot.

A recent paper about social health insurance and rural populations in the European Journal of Health Economics gives a nice analysis on the effect top-down insurance can have on rural populations. The authors, Green, Hollingsworth and Yang, look at how the roll-out of health insurance by the State has gone in China. The reason they chose his particular scheme is because it forms a natural experiment. Since it was voluntary, counties were not mandated to opt into it, and thus they did so unevenly. This makes it possible to compare data both temporally (how an area did before and after the scheme was adopted) as well as spatially (how did the experiences of areas which had adopted this scheme differ from those which had not at a given moment in time).

Background

China historically relied on the Rural Cooperative Medical System (RCMS). This was a system built upon peasant communes and was financed by communal welfare funds. However, the late 1970s saw China transition into the Household Responsibility System leading to a collapse of the communes and the RCMS leaving 90% of all peasants uninsured. This led to sky-high medical costs. There were attempts to reestablish these kinds of systems by townships and villages individually, but they did not survive for long.

Thus, the New Cooperative Medical Scheme (NCMS), a public health insurance scheme in China aimed at rural areas, was established in 2003 with the aim of reducing catastrophic health spending among the peasantry. To quote from You and Kobayashi (2009):

In 2003, 96% of rural households in China lacked medical insurance, 38% of the sick did not seek medical attention, and medical debt forced many households to reduce food consumption. The ability to pay became an important determinant of access to health care. Large negative health shocks reduced annual income in rural China by an estimated 12.4%. Health care expenses caused 2.5% of households to fall below the poverty line in 1995, and 22% of poor households attributed their poverty to illness or injury in 1998. Illness has become a leading cause of poverty in rural areas. Urban–rural differences in financing and access to health care have also risen sharply.

This scheme was voluntary because by this time, the Chinese had become wary of government taxes and public trust in governmental intentions had been running low. The scheme was created to operate on a county level rather than village or township level and all management costs are paid by the central government. The enrolled households in participating counties are taxed at a flat rate (with allowances for poorer households).

Many studies have been conducted on this programme over the past years. In general, they have shown that while access to healthcare has improved tremendously, the average out of pocket cost and the risk of catastrophic medical expenditure has not reduced. However, Green, Hollingsworth and Yang have stated that most such studies have looked at the first few years of policy implementation and have not performed any follow-up, thus it might be informative to look at how the programme has done after a wider roll-out has been implemented.

Available data indicate that this programme has gone from covering around 10% of the rural population in 2003 to to 90% in 2008 to 99% in 2014. Since it is voluntary, however, that does not guarantee that everyone covered by this scheme is enrolled into it. While the reimbursement rate varies from county to county, on average, the reimbursement rate was targeted at 50% in 2009, increased to 70% in 2011, and then to 75% in 2015. It primarily aims to push people towards primary healthcare centres in rural areas, because travel costs associated with travelling to cities for to tertiary healthcare sites was found to be fairly high. It does this by providing more generous reimbursement for getting treated at primary healthcare centres. Government expenditure was directed towards improving the rural healthcare system, training more healthcare practitioners in rural healthcare, and staffing these healthcare centres better.

Methodology

The authors get their data from the China Health and Nutrition Survey (CHNS). The key variable which they looked at was the kind of health insurance a person has. The authors compared people in counties which had not implemented NCMS with people enrolled in NCMS in order to reduce selection bias (e.g. someone deciding to not get NCMS-based insurance because they believe they are healthy enough to not need insurance or because they are too poor to afford the cess).

The authors use a logit model to estimate the effect of participating in the NCMS on medical care utilisation and out of pocket payments. The probability of out of pocket payments is modelled using a logit model and the amount paid is modelled using a generalised linear model. Unobserved confounding variables are accounted for using an Instrumental Variables approach. To quote the authors:

Following previous literature on impact evaluation of the NRCMS, insurance availability at the county level is used to derive an instrument for individual enrolment into the NRCMS. This is considered an appropriate IV because it satisfies the two requirements of validity: (1) there is a high correlation between the county NRCMS status and household insurance enrolment status, as shown in Tables 1, 2 where the coefficients of county NRMCS status in the first-stage regression are very large, and as we show later, these pass standard thresholds for detecting weak instruments; (2) the roll-out of the NRCMS across counties and over time can be treated as good as random conditional on community characteristics.

In order to integrate both spatial and temporal data, the authors look at province fixed events and time fixed events. This is able to capture differences between provinces as well as common events which affected all of them in a given year. The adoption of the NCMS by a county is used as an exogenous variable, because it has no effect on individual healthcare utilisation or expenses except through the means of the insurance. A 2-stage linear regression model is then used to evaluate the impact of insurance on the outcomes the authors are interested in, looking at different counties one at a time.

The Long-Term Effects of Expanding Social Insurance

The authors agree with previous studies which showed that there is little impact of this insurance scheme on out-of-pocket payments by patients. However, it was found that there is a marked increase in the use of village primary health centres as well as a reduction in patient load in city hospitals (which, as the authors note, just missed being statistically significant). This tells us that insurance does have the potential to change patterns of healthcare use in rural areas.

A more nuanced look at the results shows that the reduction in patient load at city hospitals tends to be driven by rich households, whose members actually had the option of going to city hospitals in the absence of insurance. These patients now seem to go towards township health centres, which are secondary-care centres. Poorer patients are the ones who seem to go towards primary care centres in villages through the NCMS. The authors also show that there is a reduction in out of pocket payments for poor households (though it is judged to be insignificant), and that people from more deprived provinces in the west seem to utilise village clinics more than those from richer provinces in the east.

While the results of this paper are in line with results from previous studies, it was found that aggregate results tend to mask effects on different slivers of the population. NCMS may have contributed towards correcting distortions int he rural healthcare system by directing people towards primary rural healthcare systems and reducing the usage of specialty tertiary care centres. However, it is not possible to verify whether this is happening due to the relative poverty of those who go to village clinics or due to an increase in quality among them. A clue that it may be the former comes from the fact that richer people tend to still go towards township secondary health centres, which seems to suggest that they are unwilling to compromise on the quality of healthcare.

