Chris Sampson’s journal round-up for 20th May 2019

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.

A new method to determine the optimal willingness to pay in cost-effectiveness analysis. Value in Health Published 17th May 2019

Efforts to identify a robust estimate of the willingness to pay for a QALY have floundered. Mostly, these efforts have relied on asking people about their willingness to pay. In the UK, we have moved away from using such estimates as a basis for setting cost-effectiveness thresholds in the context of resource allocation decisions. Instead, we have attempted to identify the opportunity cost of a QALY, which is perhaps even more difficult, but more easy to justify in the context of a fixed budget. This paper seeks to inject new life into the willingness-to-pay approach by developing a method based on relative risk aversion.

The author outlines the relationship between relative risk aversion and the rate at which willingness-to-pay changes with income. Various candidate utility functions are described with respect to risk preferences, with a Weibull function being adopted for this framework. Estimates of relative risk aversion have been derived from numerous data sources, including labour supply, lottery experiments, and happiness surveys. These estimates from the literature are used to demonstrate the relationship between relative risk aversion and the ‘optimal’ willingness to pay (K), calibrated using the Weibull utility function. For an individual with ‘representative’ parameters plugged into their utility function, K is around twice the income level. K always increases with relative risk aversion.

Various normative questions are raised, including whether a uniform K should be adopted for everybody within the population, and whether individuals should be able to spend on health care on top of public provision. This approach certainly appears to be more straightforward than other approaches to estimating willingness-to-pay in health care, and may be well-suited to decentralised (US-style) resource allocation decision-making. It’s difficult to see how this framework could gain traction in the UK, but it’s good to see alternative approaches being proposed and I hope to see this work developed further.

Striving for a societal perspective: a framework for economic evaluations when costs and effects fall on multiple sectors and decision makers. Applied Health Economics and Health Policy [PubMed] Published 16th May 2019

I’ve always been sceptical of a ‘societal perspective’ in economic evaluation, and I have written in favour of a limited health care perspective. This is mostly for practical reasons. Being sufficiently exhaustive to identify a truly ‘societal’ perspective is so difficult that, in attempting to do so, there is a very high chance that you will produce estimates that are so inaccurate and imprecise that they are more dangerous than useful. But the fact is that there is no single decision-maker when it comes to public expenditure. Governments are made up of various departments, within which there are many levels and divisions. Not everybody will care about the health care perspective, so other objectives ought to be taken into account.

The purpose of this paper is to build on the idea of the ‘impact inventory’, described by the Second Panel on Cost-Effectiveness in Health and Medicine, which sought to address the challenge of multiple objectives. The extended framework described in this paper captures effects and opportunity costs associated with an intervention within various dimensions. These dimensions could (or should) align with decision-makers’ objectives. Trade-offs invariably require aggregation, and this aggregation could take place either within individuals or within dimensions – something not addressed by the Second Panel. The authors describe the implications of each approach to aggregation, providing visual representations of the impact inventory in each case. Aggregating within individuals requires a normative judgement about how each dimension is valued to the individual and then a judgement about how to aggregate for overall population net benefit. Aggregating across individuals within dimensions requires similar normative judgements. Where the chosen aggregation functions are linear and additive, both approaches will give the same results. But as soon as we start to consider equity concerns or more complex aggregation, we’ll see different decisions being indicated.

The authors adopt an example used by the Second Panel to demonstrate the decisions that would be made within a health-only perspective and then decisions that consider other dimensions. There could be a simple extension beyond health, such as including the impact on individuals’ consumption of other goods. Or it could be more complex, incorporating multiple dimensions, sectors, and decision-makers. For the more complex situation, the authors consider the inclusion of the criminal justice sector, introducing the number of crimes averted as an object of value.

It’s useful to think about the limitations of the Second Panel’s framing of the impact inventory and to make explicit the normative judgements involved. What this paper seems to be saying is that cross-sector decision-making is too complex to be adequately addressed by the Second Panel’s impact inventory. The framework described in this paper may be too abstract to be practically useful, and too vague to be foundational. But the complexities and challenges in multi-sector economic evaluation need to be spelt out – there is no simple solution.

Advanced data visualisation in health economics and outcomes research: opportunities and challenges. Applied Health Economics and Health Policy [PubMed] Published 4th May 2019

Computers can make your research findings look cool, which can help make people pay attention. But data visualisation can also be used as part of the research process and provide a means of more intuitively (and accurately) communicating research findings. The data sets used by health economists are getting bigger, which provides more opportunity and need for effective visualisation. The authors of this paper suggest that data visualisation techniques could be more widely adopted in our field, but that there are challenges and potential pitfalls to consider.

