Causal Inference: IJE Special Issue

Papers from the December 2016 special issue of IJE are now all available online. Several are open access, and I attach these.

Philosophers who want to engage with real life science, on topics relating to causation, epidemiology, and medicine, will find these papers a great resource. So will epidemiologists and other scientists who want or need to reflect on causal inference. Most of the papers are not written by philosophers, and most do not start from standard philosophical starting points. Yet the topics are clearly philosophical. This collection would also form a great starting point for a doctoral research projects in various science-studies disciplines.

Papers 1 and 2 were first available in January. Two letters were written in response (being made available online around April) along with a response and I have included these in the list for completeness. The remaining papers were written during the course of 2016 and are now available. Many of the authors met at a Radcliffe Workshop in Harvard in December 2016. An account of that workshop may be forthcoming at some stage, but equally it may not, since not all of the participants felt that it was necessary to prolong the discussion or to share the outcomes of the workshop more widely. At some point I might simply write up my own account, by way of part-philosophical, part-sociological story.

  1. Causality and causal inference in epidemiology: the need for  a pluralistic approach‘ Jan P Vandenbroucke, Alex Broadbent and Neil Pearce. doi: 10.1093/ije/dyv341
  2. ‘The tale wagged by the DAG: broadening the scope of causal inference and explanation for epidemiology.’ Nancy Krieger and George Davey-Smith. doi: 10.1093/ije/dyw114
    1. Letter: Tyler J. VanderWeele, Miguel A. Hernán, Eric J. Tchetgen Tchetgen, and James M. Robins. Letter to the Editor. Re: Causality and causal inference in epidemiology: the need for a pluralistic approach.
    2. Letter: Arnaud Chiolero. Letter to the Editor. Counterfactual and interventionist approach to cure risk factor epidemiology.
    3. Letter: Broadbent, A., Pearce, N., and Vandenbroucke, J. Authors’ Reply to: VanderWeele et al., Chiolero, and Schooling et al.
  3. ‘Causal inference in epidemiology: potential outcomes, pluralism and peer review.’ Douglas L Weed. doi: 10.1093/ije/dyw229
  4. ‘On Causes, Causal Inference, and Potential Outcomes.’ Tyler VanderWeele. doi: 10.1093/ije/dyw230
  5. ‘Counterfactual causation and streetlamps: what is to be done?’ James M Robins and Michael B Weissman. doi: 10.1093/ije/dyw231
  6. ‘DAGs and the restricted potential outcomes approach are tools, not theories of causation.’ Tony Blakely, John Lynch and Rebecca Bentley. doi: 10.1093/ije/dyw228
  7. ‘The formal approach to quantitative causal inference in epidemiology: misguided or misrepresented?’ Rhian M Daniel, Bianca L De Stavola and Stijn Vansteelandt. doi: 10.1093/ije/dyw227
  8. Formalism or pluralism? A reply to commentaries on ‘Causality and causal inference in epidemiology.’ Alex Broadbent, Jan P Vandenbroucke and Neil Pearce. doi: 10.1093/ije/dyw298
  9. ‘FACEing reality: productive tensions between our epidemiological questions, methods and mission.’ Nancy Krieger and George Davey-Smith. doi: 10.1093/ije/dyw330

America Tour: Attribution, prediction, and the causal interpretation problem in epidemiology

Next week I’ll be visiting America to talk in Pittsburgh, Richmond, and twice at Tufts. I do not expect audience overlap so I’ll give the same talk in all venues, with adjustments for audience depending on whether it’s primarily philosophers or epidemiologists I’m talking to. The abstract is below. I haven’t got a written version of the paper that I can share yet but would of course welcome comments at this stage.

