Chris Sampson’s journal round-up for 19th June 2017

Every Monday our authors provide a round-up of some of the most recently published peer reviewed articles from the field. We don’t cover everything, or even what’s most important – just a few papers that have interested the author. Visit our Resources page for links to more journals or follow the HealthEconBot. If you’d like to write one of our weekly journal round-ups, get in touch.

Health-related resource-use measurement instruments for intersectoral costs and benefits in the education and criminal justice sectors. PharmacoEconomics [PubMed] Published 8th June 2017

Increasingly, people are embracing a societal perspective for economic evaluation. This often requires the identification of costs (and benefits) in non-health sectors such as education and criminal justice. But it feels as if we aren’t as well-versed in capturing these as we are in the health sector. This study reviews the measures that are available to support a broader perspective. The authors search the Database of Instruments for Resource Use Measurement (DIRUM) as well as the usual electronic journal databases. The review also sought to identify the validity and reliability of the instruments. From 167 papers assessed in the review, 26 different measures were identified (half of which were in DIRUM). 21 of the instruments were only used in one study. Half of the measures included items relating to the criminal justice sector, while 21 included education-related items. Common specifics for education included time missed at school, tutoring needs, classroom assistance and attendance at a special school. Criminal justice sector items tended to include legal assistance, prison detainment, court appearances, probation and police contacts. Assessments of the psychometric properties was found for only 7 of the 26 measures, with specific details on the non-health items available for just 2: test-retest reliability for the Child and Adolescent Services Assessment (CASA) and validity for the WPAI+CIQ:SHP,V2 (there isn’t room on the Internet for the full name). So there isn’t much evidence of any validity for any of these measures in the context of intersectoral (non-health) costs and benefits. It’s no doubt the case that health-specific resource use measures aren’t subject to adequate testing, but this study has identified that the problem may be even greater when it comes to intersectoral costs and benefits. Most worrying, perhaps, is the fact that 1 in 5 of the articles identified in the review reported using some unspecified instrument, presumably developed specifically for the study or adapted from an off-the-shelf instrument. The authors propose that a new resource use measure for intersectoral costs and benefits (RUM ICB) be developed from scratch, with reference to existing measures and guidance from experts in education and criminal justice.

Use of large-scale HRQoL datasets to generate individualised predictions and inform patients about the likely benefit of surgery. Quality of Life Research [PubMed] Published 31st May 2017

In the NHS, EQ-5D data are now routinely collected from patients before and after undergoing one of four common procedures. These data can be used to see how much patients’ health improves (or deteriorates) following the operations. However, at the individual level, for a person deciding whether or not to undergo the procedure, aggregate outcomes might not be all that useful. This study relates to the development of a nifty online tool that a prospective patient can use to find out the expected likelihood that they will feel better, the same or worse following the procedure. The data used include EQ-5D-3L responses associated with almost half a million unilateral hip or knee replacements or groin hernia repairs between April 2009 and March 2016. Other variables are also included, and central to this analysis is a Likert scale about improvement or worsening of hip/knee/hernia problems compared to before the operation. The purpose of the study is to group people – based on their pre-operation characteristics – according to their expected postoperative utility scores. The authors employed a recursive Classification and Regression Tree (CART) algorithm to split the datasets into strata according to the risk factors. The final set of risk variables were age, gender, pre-operative EQ-5D-3L profile and symptom duration. The CART analysis grouped people into between 55 and 60 different groups for each of the procedures, with the groupings explaining 14-27% of the variation in postoperative utility scores. Minimally important (positive and negative) differences in the EQ-5D utility score were estimated with reference to changes in the Likert scale for each of the procedures. These ranged in magnitude from 0.041 to 0.106. The resulting algorithms are what drive the results delivered by the online interface (you can go and have a play with it). There are a few limitations to the study, such as the reliance on complete case analysis and the fact that the CART analysis might lack predictive ability. And there’s an interesting problem inherent in all of this, that the more people use the tool, the less representative it will become as it influences selection into treatment. The validity of the tool as a precise risk calculator is quite limited. But that isn’t really the point. The point is that it unlocks some of the potential value of PROMs to provide meaningful guidance in the process of shared decision-making.

