Sam Watson’s journal round-up for 25th June 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.

The efficiency of slacking off: evidence from the emergency department. Econometrica [RePEc] Published May 2018

Scheduling workers is a complex task, especially in large organisations such as hospitals. Not only should one consider when different shifts start throughout the day, but also how work is divided up over the course of each shift. Physicians, like anyone else, value their leisure time and want to go home at the end of a shift. Given how they value this leisure time, as the end of a shift approaches physicians may behave differently. This paper explores how doctors in an emergency department behave at ‘end of shift’, in particular looking at whether doctors ‘slack off’ by accepting fewer patients or tasks and also whether they rush to finish those tasks they have. Both cases can introduce inefficiency by either under-using their labour time or using resources too intensively to complete something. Immediately, from the plots of the raw data, it is possible to see a drop in patients ‘accepted’ both close to end of shift and close to the next shift beginning (if there is shift overlap). Most interestingly, after controlling for patient characteristics, time of day, and day of week, there is a decrease in the length of stay of patients accepted closer to the end of shift, which is ‘dose-dependent’ on time to end of shift. There are also marked increases in patient costs, orders, and inpatient admissions in the final hour of the shift. Assuming that only the number of patients assigned and not the type of patient changes over the course of a shift (a somewhat strong assumption despite the additional tests), then this would suggest that doctors are rushing care and potentially providing sub-optimal or inefficient care closer to the end of their shift. The paper goes on to explore optimal scheduling on the basis of the results, among other things, but ultimately shows an interesting, if not unexpected, pattern of physician behaviour. The results relate mainly to efficiency, but it’d be interesting to see how they relate to quality in the form of preventable errors.

Semiparametric estimation of longitudinal medical cost trajectory. Journal of the American Statistical Association Published 19th June 2018

Modern computational and statistical methods have opened up a range of statistical models to estimation hitherto inestimable. This includes complex latent variable structures, non-linear models, and non- and semi-parametric models. Recently we covered the use of splines for semi-parametric modelling in our Method of the Month series. Not that complexity is everything of course, but given this rich toolbox to more faithfully replicate the data generating process, one does wonder why the humble linear model estimated with OLS remains so common. Nevertheless, I digress. This paper addresses the problem of estimating the medical cost trajectory for a given disease from diagnosis to death. There are two key issues: (i) the trajectory is likely to be non-linear with costs probably increasing near death and possibly also be higher immediately after diagnosis (a U-shape), and (ii) we don’t observe the costs of those who die, i.e. there is right-censoring. Such a set-up is also applicable in other cases, for example looking at health outcomes in panel data with informative dropout. The authors model medical costs for each month post-diagnosis and time of censoring (death) by factorising their joint distribution into a marginal model for censoring and a conditional model for medical costs given the censoring time. The likelihood then has contributions from the observed medical costs and their times, and the times of the censored outcomes. We then just need to specify the individual models. For medical costs, they use a multivariate normal with mean function consisting of a bivariate spline of time and time of censoring. The time of censoring is modelled non-parametrically. This setup of the missing data problem is sometimes referred to as a pattern mixing model, in that the outcome is modelled as a mixture density over different populations dying at different times. The authors note another possibility for informative missing data, which was considered not to be estimable for complex non-linear structures, was the shared parameter model (to soon appear in another Method of the Month) that assumes outcomes and dropout are independent conditional on an underlying latent variable. This approach can be more flexible, especially in cases with varying treatment effects. One wonders if the mixed model representation of penalised splines wouldn’t fit nicely in a shared parameter framework and provide at least as good inferences. An idea for a future paper perhaps… Nevertheless, the authors illustrate their method by replicating the well-documented U-shaped costs from the time of diagnosis in patients with stage IV breast cancer.