ABSTRACT

Attribution, prediction, and the causal interpretation problem in epidemiology

In contemporary epidemiology, there is a movement, part theoretical and part pedagogical, attempting to discipline and clarify causal thinking. I refer to this movement as the Potential Outcomes Aproach (POA). It draws inspiration from the work of Donald Ruben and, more recently, Judea Pearl, among others. It is most easily recognized by its use of Directed Acycylic Graphs (DAGs) to describe causal situations, but DAGs are not the conceptual basis of the POA in epidemiology. The conceptual basis (as I have argued elsewhere) is a commitment to the view that the hallmark of a meaningful causal claim is that they can be used to make predictions about hypothetical scenarios. Elsewhere I have argued that this commitment is problematic (notwithstanding the clear connections with counterfactual, contrastive and interventionist views in philosophy). In this paper I take a more constructive approach, seeking to address the problem that troubles advocates of the POA. This is the causal interpretation problem (CIP). We can calculate various quantities that are supposed to be measures of causal strength, but it is not always clear how to interpret these quantities. Measures of attributability are most troublesome here, and these are the measures on which POA advocates focus. What does it mean, they ask, to say that a certain fraction of population risk of mortality is attributable to obesity? The pre-POA textbook answer is that, if obesity were reduced, mortality would be correspondingly lower. But this is not obviously true, because there are methods for reducing obesity (smoking, cholera infection) which will not reduce mortality. In general, say the POA advocates, a measure of attributability tells us next to nothing about the likely effect of any proposed public health intervention, rendering these measures useless, and so, for epidemiological purposes, meaningless. In this paper I ask whether there is a way to address and resolve the causal interpretation problem without resorting to the extreme view that a meaningful causal claim must always support predictions in hypothetical scenarios. I also seek connections with the notorious debates about heritability.

Tobacco and epidemiology in Korea: old tricks, new answers?

Today I participated in a seminar hosted by the National Health Insurance Service (NHIS) of Korea, which is roughly the equivalent of the NHS in the UK, although the health systems differ. The seminar concerned a recent lawsuit in which tobacco companies were sued by the NHIS for the costs of treating lung cancer patients. The suit is part of a larger drive to get a grip on smoking in Korea, where over 40% of males smoke, and a packet of 20 cigarettes costs 4500 Korean Won (about USD 4.10 or UKP 2.80). The NHIS recently suffered a blow at the Supreme Court, where the ruling was somewhat luke-warm about a causal link between smoking and lung cancer in general, and moreover argued that such a link would anyway fail to prove anything about the two specific plaintiffs in the case at hand.

I was struck by the familiarity of some of the arguments that are apparently being used by the tobacco companies. For example, the Supreme Court has been convinced that diseases come in two kinds, specific and non-specific, and that since lung-cancer is a non-specific disease, it is wrong to seek to apply measures of attributability (excess/attributable fraction, population excess/attributable fraction) at all.

This is reminiscent of the use of non-specificity in the 1950s, when it was seen as a problem for the causal hypothesis that smoking causes lung cancer. It also gives rise to a strategy which is legally sound but dubious from a public health perspective, namely, first going for lung cancer, and leaving other health-risks of smoking for later. This is legally sound because lung cancer exhibits the highest relative risk of the smoking-related diseases, and perhaps it is good PR too because cancer of any kind catches the imagination. But the health burden of lung cancer is low, even in a population where smoking is relatively prevalent, since lung cancer is a rare disease even among smokers.

The health burden of heart disease, at the other end of the spectrum, is very large, and even though smoking less than doubles this risk (RR about 1.7), the base rate of heart disease is so high that this amounts to a very significant public health problem. I do not know what the right response to this complex of problems is: clearly, high-profile court cases are have an impact that extends far beyond their outcome, and also the reason that people stop smoking, or accept legislation, need not be an accurate reflection of the true risks in order for those risks to be mitigated. (If you stop smoking to avoid lung cancer, you also avoid heart disease, which is a much better reason to stop smoking from the perspective of a rational individual motivated to avoid fatal disease.) Nonetheless I am struck by the way that legal and health policy objectives interact here.

I was also interested to hear that the case of McTear was a significant blow to the Korean case because of its findings about causality, which indeed are exactly those of the Korean case. That case is not well regarded in the UK, and not authoritative (being first instance), so it is interesting – and unfortunate – that it has had an effect here.