It was also seen that the probability of incurring out of pocket payments did not go down as a result of this scheme. This may be because inpatient care, which is much more expensive than outpatient care, is only provided in tertiary hospitals, and thus the reimbursements provided by this scheme might be very low.

Policy recommendations

While the scheme has had an effect on reducing the number of people going for informal medical care to traditional doctors or self-medicating, it has not been a universal success in reducing out of pocket costs or changing healthcare utilisation. This, and the other results presented above may be of use in implementing policies in developing countries such as India:

1. In order to keep reimbursement costs manageable, it is recommended to keep a tiered reimbursement structure which incentivises visiting primary health centres over tertiary care hospitals in order to reduce congestion at the latter

2. Inpatient care needs to be subsidised in order to reduce average out of pocket payments

3. Quality of healthcare at primary care centres needs to be increased in order to make sure that people continue visiting them and that the policy actually has a positive effect

One of the greatest tragedies of COVID-19 has been the loss of life and livelihood amongst the poorest of the poor. Multiple countries have seen the breakdown of social security systems, deaths in the hundreds if not thousands, and job losses across the board. The question of lives vs livelihood has played out in different ways across the world. From the chaotic mismanagement in America to Taiwan's deft handling, though, conventional wisdom has it that the poor have borne a disproportionate brunt of the pandemic.

A recent NBER working paper by Angus Deaton challenges that notion. He argues that global inequality hasn't increased, but decreased due to the pandemic. He sets his argument out using three premises:

The number of deaths per capita has been higher in richer countries

This statement forms the crux of his paper. Conventional wisdom would have one believe that richer countries would tend to do better in pandemic prevention and control. And typically, that tends to be true. Health outcomes are better in richer countries with more developed health systems. A comprehensive study published by Johns Hopkins, the Nuclear Threat Initiative, and the Economist intelligence Unit about global health security created a set of global health indices which measure country capacity in 6 dimensions using 140 questions. Four of these dimensions are:

1. Prevention of the emergence and release of pathogens

2. Early detection and reporting for epidemics of potential international concern

3. Rapid response and mitigation of the spread of an epidemic

4. Sufficiency and robustness of the health system to treat the sick and protect health workers

Countries which tend to do well in each of these dimensions tend to be richer and wealthier. However, richer and wealthier countries also have a high mortality rate as a result of COVID-19.

So in essence, what this tells us is that countries with better health systems as measured by the Global Health Security Index have tended to do worse than countries with worse health systems, a fairly counter-intuitive result.

The author does make it clear though, that this is not his last word on this assertion. The pandemic isn't yet done devastating our species. Tanzania and Burundi have extremely low death rates: to assume that their data collection has been perfect would be wrong. Vaccines have not yet had any chance to affect deaths yet as well. Rich countries, with better health systems and more efficient systems may yet outpace poor countries in vaccine distribution. Poor countries also have the edge over rich countries in terms of demographics and weather. Poor countries also tend to be warmer: more work is done outside as compared to rich countries. In addition, countries like Taiwan, South korea, China, and many in Africa also have institutional experience of combating SARS and other infectious epidemics: experience which rich countries lack.

The author does not wish to disentangle the relationship between all these factors. At this point, all that needs to be asserted is that there is a positive relationship between number of deaths per capita and per capita income.

There is a negative relationship between predicted GDP growth and deaths per capita

The second argument comes from a straightforward look at the figures published by the IMF and the World Bank and it gives us an easy assertion. If a country has more deaths per capita, then the decrease in its GDP will be greater.

The relationship here is a lot clearer than that for deaths per capita vs income. China and India both are outliers: China has had far fewer deaths than its population would indicate, and India has had greater.

This brings the relationship between lives and economics starkly into the picture. The perceived relationship was one of trade-off between economics and saving lives. But a look at the data shows us that saving lives and economics were inexplicably linked. While this might seem somewhat obvious in retrospect, it was not obvious when lockdowns were being implemented that they would be anything but a good idea.

There is a negative relationship between per capita income and predicted GDP growth

The last assertion in this argument shows that richer countries grew slower. The relationship is fairly slight, but it does exist.

While the relationship isn't as stark as the one between deaths per capita and GDP growth, one can still see where it's going.

Income inequality has decreased across the world. If you take out China.

And this brings us to the final statement. A careful look at the data actually shows that income inequality has fallen. To explain what one means by that, one needs to understand how income inequality may be measured. From the horse's mouth itself:

My results concern two distinct measures of international income inequality, the dispersion of per capita income between countries, with each country as a unit of observation, and the dispersion of per capita income between countries, but where each country is weighted by population. Milanovic (2011, Chapters 1 and 2) has usefully labeled these inequality measures as Concept 1 and Concept 2 respectively. Concept 1 treats each country as an individual and calculates inequality between those individuals. Concept 2 pretends that each person in the world has their country’s per capita income, and then calculates inequality among all these persons. Both Concept 1 and Concept 2 are between country measures and both ignore within country inequality.

The data for plotting this graph come from the World Economic Outlook 2020 (for the solid lines) and 2019 (for the dashed lines). The 2019 document also has pre-pandemic predictions for 2020, which the graph shows. It shows the standard deviation in log per-capita income over different years. If one looks closely, one can see a dip at around 2008 during the Great Recession in all three sets of time series.

The lines on the top are population unweighted data (concept 1). The lines in the middle are weighted but excluding China, and the lines of the bottom are weighted and include China. Income inequality has been falling over the years as China and India pull themselves out of poverty. However, there is a clean break in 2020 when one looks at population-weighted data. Inequality actually did rise in the world for the first time in many many years.

The reason behind this is entirely China, which pulled itself out of the pandemic much better than had been expected. As the author says, the reason why this uptick has happened is because it has pushed 1.4 billion Chinese people ahead of 1.4 billion Indian people. Since more people are now poorer than the Chinese than are richer than them, the more China pulls away from the mean, the more inequality grows.