Do environmental factors drive obesity? Evidence from international graduate students. Health Economics [PubMedPublished 21st June 2018

‘The environment’ can encompass any number of things including social interactions and networks, politics, green space, and pollution. Sometimes referred to as ‘neighbourhood effects’, the impact of the shared environment above and beyond the effect of individual risk factors is of great interest to researchers and policymakers alike. But there are a number of substantive issues that hinder estimation of neighbourhood effects. For example, social stratification into neighbourhoods likely means people living together are similar so it is difficult to compare like with like across neighbourhoods; trying to model neighbourhood choice will also, therefore, remove most of the variation in the data. Similarly, this lack of common support, i.e. overlap, between people from different neighbourhoods means estimated effects are not generalisable across the population. One way of getting around these problems is simply to randomise people to neighbourhoods. As odd as that sounds, that is what occurred in the Moving to Opportunity experiments and others. This paper takes a similar approach in trying to look at neighbourhood effects on the risk of obesity by looking at the effects of international students moving to different locales with different local obesity rates. The key identifying assumption is that the choice to move to different places is conditionally independent of the local obesity rate. This doesn’t seem a strong assumption – I’ve never heard a prospective student ask about the weight of our student body. Some analysis supports this claim. The raw data and some further modelling show a pretty strong and robust relationship between local obesity rates and weight gain of the international students. Given the complexity of the causes and correlates of obesity (see the crazy diagram in this post) it is hard to discern why certain environments contribute to obesity. The paper presents some weak evidence of differences in unhealthy behaviours between high and low obesity places – but this doesn’t quite get at the environmental link, such as whether these behaviours are shared through social networks or perhaps the structure and layout of the urban area, for example. Nevertheless, here is some strong evidence that living in an area where there are obese people means you’re more likely to become obese yourself.


Chris Sampson’s journal round-up for 5th March 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.

Healthy working days: the (positive) effect of work effort on occupational health from a human capital approach. Social Science & Medicine Published 28th February 2018

If you look at the literature on the determinants of subjective well-being (or happiness), you’ll see that unemployment is often cited as having a big negative impact. The same sometimes applies for its impact on health, but here – of course – the causality is difficult to tease apart. Then, in research that digs deeper, looking at hours worked and different types of jobs, we see less conclusive results. In this paper, the authors start by asserting that the standard approach in labour economics (on which I’m not qualified to comment) is to assume that there is a negative association between work effort and health. This study extends the framework by allowing for positive effects of work that are related to individuals’ characteristics and working conditions, and where health is determined in a Grossman-style model of health capital that accounts for work effort in the rate of health depreciation. This model is used to examine health as a function of work effort (as indicated by hours worked) in a single wave of the European Working Conditions Survey (EWCS) from 2010 for 15 EU member states. Key items from the EWCS included in this study are questions such as “does your work affect your health or not?”, “how is your health in general?”, and “how many hours do you usually work per week?”. Working conditions are taken into account by looking at data on shift working and the need to wear protective equipment. One of the main findings of the study is that – with good working conditions – greater work effort can improve health. The Marxist in me is not very satisfied with this. We need to ask the question, compared to what? Working fewer hours? For most people, that simply isn’t an option. Aren’t the people who work fewer hours the people who can afford to work fewer hours? No attention is given to the sociological aspects of employment, which are clearly important. The study also shows that overworking or having poorer working conditions reduces health. We also see that, for many groups, longer hours do not negatively impact on health until we reach around 120 hours a week. This fails a good sense check. Who are these people?! I’d be very interested to see if these findings hold for academics. That the key variables are self-reported undermines the conclusions somewhat, as we can expect people to adjust their expectations about work effort and health in accordance with their colleagues. It would be very difficult to avoid a type 2 error (with respect to the negative impact of effort on health) using these variables to represent health and the role of work effort.

