Commentary _social_epidemiology__questionable answers and answerable questions


Published on

commentary free acces publishedin the journal of epidemiology

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Commentary _social_epidemiology__questionable answers and answerable questions

  1. 1. THE CHANGING FACE OF EPIDEMIOLOGY Editors’ note: This series addresses topics of interest to epidemiologists across a range of specialties. Commentaries start as invited talks at symposia organized by the E ­ ditors. This paper was presented at the Third North American Congress of Epidemiology held in Montreal in June 2011. Social Epidemiology Questionable Answers and Answerable Questions Sam Harpera and Erin C. Strumpf  a,b S ocial epidemiology encompasses the study of relationships between health and a broad range of social factors such as race, social class, gender, social policies, and so on. One could broadly partition the work of social epidemiology into surveillance (ie, descriptive relationships between social factors and health, tracking of health inequalities over time) and etiology (ie, causal effects of social exposures on health).1 Many social epidemiolo- gists believe these twin pursuits should ultimately serve to structure interventions aimed at reducing health-damaging social exposures or increasing exposure to social factors that enhance health. SOCIAL EPIDEMIOLOGY RISING Two decades ago one could have argued that social epidemiology was a fringe sub- discipline; now it is as prominent as any other specialization within epidemiology. Many schools of public health now offer concentrations in social epidemiology, most countries have public health goals aimed specifically at reducing social inequalities in health, and no fewer than three general and two methodological textbooks on social epidemiology are now in circulation. More importantly, social epidemiology now occupies an important place in the realm of public health policy. The World Health Organization’s landmark Commission on the Social Determinants of Health Report,2 published in 2008, attempted to bring world- wide attention to social determinants of health and to affect policy decisions in ways that would reduce social inequalities in health. The Commission’s report would not have been possible without the pathbreaking work on health inequalities that developed in the United Kingdom after the Black Report3 in 1980, which subsequently led to two full-scale reports on health inequalities in the United Kingdom.4,5 Each of these reports has used evidence assembled by social epidemiologists to advocate for policies aimed at reducing health inequalities, and yet, after the most recent report5—which found little evidence that health inequalities in England have decreased— there seems to be considerable consternation about the potential for policies to reduce social inequalities in health. Johan Mackenbach, who has produced as much evidence on health inequalities across the European region as anyone, seemed especially pessimistic: “The main conclusion therefore is that reducing health inequalities is currently beyond our means. That is the sad but inevitable conclusion from the story of the English strat- egy to reduce health inequalities.”6 Echoing these same concerns, Bambra and colleaguesFrom the aDepartment of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, Canada; and bDepartment of Economics, McGill University, Montreal, Quebec, Canada.Supported by a Chercheur Boursier Junior 1 from the Fonds de la Recherche du Québec–Santé (to S.H.) and from the Fonds de la Recherche du Québec–Santé and the Ministère de la Santé et des Services sociaux du Québec (to E.S.).Editors’ note: Related articles appear on pages 785, 787, and 790.Correspondence: Sam Harper, Department of Epidemiology, Biostatistics & Occupational Health, McGill University, 1020 Pine Avenue West, Purvis Hall, Room 34, Montreal, Quebec, Canada H3A 1A2. E-mail: © 2012 by Lippincott Williams & WilkinsISSN:1044-3983/12/2306-0795DOI:10.1097/EDE.0b013e31826d078dEpidemiology  •  Volume 23, Number 6, November 2012  |  795
  2. 2. Harper and Strumpf Epidemiology  •  Volume 23, Number 6, November 2012were dismayed by the similarity of the conclusions of each of health is not strong enough to overwhelm the effect of socialthe English reports more than 30-odd years and questioned causation.16,17 These studies typically ask the descriptive ques-the purpose of accumulating evidence from social epidemi- tion of whether those in higher status occupations have betterology, given the fact that the policy conclusions never seem health (controlling for a range of covariates), rather than theto change.7 These self-critical remarks among committed question of whether intervening to change occupational statusresearchers are concordant with evidence from policymak- would lead to improved health outcomes. Case and Paxson,18ers who charge social epidemiologists (and the broader social however, use the Whitehall data to reframe the question anddeterminants of health movement) with providing “policy-free ask whether changes in occupational grade affect changes inevidence” and the “right answers to the wrong questions.”8 health and, in a