Limitations

However, this study ought not to be the last word on this topic. It totally ignores the pain and suffering which many people have gone through, most of which has been in poorer countries. It also does not account for the number of people who have been pushed back into poverty in poorer countries as their incomes have dropped. Tales of dropping or rising inequality do not change their predicament. It also does not look at the role of institutional knowledge and strength, or for that matter, trust in the government.

In addition, the statement of "richer countries have suffered more than poorer countries" ought to be obvious: richer countries have more to lose. However, it does challenge the methodology used to assess health system readiness for epidemics. The six dimensions used to assess readiness were obviously not enough, else there would be no divide between facts on the ground and theory.

Finally, there might be other types of inequality which have grown during this time: access to opportunities, jobs, medicines, and a better future, for instance. Relegating them to a distant last in pursuit of economic growth ought not to be the driving ethos of any modern society. Even if one looks at economic growth, this study does not look at within-country inequality, which may have actually increased during this pandemic as resources concentrate in the hands of those who already have a lot.

This week's newsletter is a mix of geopolitics, COVID-19, the economics of screening and finally, the economics of antibiotic resistance. I selected four papers this week. The first paper talks about the link between COVID-19 and geopolitical risk, finding that a heavy COVID-19 burden leads to increased geopolitical risk for countries. The second and third papers examine giving mothers antenatal care in the Niger and screening for HIV in Australia respectively. Finally, the last paper looks at the load antibiotic resistance may be placing on China's GDP.

COVID-19 and Geopolitics

A paper by Wang et. al. delves into the question of whether there is any causality between the spread of COVID-19 in a country and the geopolitical risk it faces. This seemed like a very interesting question to explore: I haven't seen any other analyses done in this manner, and especially not of COVID-19. The authors correlated the geopolitical risk with new COVID-19 cases per million and new deaths per million. The Geopolitical Risk is measured by an index described in a paper by Caldara and Iacoviello. The authors describe it as:

The authors search the archives of 11 international newspapers Boston Globe, Chicago Tribune, Daily Telegraph, Financial Times, Globe and Mail, Guardian, Los Angeles Times, New York Times, Times, Wall Street Journal, and Washington Post. The authors introduce the index by calculating the news related to all news articles related to geopolitical risks. The calculation is the basis of the data at the monthly frequency.

The authors created linear models and used the cross-section dependence test and the Granger non-causality test to check causality. The results were as you'd expect: geopolitical risk seems to be affected by new COVID-19 infections and deaths. The geopolitical risk index of a country is higher the more affected it is.

Links between antenatal care attendance and out of pocket costs

Ouedraogo et. al. examine the economics of antenatal care in Zinger, Niger. It had been seen that neonatal mortality went down from 54 deaths per 1000 live births to 26 deaths per 1000 live births between 1990 - 2007. One of the reasons this might have happened is because pregnant women and children under 5 years of age are exempted from healthcare costs. However, this does not translate to reduced maternal mortality, which remains stubbornly high at 556 per 100,000 in 2016 despite the free healthcare package including antenatal care, caesarean sections, and the treatment of gynaecological cancers. According to the authors:

For the routine ANC (antenatal care), the official exemption includes governmentally covered costs for the antenatal health booklet, consultations, intermittent preventive malaria treatment with sulfadoxine-pyrimethamine, iron folic acid (IFA) supplements, laboratory tests (glucose, albumin and infections), ultrasounds and consumable items necessary for the provision of these services. Prior to 2016, the World Health Organization (WHO) recommended at least four ANC visits for uncomplicated pregnancies with the first ANC visit occurring before the 12th week of gestation, the second visit around 26 weeks, the third around 32 weeks and the fourth between 36 and 38 weeks of gestation.

The primary reason this is done is to establish social contact with the mother and child and reduce both antenatal and post-partum complications for the mother and the child. In addition, previous studies in Ethiopia, Tanzania, Bangladesh and Pakistan have shown that higher out-of-pocket costs and greater travel time have been correlated with decreased maternal health seeking. In lieu of this, the authors had three aims:

1. Assess out-of-pocket costs for pregnant women attending antenatal care (ANC)

2. Estimate time spent to attend ANC and the opportunity cost of this time

3. See the affect of the above on antenatal care attendance

In order to do this, the authors looked at rural communities in their selected area. 18 Intermediate Health Centres (IHCs) were chosen out of convenience. Approximately 77 women were chosen per IHC. Since each IHC serves multiple villages, the authors tried to ensure that the participants were selected from multiple villages as well. The authors looked at ANC attendance based on gestational age, out-of-pocket costs, time spent for an ANC visit, opportunity cost, and they controlled for socio-economic status, food security and dietary diversity.

The authors were able to cover 92 villages in their selected region. They found that:

• 75% of the women they surveyed reported going for an ANC visit.

• The women who were attending were more likely to have begun attending in the hot season (as compared to the rainy season), or had attended previously for another pregnancy

• The women attending had, on average, more formal education than the country's average and also tended to be from non-farmer families

• A majority of women sought ANC for the first time in their second trimester

• Most women walked to their ANC centre and spent less than a day attending their session

• The median opportunity cost reported was between $0.08 - $0.57

• Most women saw out-of-pocket costs during the first visit, but very few reported any such costs subsequently. The median out-of-pocket charge was around $0.38 - $0.57

There also turned out to be a positive correlation between having delivered in a health facility during a previous pregnancy and seeking antenatal care. Women who had received an insecticide-treated-net or food during the visit also tended to come back for more visits.

But the most bizzare result in this paper was that women who spent money to receive antenatal care (had out-of-pocket costs) were more likely to have attended all their required ANC sessions. This goes against what has been seen in other countries and merits further investigation.

Comparison of HIV testing regimes

A paper by Williams et. al. sets out to compare the costs of three different regimes of HIV testing, namely conventional testing, parallel testing and point-of-care testing. This study was performed in Australia, which has some fairly intimidating targets. The HIV policy in the state basically aims to achieve 90-90-90 targets. They wish to diagnose 90% of all people who have HIV in the country, get 90% of them into anti-retroviral therapy, and achieve an undetectable viral load in 90% of them.