Can present biasedness explain early onset of diabetes and subsequent disease progression? Exploring causal inference by linking survey and register data. Social Science & Medicine [PubMed] Published 26th May 2017

The term ‘irrational’ is overused by economists. But one situation in which I am willing to accept it is with respect to excessive present bias. That people don’t pay enough attention to future outcomes seems to be a fundamental limitation of the human brain in the 21st century. When it comes to diabetes and its complications, there are lots of treatments available, but there is only so much that doctors can do. A lot depends on the patient managing their own disease, and it stands to reason that present bias might cause people to manage their diabetes poorly, as the value of not going blind or losing a foot 20 years in the future seems less salient than the joy of eating your own weight in carbs right now. But there’s a question of causality here; does the kind of behaviour associated with time-inconsistent preferences lead to poorer health or vice versa? This study provides some insight on that front. The authors outline an expected utility model with quasi-hyperbolic discounting and probability weighting, and incorporate a present bias coefficient attached to payoffs occurring in the future. Postal questionnaires were collected from 1031 type 2 diabetes patients in Denmark with an online discrete choice experiment as a follow-up. These data were combined with data from a registry of around 9000 diabetes patients, from which the postal/online participants were identified. BMI, HbA1c, age and year of diabetes onset were all available in the registry and the postal survey included physical activity, smoking, EQ-5D, diabetes literacy and education. The DCE was designed to elicit time preferences using the offer of (monetary) lottery wins, with 12 different choice sets presented to all participants. Unfortunately, despite the offer of a real-life lottery award for taking part in the research, only 79 of 1031 completed the online DCE survey. Regression analyses showed that individuals with diabetes since 1999 or earlier, or who were 48 or younger at the time of onset, exhibited present bias. And the present bias seems to be causal. Being inactive, obese, diabetes illiterate and having lower quality of life or poorer glycaemic control were associated with being present biased. These relationships hold when subject to a number of control measures. So it looks as if present bias explains at least part of the variation in self-management and health outcomes for people with diabetes. Clearly, the selection of the small sample is a bit of a concern. It may have meant that people with particular risk preferences (given that the reward was a lottery) were excluded, and so the sample might not be representative. Nevertheless, it seems that at least some people with diabetes could benefit from interventions that increase the salience of future health-related payoffs associated with self-management.

Credits

Advertisements

Thesis Thursday: Miqdad Asaria

On the third Thursday of every month, we speak to a recent graduate about their thesis and their studies. This month’s guest is Dr Miqdad Asaria who graduated with a PhD from the University of York. If you would like to suggest a candidate for an upcoming Thesis Thursday, get in touch.

Title
The economics of health inequality in the English National Health Service
Supervisors
Richard Cookson, Tim Doran
Repository link
http://etheses.whiterose.ac.uk/16189

What types of inequality are relevant in the context of the NHS?

For me the inequalities that really matter are the inequalities in health outcomes, in the English context it is particularly the socioeconomic patterning of these inequalities that is of concern. The focus of health policy in England over the last 200 years has been on improving the average health of the population as well as on providing financial risk protection against catastrophic health expenditure. Whilst great strides have been made in improving average population health through various pioneering interventions including the establishment of the NHS, health inequality has in fact consistently widened over this period. Recent research suggests that in terms of quality-adjusted life expectancy the gap between people living in the most deprived fifth of neighbourhoods in the country as compared to those living in the most affluent fifth is now approximately 11 quality-adjusted life years.

However, these socio-economic inequalities in health typically accumulate across the life course and there is a limited amount that health care on its own can do to prevent these gaps from widening or indeed to close these gaps once they emerge. This is why health systems including the NHS typically focus on measuring and tackling the inequalities that they can influence even though eliminating such inequalities can have at best only modest impacts on reducing health inequality overall. These comprise of inequalities in access to and quality of healthcare as well as inequality of those health outcomes specifically amenable to healthcare.

What were the key methods and data that you used to identify levels of health inequality?

I am currently working on a project with the Ministry of Health and Family Welfare in India and it is really making me appreciate the amazingly detailed and comprehensive administrative datasets available to researchers in England. For the work underpinning my thesis I linked 10 years of data looking at every hospital admission and outpatient visit in the country with the quality and outcomes achieved for patients registered at each primary care practice, the number of doctors working at each primary care practice, general population census data, cause-specific mortality data, hospital cost data and deprivation data all at neighbourhood level. I spent a lot of time assembling, cleaning and linking these data sets and then used this data platform to build a range of health inequality indicators – some of which can be seen in an interactive tool I built to present the data to clinical commissioning groups.

As well as measuring inequality retrospectively in order to provide evidence to evaluate past NHS policies, and building tools to enable the NHS to monitor inequality going forward, another key focus of my thesis was to develop methods to model and incorporate health inequality impacts into cost-effectiveness analysis. These methods allow analysts to evaluate proposed health interventions in terms of their impact on the distribution of health rather than just their impact on the mythical average citizen. The distributional cost-effectiveness analysis framework I developed is based on the idea of using social welfare functions to evaluate the estimated health distributions arising from the rollout of different health care interventions and compute the equity-efficiency trade-offs that would need to be made in order to prefer one intervention over another. A key parameter in this analysis required in order to make equity-efficiency trade-offs is the level of health inequality aversion. This parameter was quite tricky to estimate with methods used to elicit it from the general public being prone to various framing effects. The preliminary estimates that I used in my analysis for this parameter suggested that at the margin the general public thought people living in the most deprived fifth of neighbourhoods in the country deserve approximately 7 times the priority in terms of health care spending as those who live in the most affluent fifth of neighbourhoods.