The event was an extremely good-spirited affair, and the other speakers had some interesting things to say. My book, in Korean, received a significant plug, not least, I suspect, because the audience not understanding much of my talk, were repeatedly referred to it for more detail. The most shocking thing about the event was to hear the same obfuscatory strategies that are now history in Europe and America being used, to good effect, by the very same companies in this part of the world. It is one thing to defend a case on grounds that one believes, but there is not anyone who still reasonably believes that smoking does not cause lung cancer, which seems to be the initial burden that plaintiffs in this sort of case need to prove. It is a bit like being asked to begin your case against a scaffolder who dropped a metal bar on your head with a proof of the law of gravity, and then being asked to prove that the general evidence concerning gravity proves that gravity was the cause in this particular case, given that not all downward motions are caused by gravity. – Not exactly like that, of course, but not exactly unlike, either.

On the positive side, I am hoping that a clear explanation of the reasoning behind the PC Inequality that I favour might help with the next stage of the case, although I am unclear what that stage might be.

A Tale of Two Papers

I’m on my way back from the World Epi Congress in Anchorage, where causation and causal inference have been central topics of discussion. I wrote previously about a paper (Hernan and Taubman 2008) suggesting that obesity is not a cause of mortality. There is another, more recent paper published in July of this year, suggesting, more or less, that race is not a cause of health outcomes – or at least that it’s not a cause that can feature in causal models (Vanderweele and Robinson 2014). I can’t do justice to the paper here, of course, but I think this is a fair, if crude, summary of the strategy.

This paper is an interesting comparator for the 2008 obesity paper (Hernan and Taubman 2008). It shares the idea that there is a close link between (a) what can be humanly intervened on, (b) what counterfactuals we can entertain, and (c) what causes we can meaningfully talk about. This is a radical view about causation, much stronger than any position held by any contemporary philosopher of whom I’m aware. Philosophers who do think that agency or intervention are central to the concept of causation treat the interventions as in-principle ones, not things humans could actually do.

Yet feasibility of manipulating a variable really does seem to be a driver in this literature. In the paper on race, the authors consider what variables form the subject of humanly possible interventions, and suggest that rather than ask about the effect of race, we should ask what effect is left over after these factors are modelled and controlled for, under the umbrella of socioeconomic status. That sounds to me a bit like saying that we should identify the effects of being female on job candidates’ success by seeing what’s left after controlling for skirt wearing, longer average hair length, shorter stature, higher pitched voice, female names, etc. In other words, it’s very strange indeed. Perhaps it could be useful in some circumstances, but it doesn’t really get us any further with the question of interest – how to quantify the health effects of race, sex, and so forth.

Clearly, there are many conceptual difficulties with this line of reasoning. A good commentary was published with the paper (Glymour and Glymour 2014) which really dismantles the logic of the paper. But I think there are a number of deeper and more pervasive misunderstandings to be cleared up, misunderstandings which help explain why papers like this are being written at all. One is confusion between causation and causal inference; another is confusion between causal inference and particular methods of causal inference; and a third is a mix-up between fitting your methodological tool to your problem, and your problem to your tool.

The last point is particularly striking. What’s so interesting about these two papers (2008 & 2014) is that they seem to be trying to fit research problems to methods, not trying to develop methods to solve problems – even though this is ostensibly what they (at least VW&R 20114) are trying to do. To me, this is strongly reminiscent of Thomas Kuhn’s picture of science, according to which an “exemplary” bit of science occurs, and initiates a “paradigm”, which is a shared set of tools for solving “puzzles”. Kuhn was primarily influenced by physics, but this way of seeing things seems quite apt to explain what is otherwise, from the outside, really quite a remarkable, even bizarre about-turn. Age, sex, race – these are staple objects of epidemiological study as determinants of health; and they don’t fit easily into the potential outcomes paradigm. It’s fascinating to watch the subsequent negotiation. But I’m quite glad that it doesn’t look like epidemiologists are going to stop talking about these things any time soon.

References

Glymour C and Glymour MR. 2014. ‘Race and Sex Are Causes.’ Epidemiology 25 (4): 488-490.

Hernan M and Taubman S. 2008. ‘Does obesity shorten life? The importance of well-defined interventions to answer causal questions.’ International Journal of Obesity 32: S8–S14.

VanderWeele TJ and Robinson WR. 2014. ‘On the Causal Interpretation of Race in Regressions Adjusting for Confounding and Mediating Variables.’ Epidemiology 25(4): 473-484.