In Brisbane, where this study was conducted, the conventional method is an ELISA screen followed by a confirmatory Western Blot. The point-of-care (POC) method is the Direct HIV Combo (DHC) Assay in the form of an HIV self-test, and the parallel method is when an ELISA screen and a DHC Assay are performed simultaneously by an appropriately trained professional. The conventional method is generally seen in hospitals and clinics and is extremely accurate. However, the confirmatory Western Blot requires the patient to come in for a second visit. The parallel method has the advantage of using two screening tests and thus being more accurate as well as being performed in one patient visit. The POC test is the least accurate, but it can be performed at home.

The authors realised that conventional testing hs extremely expensive and the main cost is that of the medical staff. The use of volunteer nurses in this position reduces the cost to a third. Parallel testing reduces costs further because there are no follow-on visits. And finally, POC testing is the cheapest. The authors suggest that the use of POC tests might be the most appropriate here because the incidence of HIV was low enough in the population that a less efficient screening method was fairly cost-effective.

There are some major questions which come to mind at this conclusion. Since the authors don't describe the rate of false-positives when using POC tests, one wonders at the cost the confirmatory testing of these false positives would generate.

The Economic Burden of Antibiotic Resistance

Zhen et. al. take a look at the economic burden of antibiotic resistance as measured in China. They examined the effects of antibiotic resistance on hospital mortality, length of stay, and costs. According to the authors:

China was the second largest consumer of antibiotics in 2010 in the globe. In primary care facilities, 52.9% of outpatient prescriptions and 77.5% of inpatient prescriptions contained antibiotics, of which, only 39.4% and 24.6% were considered appropriate respectively. Among BRICS countries (Brazil, Russia, India, China, and South Africa), more than 57.0% of the increase in antibiotic consumption in hospital occurred in China between 2000 and 2010.

China is now leading the world in terms of antibiotic resistant bacterial strains. In the EU and the EEA, antibiotic resistance's costs were seen to be between 1.1 to 1.5 billion Euros. China's costs might prove to be even bigger.

In order to get the costs, the authors collected data from 3 sites in the Zhejiang Province and one in the Shandong Province. They performed a retrospective analysis.

In this study, we included patients with infection or colonisation caused by S. aureus, E. faecalis, E. faecium, E. coli, K. pneumonia, P. aeruginosa, and A. baumannii. We classified the infection or colonisation cases into susceptible episodes and resistant episodes (including resistant and intermediate isolates) based on the susceptibility test results for the patient specimens.

They defined Single Drug Resistance (SDR) as resistance to at least one antibiotic in one or two antibiotic categories, and Multiple Drig Resistance as resistance to antibiotics in three or more categories.

The economic costs of patients suffering from antibiotic resistance were obtained by multiplying the number of such patients with the average medical cost of a typical patient in a Chinese hospital.

The authors estimated that around 27% of all patients in hospitals had some form of antibiotic resistance and cost China around $77 billion, the same figure as that for the USA.

However there are some limitations in this study which the authors also acknowledge. The first is that the economic costs might not have been estimated properly because the hospitals chosen by the authors may not be representative of actual costs in China. The second is that this study only estimates the effect of antibiotic resistance in hospitals, but antibiotic resistance tends to be felt outside hospitals and in the wider community as well. Finally, the study does not estimate the costs of antibiotic resistance upon those who get colonised by antibiotic resistant bacteria but have not been infected by them.

Happy new year everyone! The first week of the year saw my gaze drawn to a number of interesting papers, some dealing with COVID-19, some with less newsworthy, but more prosaic topics. One paper looks at the role of health systems in combating COVID-19, where the authors conclude that the only variable they could detect which mattered in the effectiveness of a health system's response was the number of hospital beds per capita. Another paper looks at pay for performance programmes in healthcare and their effect on health systems, healthcare providers, and consumers, while a third looks at the ways in which insurance providers use patient-reported data.

We then go on to a paper which discusses genomic detection for non-small-cell lung cancer before coming back to COVID-19. Out of the last four papers, two discuss food systems and information systems in the light of COVID-19, and the other two one discuss the impact of money on healthcare from the perspective of individuals. One looks at the impact of formal credit constraints on rural Chinese people, while the other looks at unexpected windfalls in the form of lottery winnings in Urban Singapore.

The Role of the Health System in Combating COVID-19

This paper by Bayraktar et. al. discusses the role of different health systems in dealing with COVID-19. The responses of different governments towards this unprecedented health crisis have been starkly variant. On one hand, you had China, which imposed draconian lockdowns and conducted mandatory testing on everyone in the city. On the other, you had the United States where the response was both patchwork and badly-coordinated. However, as the authors note:

When the countries with the highest COVID-19 cases ratio are considered, the highest death/case rates belong to France, Italy and the United Kingdom, respectively. On the other hand, the lowest death/case ratios belong to Russia, Turkey and India, respectively. While, the countries with the highest ratio of COVID-19 recovered/cases are China, Germany and Turkey, the countries with the lowest ratio countries are the UK, USA and France, respectively. Turkey is the most successful country in terms of the recovered and death rates. The lowest performances in these two indicators belong to France and the UK, respectively.

The individual country-level responses are fairly interesting to note:

• France had issues in getting its health workers into position because a number of them were striking. Despite the fact that France has a good number of hospital beds per capita, this situation combined with PPE shortages led to a somewhat shambolic initial response

• The United Kingdom had the lowest number of beds and doctors per capita after the United States. It also had (what the authors refer to as) an insufficient number of healthcare workers and PPE kits due to fiscal pressure from the government's austerity policies. In addition, the herd immunity gambit did not work out either

• Italy and Spain called back retired personnel in order to make up for manpower shortage and to control the alarming level of spread that was seen. However, both countries experienced a medicine and equipment shortage as well. Spain also saw too many frontline workers get infected with COVID-19, causing non-specialists to be switched over to COVID duty without adequate training

• The United States was not judged in this paper because the health system, by its very nature, is fragmented and thus cannot be judged monolithically. There is a great deal of variance between states in terms of number of doctors, hospital beds, etc. per capita. However, it is notable that the USA continues to face an issue of insifficient numbers of medical staff as well as PPE