Does your PhD work enable us to attach a ‘cost’ to inequality, and ‘value’ to policies that reduce it?

As budding economists, we are ever cautious to distinguish association and causation. My thesis starts by estimating the cost associated with inequality to the NHS. That is the additional cost to the NHS spent on treating the excess morbidity in those living in relatively deprived neighbourhoods. I estimated the difference between the actual NHS hospital budget and what the cost would have been if everybody in the country had the morbidity profile of those who live in just the most affluent fifth of neighbourhoods. For inpatient hospital costs this difference came to £4.8 billion per year and widening this to all NHS costs this came to £12.5 billion per year approximately a fifth of the total NHS budget. I looked both cross-sectionally and also modelled lifetime estimated health care use and found that even over their entire lifetimes people living in more deprived neighbourhoods consumed more health care despite their substantially shorter life expectancies.

This cost is of course very different to the value of policies to reduce inequality. This difference arises for two main reasons. First, my estimates were not causal but rather associations so we are unable to conclude that reducing socioeconomic inequality would actually result in everybody in the country gaining the morbidity profile of those living in the most affluent fifth of neighbourhoods. Second and perhaps more significantly, my estimates do not value any of the health benefits that would result from reducing health inequality they just count the costs that could be saved by the NHS due to the excess morbidity avoided. The value of these health benefits forgone in terms of quality adjusted life years gained would have to be converted into monetary terms using an estimate of willingness to pay for health and added to these cost savings (which themselves would need to be converted to consumption values) to get a total value of reducing inequality from a health perspective. There would also, of course, be a range of non-health impacts of reducing inequality that would need to be accounted for if this exercise were to be comprehensively conducted.

In simple terms, if the causal link between socioeconomic inequality and health could be determined then the value to the health sector of policies that could substantially reduce this inequality would likely be far greater than the costs quoted here.

How did you find the PhD-by-publication route? Would you recommend it?

I came to academia relatively late having previously worked in both the government and the private sector for a number of years. The PhD by publication route suited me well as it allowed me to get stuck into a number of projects, work with a wide range of academics and build an academic career whilst simultaneously curating a set of papers to submit as a thesis. However, it is certainly not the fastest way to achieve PhD status, my thesis took 6 years to compile. The publication route is also still relatively uncommon in England and I found both my supervisors and examiners somewhat perplexed about how to approach it. Additionally, my wife who did her PhD by the traditional route assures me that it is not a ‘proper’ PhD!

For those fresh out of an MSc programme the traditional route probably works well, giving you the opportunity to develop research skills and focus on one area in depth with lots of guidance from a dedicated supervisor. However, for people like me who probably would never have got around to doing a traditional PhD, it is nice that there is an alternative way to acquire the ‘Dr’ title which I am finding confers many unanticipated benefits.

What advice would you give to a researcher looking to study health inequality?

The most important thing that I have learnt from my research is that health inequality, particularly in England, has very little to do with health care and everything to do with socioeconomic inequality. I would encourage researchers interested in this area to look at broader interventions tackling the social determinants of health. There is lots of exciting work going on at the moment around basic income and social housing as well as around the intersection between the environment and health which I would love to get stuck into given the chance.

Why insurance works better with some adverse selection

Adverse selection, a process whereby low-risk individuals drop out of the insurance pool, leaving only high-risk individuals, arises when the individuals purchasing insurance have better information regarding their risk status than does the insurer. […] In the limit, adverse selection can make insurance markets unsustainable. Even short of the market disappearing altogether… The market cannot offer a full set of insurance contracts, reducing allocative efficiency.

The story summarised above (by Jeremiah Hurley) is familiar to all health economists. Adverse selection is generally understood to be a universal problem for efficiency in health insurance (and indeed all insurance), which should always be avoided or minimised, or else traded off against other objectives of equity. In my book, Loss Coverage: Why Insurance Works Better with Some Adverse Selection, I put forward a contrary argument that a modest degree of adverse selection in insurance can increase efficiency.

My argument depends on two departures from canonical models of insurance, both realistic. First, I assume that not all individuals will buy insurance when it is risk-rated; this is justified by observation of extant markets (e.g. around 10% of the US population has no health insurance, and around 50% have no life insurance). Second, my criterion of efficiency is based not on Pareto optimality (unsatisfactory because it says so little) or utilities (unsatisfactory because always unobservable), but on ‘loss coverage.’