Decision modelling is an obvious context in which to use data visualisation, because models tend to involve large numbers of simulations. Dynamic visualisations can provide a means by which to better understand what is going on in these simulations, particularly with respect to uncertainty in estimates associated with alternative model structures or parameters. If paired with interactive models and customised dashboards, visualisation can make complex models accessible to non-expert users. Communicating patient outcomes data is also highlighted as a potential application, aiding the characterisation of differences between groups of individuals and alternative outcome measures.

Yet, there are barriers to wider use of visualisation. There is some scepticism about bias in underlying analyses, and end users don’t want to be bamboozled by snazzy graphics. The fact that journal articles are still the primary mode of communicating research findings is a problem, as you can’t have dynamic visualisations in a PDF. There’s also a learning curve for analysts wishing to develop complex visualisations. Hopefully, opportunities will be identified for two-way learning between the health economics world and data scientists more accustomed to data visualisation.

The authors provide several examples (static in the publication, but with links to live tools), to demonstrate the types of visualisations that can be created. Generally speaking, complex visualisations are proposed as complements to our traditional presentations of results, such as cost-effectiveness acceptability curves, rather than as alternatives. The key thing is to maintain credibility by ensuring that data visualisation is used to describe data in a more accurate and meaningful way, and to avoid exaggeration of research findings. It probably won’t be long until we see a set of good practice guidelines being developed for our field.


Thesis Thursday: David Mott

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 David Mott who has a PhD from Newcastle University. If you would like to suggest a candidate for an upcoming Thesis Thursday, get in touch.

How do preferences for public health interventions differ? A case study using a weight loss maintenance intervention
Luke Vale, Laura Ternent
Repository link

Why is it important to understand variation in people’s preferences?

It’s not all that surprising that people’s preferences for health care interventions vary, but we don’t have a great understanding of what might drive these differences. Increasingly, preference information is being used to support regulatory decisions and, to a lesser but increasing extent, health technology assessments. It could be the case that certain subgroups of individuals would not accept the risks associated with a particular health care intervention, whereas others would. Therefore, identifying differences in preferences is important. However, it’s also useful to try to understand why this heterogeneity might occur in the first place.

The debate on whose preferences to elicit for health state valuation has traditionally focused on those with experience (e.g. patients) and those without (e.g. the general population). Though this dichotomy is problematic; it has been shown that health state utilities systematically differ between these two groups, presumably due to the difference in relative experience. My project aimed to explore whether experience also affects people’s preferences for health care interventions.

How did you identify different groups of people, whose preferences might differ?

The initial plan for the project was to elicit preferences for a health care intervention from general population and patient samples. However, after reviewing the literature, it seemed highly unlikely that anyone would advocate for preferences for treatments to be elicited from general population samples. It has long been suggested that discrete choice experiments (DCEs) could be used to incorporate patient preferences into decision-making, and it turned out that patients were the focus of the majority of the DCE studies that I reviewed. Given this, I took a more granular approach in my empirical work.

We recruited a very experienced group of ‘service users’ from a randomised controlled trial (RCT). In this case, it was a novel weight loss maintenance intervention aimed at helping obese adults that had lost at least 5% of their overall weight to maintain their weight loss. We also recruited an additional three groups from an online panel. The first group were ‘potential service users’ – those that met the trial criteria but could not have experienced the intervention. The second group were ‘potential beneficiaries’ – those that were obese or overweight and did not meet the trial criteria. The final group were ‘non-users’ – those with a normal BMI.

What can your study tell us about preferences in the context of a weight loss maintenance intervention?

The empirical part of my study involved a DCE and an open-ended contingent valuation (CV) task. The DCE was focused on the delivery of the trial intervention, which was a technology-assisted behavioural intervention. It had a number of different components but, briefly, it involved participants weighing themselves regularly on a set of ‘smart scales’, which enabled the trial team to access and monitor the data. Participants received text messages from the trial team with feedback, reminders to weigh themselves (if necessary), and links to online tools and content to support the maintenance of their weight loss.

The DCE results suggested that preferences for the various components of the intervention varied significantly between individuals and between the different groups – and not all were important. In contrast, the efficacy and cost attributes were important across the board. The CV results suggested that a very significant proportion of individuals would be willing to pay for an effective intervention (i.e. that avoided weight regain), with very few respondents expressing a willingness to pay for an intervention that led to more than 10-20% weight regain.

Do alternative methods for preference elicitation provide a consistent picture of variation in preferences?

Existing evidence suggests that willingness to pay (WTP) estimates from CV tasks might differ from those derived from DCE data, but there aren’t a lot of empirical studies on this in health. Comparisons were planned in my study, but the approach taken in the end was suboptimal and ultimately inconclusive. The original plan was to obtain WTP estimates for an entire WLM intervention using the DCE and to compare this with the estimates from the CV task. Due to data limitations, it wasn’t possible to make this comparison. However, the CV task was a bit unusual because we asked for respondents’ WTP at various different efficacy levels. So instead the comparison made was between average WTP values for a percentage point of weight re-gain. The differences were statistically insignificant.