• Turkey and Germany have a fairly good number of doctors and beds per capita. Both countries called up retired doctors and part-time workers respectively to work more and put money into getting COVID treatment to as many people as possible

• Russia was also seen to have a good number of hospital beds as well as medical personnel per capita. Regardless, Russia pressed medical students into action and provided free COVID care to anyone who required it, building temporary hospitals where needed

• China was able to impose strict lockdowns and perform extremely effective contact tracing. When this failed, then the building of temporary hospitals and care centres was their way forward

• Brazil has a state healthcare system (SUS) which covers everyone, with around 80% of the population beholden to the SUS for their healthcare needs. While it does have an advantage in fighting COVID-19 because of its younger population, the prevalence of comorbidities is fairly high

The authors have limited their comments to a small number of countries, but their analysis covers all members of the OECD and a few non-members as well. In order to understand which facets of a health system helped in fighting COVID-19, the authors performed an analysis using artificial neural networks to infer the importance of the following variables:

1. Number of hospital beds per capita

2. Number of doctors per capita

3. Life expectancy at 60

4. Presence of Universal Health Services

5. Share of Health Expenditure in GDP

They come to a number of conclusions:

1. Spending per capita is a bad way of predicting the performance of a health system. The efficiency of a health system (no measure has been given by the authors for calculating this) also makes a difference

2. The prior institutional experience of a country with infectious diseases seems to be a variable worth considering (not considered in this study)

3. The speed of the initial response was very important. The faster the initial response, the better the action on the pandemic

4. The presence or absence of universal healthcare did not matter

5. The only variable which really mattered was the number of hospital beds per capita

This study is, by necessity, limited. The authors identify a number of variables which can be examined in further studies so as to understand how a health system ought to be designed to prevent a pandemic from overwhelming it.

A review of P4P programmes in LMICs

Singh et. al. perform a realist review of Pay-For-Performance (P4P) programmes in Low and Medium Income Countries (LMICs). They aim to fill the gap left by existing systematic reviews in understanding which parts of a health system contribute to the success or failure of P4P programmes.

P4P means that there is a provision of financial incentives to health professionals contingent on them reaching some types of pre-set performance targets. Notably, in LMICs, P4P programmes also seem to also involve:

• A shift to e-Health systems

• New systems of performance and data verification

• Increased financial decentralisation

Specifically, the authors aimed to answer these questions:

1. What were the effects of P4P programmes in LMICs?

2. What contextual factors and mechanisms influence health outcomes when P4P is introduced in health systems in LMICs?

3. Why and how do these factors and mechanisms influence these outcomes?

It was found that P4P schemes tend to encourage health providers (hospitals, polyclinics, etc.) to invest more time in demand-side strategies to achieve incentivised P4P targets, so they try to offer attractive schemes to draw patients in. In doing so, contrary to what one might expect, providers tend to stick extremely closely to clinical guidelines. This causes quality of care to improve because providers start increasing the time they spend per patient and they also improve the availability of drugs and medical technologies. This has often caused patient trust to rise in these programmes. However, the authors do note that this may be linked to how much money the patient has in his or her pocket.

This also has a significant affect on providers. It tends to increase cash at hand and thus increase financial autonomy, and it also has the possibility of lowering the fees charged for services. Unfortunately, it has also been observed to shift attention from procedures not covered under P4P, causing overall quality of care to diminish.

All this leads to a net rise in patient satisfaction because P4P tends to cause healthcare facilities to invest in infrastructure. This leads to increased availability of services and less friction, which often leads to smaller out of pocket fees (caveat: the authors note that increased quality of services tends to lead to an increased willingness to pay more for them in patients). It also leads to increased provider responsiveness.

P4P programmes have many positive effects on health providers as well. Their productivity was seen to uniformly increase across LMICs. There's also an increase in standards of governance and accountability, espcially where administrative roles are well-defined. A major part of this has to do with timely payments, which have a positive association with increased productivity. In such cases, healthcare providers have to depend upon the efficiency of the banking system in their area, procedural efectiveness of their own administration, and the speed with which the donor (the one paying for the P4P programme) disburses money.

Unfortunately, this does not necessarily lead to increased health provider motivation. The authors state that fiscal and administrative autonomy combined with the introduction of P4P programmes themselves may contribute to increased motivation. The extra income from P4P is also seen as a net positive. However, many health providers might not get to keep this money if they are not given financial autonomy and so may not have any motivation to improve. The more decentralised the health system, the greater the positive effect on health providers.

There is a negative side to P4P programmes as well. They tend to encourage providers to game the system and misreport data in order to make more money, especially in countries with inadequate verification mechanisms and low-quality checks and balances. They can also lead to a net increase in prices of services if price are already at rock bottom. And finally, P4P programmes tend to work well in countries with good existing health systems, and ought to be seen as part of a multi-pronged strategy to improve systemic capability.

Understanding the use of patient reported data by insurance agencies

Neubert et. al. deal with a very important topic for our times. They review various studies which talk about the kinds of self-reported health data collected by insurers on patients and how it is utilised. In addition, they also try to see what the motivations of actually collecting these data are so as to understand how this paradigm fits into an insurance agency's wider goals. In order to do so, the authors talk about two very important concepts. Quality of care (QoC), and value-based healthcare. In other words, they may be thought of as variables which look at patient satisfaction/outcomes and bang for buck respectively.

The authors find that insurers look for two types of indicators in general:

• PROMS - Patient Reported Outcome Measures, which tend to measure how patients view the progress of their medical trysts ("I feel better after having followed the doctor's advice, but I think I still need another session," or "I'd rate my pain a 4/10, down from a 6/10 a week ago.")

• PREMS - Patient Reported Experience Measures ("I liked the way the hospital handled MRI scanning," or "I loved the fact that they have experts from three different fields handling prenatal counseling.")

There are other indicators which insurance agencies take into account as well, such as structural indicators (the presence of specific technologies and programmes such as an MRI machine or an HIV/AIDS programme), process indicators (how long did it take to get a certain type of job done and whether it as efficient) and clinical outcome indicators (did the patient survive, have they been put on drugs to manage something chronically). Structural indicators tend to be the least used and generally tend to be reported along with process and clinical outcome indicators, both of which were much more common.