Snakes, statistics, and goals for the goal-setters

Cesar Victora gave a very interesting talk earlier today concerning the International Epidemiology Association’s position paper on the UN’s Sustainable Development Goals, which are currently being drafted (to replace the Millennium Development Goals post-2015). Victora is President of the IEA, for a few more hours at least (the new President takes office this evening). Many of his points were reiterated by the next speaker, Theodor Abelin, and in questions from the floor. There were no audible voices of dissent. (The talk reflects a fuller position paper, available here.)

The point that stayed with me most from Victora’s rich talk was the importance of relating goals to appropriate measurement techniques. My own interest in epidemiology has tended to focus on efforts to identify causes (“analytic” epidemiology), since causation is a natural magnet for philosophical interest. But measurement is also a focus of philosophical interest, and Victora nicely pointed out that “descriptive” epidemiology – the business of measuring things like maternal mortality rate, for example – is extremely important if these Sustainable Development Goals are to be effective. A country cannot be held to a goal that cannot be measured, and it cannot be fairly be held to a goal when progress towards that goal is estimated rather than measured.

For example, I was not surprised to learn that in many countries where maternal mortality is high, data on maternal mortality rates (MMRs) are scarce. What did surprise me was hearing about the calculations that some august international organisations perform in the absence of data. A calculation is performed involving GDP per capita, general fertility rate and skilled birth attendance. MMR is estimated as a function of these and perhaps some other similar variables. This means that if the country goes through a recession, the estimated MMR will automatically go up. – Perhaps is really will go up, but it seems strange to think of that calculation as a measurement, at least in the absence of extremely good evidence for the reliability of the estimating equation – evidence which, of course, we don’t have.

MMR is measurable, of course. The problem with MMR is simply a lack of data, and this problem afflicts a large class of conditions. As Victora put it in relation to snakebite: “Where we have snakes, we don’t have statistics, and where we have statistics, we don’t have snakes.”

However, Victora’s most penetrating critique of the SDGs concerned the setting of goals in the absence of clear ideas about how progress towards the goals will be measured. The health-related goal is as follows:

Goal 3. Ensure healthy lives and promote well-being for all at all ages” (from the Outcome Document)

This overarching goal is broken down into 13 subgoals, some of which are very loosely specified. For instance, how are we to tell whether a country has managed to “strengthen prevention and treatment of substance abuse, including narcotic drug abuse and harmful use of alcohol”? Ironically, those goals that are most clearly specified are wildly unattainable, such as halving global deaths and injuries from road traffic accidents by 2020. Those that are not well specified present measurement challenges for epidemiologists.

This made me wonder whether a body like the IEA could itself set some “goals for the goal-setters” – that is, criteria which any health-related goal must meet if, in the professional opinion of the IEA, they are to be useful. The simplest such criterion would be that outcomes must be specified in terms of a recognised epidemiological measure (mortality, for instance). Another might be to accompany each goal with information (perhaps in a corresponding entry in an appendix) concerning the trend over the past similar period: so if the goal is the halve road traffic deaths in 15 years, or 25, information on the growth of road traffic deaths over the past 15 or 25 years might be included. Goals of this kind will always be political, but there might be agreement on a set of simple rules for setting such goals, and if such rules existed, this might pull epidemiologists closer in to the goal-setting process – a kind of politicking which, as one of the questioners pointed out, is not part of standard epidemiological training.

 

“Risk Relativism” paper accepted – thanks to those who commented

Just a note to thank those who offered comments on the revisions of “Risk relativism and physical law”. This has now been accepted by Journal of Epidemiology and Community Health, where it will feature as part of a “Debate” with invited comments from couple of epidemiologists. Not entirely sure when, since they will presumably have to write their comments now. Anyway I really appreciate the feedback I got on this one – definitely improved the final result. Thanks.

Comments sought: Risk relativism and physical law

The attached is a revise and resubmit, and will form part of a Debate in Journal of Epidemiology and Community Health. I have until 25 July to submit. Comments are very welcome. Text is 2100 words. Feel free to comment/track changes in the doc if so inclined.

2014-07-19 Risk relativism and physical law – version for comment