In its simplest form, loss coverage is the expected fraction of the population’s losses which is compensated by insurance.

Since the purpose of insurance is to compensate the population’s losses, I argue that higher loss coverage is more efficient than lower loss coverage. Under this criterion, insurance of one high risk will contribute more to efficiency than insurance of one low risk. This is intuitively reasonable: higher risks are those who most need insurance!

If this intuition is accepted, the orthodox arguments about adverse selection seem to overlook one point. True, adverse selection leads to a higher average price for insurance and a fall in numbers of individuals insured. But it also leads to a shift in coverage towards higher risks (those who need insurance most). If this shift in coverage is large enough, it can more than outweigh the fall in numbers insured, so that loss coverage is increased.

My argument can be illustrated by the following toy example. The numbers are simplified and exaggerated for clarity, but the underlying argument is quite general.

Consider a population of just ten risks (say lives), with three alternative scenarios for insurance risk classification: risk-differentiated premiums, pooled premiums (with some adverse selection), and pooled premiums (with severe adverse selection). Assume that all losses and insurance cover are for unit amounts (this simplifies the discussion, but it is not necessary).

The three scenarios are represented in the three panels of the illustration. Each ‘H’ represents one higher risk and each ‘L’ represents one lower risk. The population has the typical predominance of lower risks: a lower risk-group of eight risks each with probability of loss 0.01, and a higher risk-group of two risks each with probability of loss 0.04.

In Scenario 1, risk-differentiated premiums (actuarially fair premiums) are charged. The demand response of each risk-group to an actuarially fair price is the same: exactly half the members of each risk-group buy insurance. The shading shows that a total of five risks buy insurance.

Scenario 1

 

The weighted average of the premiums paid is (4 x 0.01 +1 x 0.04)/5 = 0.016. Since higher and lower risks are insured in the same proportions as they exist in the population, there is no adverse selection.

Exactly half the population’s expected losses are compensated by insurance. I describe this as ‘loss coverage’ of 50%. (The calculation is (4 x 0.01 + 1x 0.04) / (8 x 0.01 + 2 x 0.04) = 0.50.)

In Scenario 2, risk classification has been banned, and so insurers have to charge a common pooled premium to both higher and lower risks. Higher risks buy more insurance, and lower risks buy less (adverse selection). The pooled premium is set as the weighted average of the true risks, so that expected profits on low risks exactly offset expected losses on high risks. This weighted average premium is (1 x 0.01 +2 x 0.04)/3 = 0.03. The shading symbolises that that three risks (compared with five previously) buy insurance.

Scenario 2

 

Note that the weighted average premium is higher in Scenario 2, and the number of risks insured is lower. These are the essential features of adverse selection, which Scenario 2 accurately and completely represents. But there is a surprise: despite the adverse selection in Scenario 2, the expected losses compensated by insurance for the whole population are now higher. That is, 56% of the population’s expected losses are now compensated by insurance, compared with 50% before. (The calculation is (1 x 0.01 + 2 x 0.04) / (8x 0.01 + 2 x 0.04) = 0.56.)

I argue that Scenario 2, with a higher expected fraction of the population’s losses compensated by insurance – higher loss coverage – is more efficient than Scenario 1. The superiority of Scenario 2 arises not despite adverse selection, but because of adverse selection.

At this point an economist might typically retort that that the lower numbers insured in Scenario 2 compared with Scenario 1 is suggestive of lower efficiency. However, it seems surprising that an arrangement such as Scenario 2, under which more risk is voluntarily traded and more losses are compensated, is always disparaged as less efficient.

A ban on risk classification can also reduce loss coverage, if the adverse selection which the ban induces becomes too severe. This possibility is illustrated in Scenario 3. Adverse selection has progressed to the point where only one higher risk, and no lower risks, buys insurance. The expected losses compensated by insurance for the whole population are now lower. That is, 25% of the population’s expected losses are now compensated by insurance, compared with 50% in Scenario 1, and 56% in Scenario 2. (The calculation is (1 x 0.04) / (8x 0.01 + 2 x 0.04) = 0.25.)

Scenario 3

 

These scenarios suggest that banning risk classification can increase loss coverage if it induces the `right amount’ of adverse selection (Scenario 2), but reduce loss coverage if it generates `too much’ adverse selection (Scenario 3). Which of Scenario 2 or Scenario 3 actually prevails depends on the demand elasticities of higher and lower risks.

The argument illustrated by the toy example applies broadly. It does not depend on any unusual choice of numbers for the example. The key idea is that loss coverage – and hence, I argue, efficiency – is increased by a modest degree of adverse selection.