Insurers tend to use this data in multiple ways. PREMS and PROMS are used a lot when deciding to purchase certain services from specific service providers. QoC depends on data availability so it might not be used to make decisions here even though it might be a preferred metric. Value-based healthcare is generally assessed using cost and volume data, especially when going for a selective contracting agreement.

A selective contracting arrangement (SCA) is an arrangement for the payment of predetermined fees or reimbursement levels for covered services by the carrier to preferred providers or preferred provider organizations (PPOs)

The authors also note that different PREMs and PROMs are used for selective contracting and P4P programmes. For the first, all you wish to do is to compare providers with each other. However, for the seocnd you wish to compare providers to an objective benchmark or standard.

Insurance agencies use this data for the purposes of quality assurance and quality improvement. Three perspectives emerge about QoC measures in particular:

• They're used as an ancillary instrument to inform decisions of insurers

• They're used as a means of supporting and enhancing quality improvement via benchmarking providers

• PREMs tend to be used for this task more than other indicators

QoC tends to be a major part of healthcare, and so insurers spend a lot of time on it. They look at measures like effectiveness, efficiency, access, patient-centredness, equity, and safety.

On the other hand, PROMs and claims data are used to get information about populations at risk. All this data is then used to develop new products and programmes.

The authors make several interesting comments which I wish to highlight:

• PROMs tend to be the most popular way of collecting patient-reported health data, followed by PREMs. Structural, process, and clinical outcome indicators trail these two

• The breadth to which this data is used varies wildly across insurers. Insurers need to have a robust system to collect this data, the culture of the insurer ought not to be paternalistic ("I know what's best for patients, there's no point in asking them"), the laws of the jurisdiction needs to allow for collection of needed data, and market forces need to exert pressure to improve services

• PROMs and PREMs are not a stable set of indicators, new measures are continuously being added to these and old measures being removed

• Health insurers are starting to become drivers of provider performance

• There is a great deal of heterogeneity in the terms used in the literature regarding these measures

The authors also report the following limitations which health insurers grapple with:

However, our study highlights three key aspects that hinder a more robust use of such data in a health insurer’s business. First, the insurers’ use of patient-reported data is affected by a large technological and methodological heterogeneity that inhibits the transferability of innovative and effective initiatives across contexts. Second, the varying terminology of constructs used by the many stakeholders with whom an insurer interacts. Third, the involvement of insured people by insurers in the development of patient-reported measures and decision-making in regard to a health insurer’s strategy and practices is still limited. To overcome these hindering factors, health insurers are advised to be more explicit in regard to the role they want to play within the health system and society at large. In addition, health insurers should have a clear scope about the use and actionability of patient-reported measures, and further involve insurees to the extent where it is feasible and deemed necessary.

Comprehensive profiling for non-small-cell lung cancer: health and budget impact

Johnston et. al. write about the health and budget impacts of genomic testing for detecting mutations pertinent to non-small-cell lung cancer. This study was funded by a company which aims to sell a solution for this type of genomic profiling. Specifically, the authors focus on the Foundation One CDx and Liquid tests, which are assays used to profile multiple known genomic mutations in a single assay. CDx uses a small amount of tissue to perform the assay and Liquid requires a blood sample.

The authors created a model using the societal perspective in Ontario, Canada, and took workplace productivity into account for this study. They saw that the use of Foundation Medicine Tests was associated with an increase in budget impact as well as increased life years and workplace productivity. The authors note that multiple gene testing generally leads to a decrease in turnaround time and improved detection rates, but the advantage of using Foundation Medicine products was in smaller amounts of initial tissue sample required. Their study also showed an increase in the use of targeted therapies when using Foundation Medicine products, which they hypothesize has the potential to increase quality of life for patients as compared to traditional broad-spectrum chemotherapy.

Unfortunately, I would take this study with a grain of salt because of the involvement of Foundation Medicine itself in these studies. Certain assumptions were based on clinical experience, and while I don't think there was any reason to suspect academic dishonesty in this paper, I would still wait for independent corroboration of these results.

Food consumption patterns and practices during COVID-19

Murphy et. al. examine changing food habits during the COVID-19 pandemic and the effect it had on the quality of diets and the food system. Their hope is that this information may be used to increase food supply chain resilience. They base their study upon observations in selected cohorts in Ireland, Great Britain, New Zealand, and the United States of America. Their results are fairly straightforward:

• Ireland and Great Britain saw increases in the use of fresh ingredients for preparing meals

• Ireland, Great Britain and New Zealand saw a reduction in food waste, a reduction in takeout and ready-made food, and an increase in baking

• There were reports of there being increased difficulty in all countries in finding ingredients as well as in bulk buying

• The authors report that people in all countries began to increasingly plan their food shopping by making lists, for example

• Great Britain saw an increase in fruit-eating, while Ireland, New Zealand and Great Britain all saw an increase in vegetable eating

• The consumption of saturated fats also went up in all three countries

Ireland saw the greatest change towards eating vegetables, the use of fresh ingredients, reducing food waste, and other positive indicators mentioned in this paper. The authors note that the Irish cohort which was used for this study was predominantly comprised of young, highly-educated females, and so results may have been skewed in its favour. Ireland and New Zealand also reported the lowest difficulty in finding and sourcing ingredients. It is also interesting to see how the United States is notably absent when talking about positive food trends, but it too suffered from similar food shortages as suffered in other countries.

The main takeaway from this paper was that countries which had more targeted and focused restrictions in place were better at managing their food supply chains.

The relationship between formal credit constraints and health status in China

Yang et. al. use the Chinese Household Income Project (CHIP) data to link rural Chinese households' self-reported health status with the formal credit constraints they face from financial institutions. Typically, research in this area focuses on individual demographic attributes, social attributes, the behaviour of individuals, and the impact of environmental quality and hazards. However, this study looks at a different link. In the authors' own words:

Our main contributions are as follows: first, by examining the impact of formal credit constraints on rural residents’ health, we contribute to two strands of literature: credit constraints and individual health. Second, this study clarifies the mechanism of credit constraints on rural residents’ health theoretically and empirically. Third, this study provides a decision-making reference for the government to improve rural residents’ health policy from credit constraints and credit supply improvement. This study is the first attempt to empirically analyze the impact of credit constraints on rural residents’ health in China to the best of our knowledge.