Are some people’s preferences ‘better defined’ than others’?

We hypothesised that those with experience of the trial intervention would have ‘better defined’ preferences. To explore this, we compared the data quality across the different user groups. From a quick glance at the DCE results, it is pretty clear that the data were much better for the most experienced group; the coefficients were larger, and a much higher proportion was statistically significant. However, more interestingly, we found that the most experienced group were 23% more likely to have passed all of the rationality tests that were embedded in the DCE. Therefore, if you accept that better quality data is an indicator of ‘better defined’ preferences, then the data do seem reasonably supportive of the hypothesis. That being said, there were no significant differences between the other three groups, begging the question: was it the difference in experience, or some other difference between RCT participants and online panel respondents?

What does your research imply for the use of preferences in resource allocation decisions?

While there are still many unanswered questions, and there is always a need for further research, the results from my PhD project suggest that preferences for health care interventions can differ significantly between respondents with differing levels of experience. Had my project been applied to a more clinical intervention that is harder for an average person to imagine experiencing, I would expect the differences to have been much larger. I’d love to see more research in this area in future, especially in the context of benefit-risk trade-offs.

The key message is that the level of experience of the participants matters. It is quite reasonable to believe that a preference study focusing on a particular subgroup of patients will not be generalisable to the broader patient population. As preference data, typically elicited from patients, is increasingly being used in decision-making – which is great – it is becoming increasingly important for researchers to make sure that their respondent samples are appropriate to support the decisions that are being made.

Chris Sampson’s journal round-up for 15th October 2018

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.

Reliability and validity of the contingent valuation method for estimating willingness to pay: a case of in vitro fertilisation. Applied Health Economics and Health Policy [PubMed] Published 13th October 2018

In vitro fertilisation (IVF) is a challenge for standard models of valuation in health economics. Mostly, that’s because, despite it falling within the scope of health care, and despite infertility being a health problem, many of the benefits of IVF can’t be considered health-specific. QALYs can’t really do the job, so there’s arguably a role for cost-benefit analysis, and for using stated preference methods to determine the value of IVF. This study adds to an existing literature studying willingness to pay for IVF, but differs in that it tries to identify willingness to pay (WTP) from the general population. This study is set in Australia, where IVF is part-funded by universal health insurance, so asking the public is arguably the right thing to do.

Three contingent valuation surveys were conducted online with 1,870 people from the general public. The first survey used a starting point bid of $10,000, and then, 10 months later, two more surveys were conducted with starting point bids of $4,000 and $10,000. Each included questions for a 10%, 20%, and 50% success rate. Respondents were asked to adopt an ex-post perspective, assuming that they were infertile and could conceive by IVF. Individuals could respond to starting bids with ‘yes’, ‘no’, ‘not sure’, or ‘I am not willing to pay anything’. WTP for one IVF cycle with a 20% success rate ranged from $6,353 in the $4,000 survey to $11,750 in the first $10,000 survey. WTP for a year of treatment ranged from $18,433 to $28,117. The method was reliable insofar as there were no differences between the first and second $10,000 surveys. WTP values corresponded to the probability of success, providing support for the internal construct validity of the survey. However, the big difference between values derived using the alternative starting point bids indicates a strong anchoring bias. The authors also tested the external criterion validity by comparing the number of respondents willing to pay more than $4,000 for a cycle with a 20% success rate (roughly equivalent to the out of pocket cost in Australia) with the number of people who actually choose to pay for IVF in Australia. Around 63% of respondents were willing to pay at that price, which is close to the estimated 60% in Australia.

This study provides some support for the use of contingent valuation methods in the context of IVF, and for its use in general population samples. But the anchoring effect is worrying and justifies further research to identify appropriate methods to counteract this bias. The exclusion of the “not sure” and “I will not pay anything” responses from the analysis – as ‘non-demanders’ – arguably undermines the ‘societal valuation’ aspect of the estimates.

Pharmaceutical expenditure and gross domestic product: evidence of simultaneous effects using a two‐step instrumental variables strategy. Health Economics [PubMed] Published 10th October 2018

The question of how governments determine spending on medicines is pertinent in the UK right now, as the Pharmaceutical Price Regulation Scheme approaches its renewal date. The current agreement includes a cap on pharmaceutical expenditure. It should go without saying that GDP ought to have some influence on how much public spending is dedicated to medicines. But, when medicines expenditure might also influence GDP, the actual relationship is difficult to estimate. In this paper, the authors seek to identify both effects: the income elasticity of government spending on pharmaceuticals and the effect of that spending on income.