Agreement between retrospectively and contemporaneously collected patient-reported outcome measures (PROMs) in hip and knee replacement patients. Quality of Life Research [PubMed] Published 26th February 2018

The use of patient-reported outcomes (PROMs) in elective care in the NHS has been a boon for researchers in our field, providing before-and-after measurement of health-related quality of life so that we can look at the impact of these interventions. But we can’t do this in emergency care because the ‘before’ is never observed – people only show up when they’re in the middle of the emergency. But what if people could accurately recall their pre-emergency health state? There’s some evidence to suggest that people can, so long as the recall period is short. This study looks at NHS PROMs data (n=443), with generic and condition-specific outcomes collected from patients having hip or knee replacements. Patients included in the study were additionally asked to recall their health state 4 weeks prior to surgery. The authors assess the extent to which the contemporary PROM measurements agree with the retrospective measurements, and the extent to which any disagreement relates to age, socioeconomic status, or the length of time to recall. There wasn’t much difference between contemporary and retrospective measurements, though patients reported slightly lower health on the retrospective questionnaires. And there weren’t any compelling differences associated with age or socioeconomic status or the length of recall. These findings are promising, suggesting that we might be able to rely on retrospective PROMs. But the elective surgery context is very different to the emergency context, and I don’t think we can expect the two types of health care to impact recollection in the same way. In this study, responses may also have been influenced by participants’ memories of completing the contemporary questionnaire, and the recall period was very short. But the only way to find out more about the validity of retrospective PROM collection is to do more of it, so hopefully we’ll see more studies asking this question.

Adaptation or recovery after health shocks? Evidence using subjective and objective health measures. Health Economics [PubMed] Published 26th February 2018

People’s expectations about their health can influence their behaviour and determine their future health, so it’s important that we understand people’s expectations and any ways in which they diverge from reality. This paper considers the effect of a health shock on people’s expectations about how long they will live. The authors focus on survival probability, measured objectively (i.e. what actually happens to these patients) and subjectively (i.e. what the patients expect), and the extent to which the latter corresponds to the former. The arguments presented are couched within the concept of hedonic adaptation. So the question is – if post-shock expectations return to pre-shock expectations after a period of time – whether this is because people are recovering from the disease or because they are moving their reference point. Data are drawn from the Health and Retirement Study. Subjective survival probability is scaled to whether individuals expect to survive for 2 years. Cancer, stroke, and myocardial infarction are the health shocks used. The analysis uses some lagged regression models, separate for each of the three diagnoses, with objective and subjective survival probability as the dependent variable. There’s a bit of a jumble of things going on in this paper, with discussions of adaptation, survival, self-assessed health, optimism, and health behaviours. So it’s a bit difficult to see the wood for the trees. But the authors find the effect they’re looking for. Objective survival probability is negatively affected by a health shock, as is subjective survival probability. But then subjective survival starts to return to pre-shock trends whereas objective survival does not. The authors use this finding to suggest that there is adaptation. I’m not sure about this interpretation. To me it seems as if subjective life expectancy is only weakly responsive to changes in objective life expectancy. The findings seem to have more to do with how people process information about their probability of survival than with how they adapt to a situation. So while this is an interesting study about how people process changes in survival probability, I’m not sure what it has to do with adaptation.

3L, 5L, what the L? A NICE conundrum. PharmacoEconomics [PubMed] Published 26th February 2018

In my last round-up, I said I was going to write a follow-up blog post to an editorial on the EQ-5D-5L. I didn’t get round to it, but that’s probably best as there has since been a flurry of other editorials and commentaries on the subject. Here’s one of them. This commentary considers the perspective of NICE in deciding whether to support the use of the EQ-5D-5L and its English value set. The authors point out the differences between the 3L and 5L, namely the descriptive systems and the value sets. Examples of the 5L descriptive system’s advantages are provided: a reduced ceiling effect, reduced clustering, better discriminative ability, and the benefits of doing away with the ‘confined to bed’ level of the mobility domain. Great! On to the value set. There are lots of differences here, with 3 main causes: the data, the preference elicitation methods, and the modelling methods. We can’t immediately determine whether these differences are improvements or not. The authors stress the point that any differences observed will be in large part due to quirks in the original 3L value set rather than in the 5L value set. Nevertheless, the commentary is broadly supportive of a cautionary approach to 5L adoption. I’m not. Time for that follow-up blog post.