The authors used a theoretical framework derived with the help of two perspectives: credit demand perspective, and credit supply perspective. The credit demand perspective relies upon the theory of credit demand repression. Demand repression tends to happen, according to the authors, when financial institutions set high transaction costs and stringent credit conditions for disbursing loans. This leads to an increase in loan rejection rates. The credit supply perspective deals with the credit rationing theory. As the authors explain, under conditions of fixed interest rates, facing adverse selection and information asymmetry, banks are forced to adopt conditions like handling fees etc. which price consumers out of the market.

These constraints then tend to disrupt income and consumption while increasing economic vulnerability. This can take many forms, for example, lack of credit might cause someone to forgo higher education because there is no way of paying the requisite fees. This hinders promotion of human capital. It also reduces the ability to plan across long time-scales. This has, among other things, a negative impact on entrepreneurship.

In order to understand this relationship empirically, the authors utilised 3 variables: self-related health, days in which illness was reported, and formal credit constraints. These variables were controlled with respect to age, gender, ethnicity, etc. The authors found that 27.02% of rural residents in their sample were found to have credit constraints. In their own words:

Formal credit constraints reduce the probability that a rural resident judges their health condition as good and increases the probability that a rural resident judges their health condition as bad.

An ordered probit model showed the causality beween credit constraints and self-reported health, and a tobit model showed the causality between credit constraints and the number of days taken off from work due to illness.

The authors conclude by saying that these results show that it is a good idea to reduce credit constraints, push people towards formal lending mechanisms through trust-building programmes, and to train people to choose profitable enterprises to sink their money in.

However, the study misses out on a major pillar: the presence of informal credit sources and informal credit constraints (this is a gap which the authors are aware of). A future study ought to also characterise informal credit sources and constraints and see whether their relationship with health status is any different.

The role of governments in information dissemination

Lu et. al. examine the role of governments in information dissemination in the context of COVID-19. The authors state that information travels faster than the virus, and so timely dissemination of information is extremely important. However, disseminating false information, either deliberately or mistakenly, can lead to major issues. Misinformation tends to have a natural sort of resonance with public opinions: people often get illogical or fail to understand the different between true and false statements.

The government has a central role in blocking and spreading information. Minimal blocking of information leads to an open information network but it can lead to uncontrolled spread of rumours. This has the potential to prevent pandemic prevention efforts. To quote the paper:

It is always believed that government should perform as a central node to disclose accurate and up-to-date information to the entire society, so as to keep the public away from untruthful information and prompt the public to make informed decisions about health protection. However, in the real world, governments do face time constraints and the trade-off between being accurate and being up-to-date in terms of information disclosing, which is not considered in classical information theory. Because a highly infectious disease caused by unknown viruses with great externality, such as COVID-19, spreads together with information of varying qualities (truthfulness, accuracy, etc.), it is highly probable that the disease has already contaminated the society before low-quality information is purged. In this case, the government has no way to disclose accurate information in time, resulting in the loss of public trust and raising the doubt of the public on the governing capacity of the government, which will accelerate epidemic outbreak. Therefore, governments need to not only decide when to inject information into the network, but also whether to follow the tenet that governments do not and should not block information spreading at any circumstance.

The major conclusion in literature on this topic has been that governments ought not to interfere with the spread of information in any way in order to maximise welfare. Unfortunately this requires two assumptions to be true. One is that the publishers of this information are perfectly competitive (a perfectly open and fair marketplace) and two, that there is no time constraint. These assumptions do not really hold in the real world. The authors' conclusions are somewhat illuminating:

We introduce the non-dualism of information and the heterogeneity of nodes’ behaviors into the epidemic model and conduct a simulation to reveal the information intervention dilemma faced by the government between information disclosing and blocking. We find that governments face a trade-off between speed and accuracy in information disclosing; and the optimal strategy is contingent on varying conditions in information blocking. The optimal combination of disclosing and blocking is highly sensitive to the government preference and its governance capacity. Governments that are only responsible for the outcome of intervention will focus unilaterally on the accuracy at the expense of speed; a risk-averse government that intends to minimize the maximum infection rate under uncertain scenarios will impose a more restrictive blocking; and the most restrictive blocking strategy might be the best for governments with lower capability and credibility.

It truly makes one wonder where one's own government stands.

The effect of lottery wins on health

Kim and Koh examine the effect of exogenous deluges of money on self-reported health statuses. In order to do so, they focus on lottery wins in Singapore. Their results show that a win of S$10,000 led to a measurable increase in self-reported health status. It does not affect the amount of money spent on healthcare, cigarettes, etc. at all. It also does not discourage the lottery winner from going to work.

The authors conclude that the psychological effects of the lottery win and the general increase in household spending (the latter of which can be measured) may account for the increase in health status of a an individual.

I've had to to a bunch of reading about systems thinking with respect to healthcare recently, sparking an intense interest in the topic. You tend to find a lot of literature online talking about it. Systems thinking is described as an approach to problem solving in which one thinks about the whole system as being comprised of constituent parts and concentrates on parts and relationships both.

However, that doesn't really tell us much about what it is and how one goes about using this paradigm. If one wants to really get into what systems thinking is about, a more academic source might be handier. For the purposes of this article, I refer to a paper called "A Definition of Systems Thinking: A Systems Approach" by Arnold and Wade from 2015^[1] which goes about the process of definition in a very interesting way.

So what really is systems thinking?

We can start by saying that systems thinking is, really, nothing more than a system of thinking about systems. If that sounds like a circular concept, buckle up. Systems tend to be the easiest to define using their function or their purpose; but that is, often, the hardest part of a system to tease out. In this case, fortunately, it is easy. The function of systems thinking is to think about systems!