The authors use a variety of data sources from the World Health Organization, World Bank, and International Monetary Fund to construct an unbalanced panel for 136 countries from 1995 to 2006. To get around the challenge of two-way causality, the authors implement a two-step instrumental variable approach. In the first step of the procedure, a model estimates the impact of GDP per capita on government spending on pharmaceuticals. International tourist receipts are used as an instrument that is expected to correlate strongly with GDP per capita, but which is expected to be unrelated to medicines expenditure (except through its correlation with GDP). The model attempts to control for health care expenditure, life expectancy, and other important country-specific variables. In the second step, a reverse causality model is used to assess the impact of pharmaceutical expenditure on GDP per capita, with pharmaceutical expenditure adjusted to partial-out the response to GDP estimated in the first step.

The headline average results are that GDP increases pharmaceutical expenditure and that pharmaceutical expenditure reduces GDP. A 1% increase in GDP per capita increases public pharmaceutical expenditure per capita by 1.4%, suggesting that pharmaceuticals are a luxury good. A 1% increase in public pharmaceutical expenditure is associated with a 0.09% decrease in GDP per capita. But the results are more nuanced than that. The authors outline various sources of heterogeneity. The positive effect of GDP on pharmaceutical expenditure only holds for high-income countries and the negative effect of pharmaceutical expenditure on GDP only holds for low-income countries. Quantile regressions show that income elasticity decreases for higher quantiles of expenditure. GDP only influences pharmaceutical spending in countries classified as ‘free’ on the index of Economic Freedom of the World, and pharmaceutical expenditure only has a negative impact on GDP in countries that are ‘not free’.

I’ve never come across this kind of two-step approach before, so I’m still trying to get my head around whether the methods and the data are adequate. But a series of robustness checks provide some reassurance. In particular, an analysis of intertemporal effects using lagged GDP and lagged pharmaceutical expenditure demonstrates the robustness of the main findings. Arguably, the findings of this study are more important for policymaking in low- and middle-income countries, where pharmaceutical expenditures might have important consequences for GDP. In high-income (and ‘free’) economies that spend a lot on medicines, like the UK, there is probably less at stake. This could be because of effective price regulation and monitoring, and better adherence, ensuring that pharmaceutical expenditure is not wasteful.

Parental health spillover in cost-effectiveness analysis: evidence from self-harming adolescents in England. PharmacoEconomics [PubMed] [RePEc] Published 8th October 2018

Any intervention has the potential for spillover effects, whereby people other than the recipient of care are positively or negatively affected by the consequences of the intervention. Where a child is the recipient of care, it stands to reason that any intervention could affect the well-being of the parents and that these impacts should be considered in economic evaluation. But how should parental spillovers be incorporated? Are parental utilities additive to that of the child patient? Or should a multiplier effect be used with reference to the effect of an intervention on the child’s utility?

The study reports on a trial-based economic evaluation of family therapy for self-harming adolescents aged 11-17. Data collection included EQ-5D-3L for the adolescents and HUI2 for the main caregiver (86% mothers) at baseline, 6-month follow-up, and 12-month follow-up, collected from 731 patient-parent pairs. The authors outline six alternative methods for including parental health spillovers: i) relative health spillover, ii) relative health spillover per treatment arm, iii) absolute health spillover, iv) absolute global health spillover per treatment arm, v) additive accrued health benefits, and vi) household equivalence scales. These differ according to whether parental utility is counted as depending on adolescent’s utility, treatment allocation, the primary outcome of the study, or some combination thereof. But the authors’ primary focus (and the main contribution of this study) is the equivalence scale option. This involves adding together the spillover effects for other members of the household and using alternative weightings depending on the importance of parental utility compared with adolescent utility.

Using Tobit models, controlling for a variety of factors, the authors demonstrate that parental utility is associated with adolescent utility. Then, economic evaluations are conducted using each of the alternative spillover accounting methods. The base case of including only adolescents’ utility delivers an ICER of around £40,453. Employing the alternative methods gives quite different results, with the intervention dominated in two of the cases and an ICER below £30,000 per QALY in others. For the equivalence scale approach, the authors employ several elasticities for spillover utility, ranging from 0 (where parental utility is of equivalent value to adolescent utility and therefore additive) to 1 (where the average health spillover per household member is estimated for each patient). The ICER estimates using the equivalence scale approach ranged from £27,166 to £32,504. Higher elasticity implied lower cumulated QALYs.

The paper’s contribution is methodological, and I wouldn’t read too much into the magnitude of the results. For starters, the use of HUI2 (a measure for children) in adults and the use of EQ-5D-3L (a measure for adults) in the children is somewhat confusing. The authors argue that health gains should only be aggregated at the household level if the QALY gain for the patient is greater or equal to zero, because the purpose of treatment is to benefit the adolescents, not the parents. And they argue in favour of using an equivalence scale approach. By requiring an explicit judgement to set the elasticity within the estimation, the method provides a useful and transparent approach to including parental spillovers.