On the commensurability of efficiency

In this week’s round-up, I highlighted a recent paper in the journal Cambridge Quarterly of Healthcare Ethics. There are some interesting ideas presented regarding the challenge of decision-making at the individual patient level, and in particular a supposed trade-off between achieving efficiency and satisfying health need.

The gist of the argument is that these two ‘values’ are incommensurable in the sense that the comparative value of two choices is ambiguous where the achievement of efficiency and need satisfaction needs to be traded. In the journal round-up, I highlighted 2 criticisms. First, I suggested that efficiency and health need satisfaction are commensurable. Second, I suggested that the paper did not adequately tackle the special nature of microlevel decision-making. The author – Anders Herlitz – was gracious enough to respond to my comments with several tweets.

Here, I’d like to put forth my reasoning on the subject (albeit with an ignorance of the background literature on incommensurability and other matters of ethics).

Consider a machine gun

A machine gun is far more efficient than a pistol, right? Well, maybe. A machine gun can shoot more bullets than a pistol over a sustained period. Likewise, a doctor who can treat 50 patients per day is more efficient than a doctor who can treat 20 patients per day.

However, the premise of this entire discussion, as established by Herlitz, is values. Herlitz introduces efficiency as a value and not as some dispassionate indicator of return on input. When we are considering values – as we necessarily are when we are discussing decision-making and more generally ‘what matters’ – we cannot take the ‘more bullets’ approach to assessing efficiency.

That’s because ‘more bullets’ is not what we mean when we talk about the value of efficiency. The production function is fundamental to our understanding of efficiency as a value. Once values are introduced, it is plain to see that in the context of war (where value is attached to a greater number of deaths) a machine gun may very well be considered more efficient. However, bearing a machine gun is far less efficient than bearing a pistol in a civilian context because we value a situation that results in fewer deaths.

In this analogy, bullets are health care and deaths are (somewhat confusingly, I admit) health improvement. Treating more people is not better because we want to provide more health care, but because we want to improve people’s health (along with some other basket of values).

Efficiency only has value with respect to the outcome in whose terms it is defined, and is therefore always commensurable with that outcome. That is, the production function is an inherent and necessary component of an efficiency to which we attach value.

I believe that Herlitz’s idea of incommensurability could be a useful one. Different outcomes may well be incommensurable in the way described in the paper. But efficiency has no place in this discussion. The incommensurability Herlitz describes in his paper seems to be a simple conflict between utilitarianism and prioritarianism, though I don’t have the wherewithal to pursue that argument so I’ll leave it there!

Microlevel efficiency trade-offs

Having said all that, I do think there could be a special decision-making challenge regarding efficiency at the microlevel. And that might partly explain Herlitz’s suggestion that efficiency is incommensurable with other outcomes.

There could be an incommensurability between values that can be measured in their achievement at the individual level (e.g. health improvement) and values that aren’t measured with individual-level outcomes (e.g. prioritisation of more severe patients). Those two outcomes are incommensurable in the way Herlitz described, but the simple fact that we tend to think about the former as an efficiency argument and the latter as an equity argument is irrelevant. We could think about both in efficiency terms (for example, treating n patients of severity x is more efficient than treating n-1 patients of severity x, or n patients of severity x-1), we just don’t. The difficulty is that this equity argument is meaningless at the individual level because it relies on information about outcomes outside the microlevel. The real challenge at the microlevel, therefore, is to acknowledge scope for efficiency in all outcomes of value. The incommensurability that matters is between microlevel and higher-level assessments of value.

As an aside, I was surprised that the Rule of Rescue did not get a mention in the paper. This is a perfect example of a situation in which arguments that tend to be made on efficiency grounds are thrown out and another value (the duty to save an immediately endangered life) takes over. One doesn’t need to think very hard about how Rule of Rescue decision-making could be framed as efficient.

In short, efficiency is never incommensurable because it is never an end in itself. If you’re concerned with being more efficient for the sake of being more efficient then you are probably not making very efficient decisions.