To go deeper, we need to define what a system is. The Oxford Dictionary tells us that a system is "a set of things working together as parts of a mechanism or an interconnecting network; a complex whole." In other words, it is a set of things (let's call them elements) which are connected to each other, signifying relationships. So systems thinking is a framework which gives us the tools to understand how a system functions.

Why is this needed?

Understanding systems turns out to be fundamental to nearly every task we manage because systems surround us. Nothing in this world is completely independent. Society, the government, electricity transmission, the shipping industry, the ecosystem of a pond, the climate are all examples of systems. Understanding each of them is an art in of itself: understanding their effects on each other and the surrounding world might be beyond the world's fastest supercomputer!

Hold on, you might say at this point. Don't we already understand these things? Aren't we already fluent in our understanding of electricity transmission systems? Didn't we design the government? Isn't the shipping industry the bedrock of all modern consumerism?

The answer is kind of, to all of those. Or rather, it depends on how we define these systems. Sure, we understand electricity transmission systems. But a huge element of these systems is the group of consumers who sit pretty at the other end, the ones who actually consume the electricity being generated and transmitted. Another big part happens to be the workers who maintain these lines. Another element of this system might be politicians (depending on your country) who may have a vested interest in making sure that the constituents of their choice get the generated electricity. Claiming that we understand all these actors and the relationships between them is more than can be said by most people. Similarly, it would be the height of folly for any man or woman to claim that they understand the inner workings of the government or the economic eddies that govern shipping.

The classical approach to problems such as this is called reductionism. Most wielders of the scientific method are intimately familiar with this school of thought. Break a complex problem into its constituent parts, understand each part, and lo behold! The complex, intractable problem is now reduced to several manageable issues.

This tends to be one part of systems thinking. Knowing how to define the elements of a system requires one to be educated in reductionism, but understanding their relationships tends to be a problem of exponentially increasing complexity as the number of elements of a system increase.

So let's start with the easier part of systems thinking. Reducing it to its elements.

Elements of systems thinking

Stave and Hopper (2007) give us a set of elements with which we may start understanding systems thinking^[2]:

1. Recognizing Interconnections

2. Identifying Feedback

3. Understanding Dynamic Behavior

4. Differentiating types of flows and variables

5. Using Conceptual Models

6. Creating Simulation Models

7. Testing Policies

Arnold and Wade (2015) add to and subtract from these elements to give us the following:

1. Recognizing Interconnections: This tends to be the most basic systems thinking still. One needs to be able to identify the key parts of a system and figure out the key relationships between them

2. Identifying and Understanding Feedback: A bunch of these relationships form feedback loops, where two elements may feed into each other. These need to be understood and evaluated separately from the others

3. Understanding System Structure: This is a step above the other two elements, in which one takes these key relationships and understands how the parts of a system fit together so as to describe the system as a whole

4. Differentiating Types of Stocks, Flows, Variables: A stock is a pool of a given resource in a system. A flow would be the inflow or outflow of that resource. A variable would be something like flow-rate or the maximum quantity of a stock. Once the system structure is understood, stocks, flows and variables need to be quantified and described separately

5. Identifying and Understanding Non-Linear Relationships: This has been separated out from the previous element by Arnold and Wade. In particular, this element is also about measuring stocks and flows, but it measures those which do not have linear relationships with each other

6. Understanding Dynamic Behavior: All the elements and relationships which have been understood interact with each other and may behave unpredictably. Such behaviour is called dynamic behaviour. Understanding this is one of the core functions of systems thinking

7. Reducing Complexity by Modeling Systems Conceptually: While complexity is part and parcel of most systems, understanding them often requires stripping away that complexity and observation from different angles. In other words, one needs to know how to model a system so as to be able to explain it well

8. Understanding Systems at Different Scales: And finally, this skill is the ability to understand systems at different scales. Each element of a system may be a system itself, and the system being viewed is probably an element in a more complex system. Being aware of these details can lead to better insights

Relationships

Now that we know all the elements, we can see what the relationships between them are like. The easiest way to understand it is to map it out (adapted from Arnold and Wade, 2015).

This is a slightly complicated map. In short, the thick arrows with solid borders represent strong relationships, and the thin arrows with dotted borders represent weak relationships. The strong relationships are as follows:

1. Understanding system structure strongly enhances understanding of dynamic system behaviour

2. Understanding a system's structure leads to developing conceptual models properly

3. A conceptual understanding of the system may lead to insights about the system's role in as an element in bigger systems as well as the complexity of its constituent elements

The weak relationships come out thus:

1. Understanding dynamic behaviour of a system may prompt one to go back and take a second look at the elements and relationships already identified

2. It is often worth going back and re-identifying elements and relationships when making conceptual systems

3. Understanding a system conceptually makes it easier to identify and seek out dynamic behaviour

4. Once a person is able to start understanding systems at different scales, it can reveal additional elements and relationships one might not have picked up at the start of the exercise

5. Understanding dynamic behaviour can allow the incorporation of dynamic models when creating a conceptual system

6. Knowledge of the different scales at which a system operates can help in enhancing the accuracy of any conceptual models one might like to use

All the four major elements in the diagramme strongly improve one's ability to identify systems, predict their behaviours, and devise modifications if and when needed.

Definition of systems thinking

So now that we have come this far, we can understand the definition of systems thinking:

Systems thinking is a set of synergistic analytic skills used to improve the capability of identifying and understanding systems, predicting their behaviors, and devising modifications to them in order to produce desired effects. These skills work together as a system. - Arnold and Wade (2015)

These analytical skills and their interconnections are easy enough to understand, and they make complete sense when seen from the lens of systems thinking itself.

In the next post in this series, we shall look at some systems thinking frameworks used in healthcare followed by an intervention strategy informed by the same.

1. Arnold, R.D. and Wade, J.P., 2015. A definition of systems thinking: A systems approach. Procedia computer science, 44(2015), pp.669-678. ↩︎

2. Stave, K. and Hopper, M., 2007, July. What constitutes systems thinking? A proposed taxonomy. In 25th International Conference of the System Dynamics Society. ↩︎