parallel fashion, whether changes in health affect changes in occupational grade. Using a fixed-effects CRITIQUES OF THE SOCIAL APPROACH design that controls for unmeasured time-invariant individual The gap between the questions policymakers want characteristics, they find no evidence that civil service gradeanswered and the kinds of questions social epidemiologists affects health, but some evidence that health affects futurehave been answering is evident in a series of commentar- employment. They also find that future civil service gradeies on the most recent report9 by Michael Marmot on health predicts current health, suggesting health-related selection orinequalities in the United Kingdom. While social epide- unmeasured confounding. Although this identification strat-miologists largely praised the report (some even chided the egy trades off a lot of precision for a reduction in bias,19 it isUK committee for not being radical enough in their policy probably closer to the kind of question that we would ask ifrecommendations),10 others questioned whether the report we could design a randomized trial to evaluate the effects ofhad provided sufficient evidence to back its strong policy occupational grade on health.recommendations.11,12 Critics accused social epidemiologists A second example comes from a randomized trialof being “casual about causality” and argued that there was, called the Moving to Opportunity study. The backdrop herein fact, little evidence to suggest causal effects of income or is an abundance of observational research showing thateducation on health (although there is better evidence for the neighborhood characteristics such as poverty concentrationassociation between early-life conditions and health in later are associated with a wide range of health and sociallife). The discrepant readings of this evidence may be a conse- outcomes.20 In the randomized study, nearly 5000 familiesquence of social epidemiologists generally answering descrip- living in high-poverty urban neighborhoods in five U.S. citiestive questions (whether socially disadvantaged individuals were randomly assigned one of two treatments, one of whichhave poorer health), but interpreting their findings causally required relocating to a more prosperous neighborhood.(whether socially disadvantaged individuals would have bet- The intervention was successful in reducing exposure toter health if they were to become advantaged, or vice versa). concentrated poverty, but intention-to-treat (ITT) estimatesSuch interpretation implies both an intervention and a causal on social and health outcomes (in the presence of substantialeffect of that intervention. Social epidemiologists often pres- non-compliance) were largely null.21 Sociologists Clampet-ent their results in counterfactual terms that imply causation: Lundquist and Massey used the same data but analyzed itif clerical workers had the same mortality as administrators,13 as an observational study, arguing that the ITT estimate canif blacks had the same mortality as whites,14 if everyone had “measure the effects of the policy initiative, but is not wellthe same mortality rates as those with a university educa- suited to capturing neighborhood effects.”22 This quote seemstion,15 the reduction in disease burden would be substantial. to reflect some of the tension mentioned above between theFurthermore, such counterfactuals often do not correspond to kind of evidence wanted by policymakers (causal effectany known or feasible interventions. We suggest that the epi- of the policy) and the kind of evidence being delivered bydemiologic question should be reframed and that changing the social epidemiology. The investigators found evidence thatquestion can make an important difference both to the answer increased exposure to low poverty areas is beneficial—one gets about the impact of social conditions on health and although by breaking the randomization they reintroducedto one’s ability to provide useful information to policymakers. precisely the same kinds of selection effects that led to theSome examples follow. concerns cited above about being “casual about causality.”23 This does not mean that randomized experiments are the only REFRAMING THE QUESTION solution. Sampson and colleagues,24 for example, were able to The UK Whitehall studies have provided numerous “recover” the randomization study estimates on verbal IQ byanalyses demonstrating positive associations between civil explicitly mimicking the design of a trial in observational dataservice occupational grade and health, but concerns are often from Chicago neighborhoods.raised about the potential for social gradients to arise because A third example of reframing concerns the effect ofof unmeasured confounding or health-related selection. education on mortality. Countless social epidemiologic stud-Whitehall investigators have largely attributed their results to ies have followed persons with high versus low educationsocial causation, arguing that any social selection based on and found substantially increased mortality among the less796  | © 2012 Lippincott Williams & Wilkins
  3. 3. Epidemiology  •  Volume 23, Number 6, November 2012 Social Epidemiologyeducated—an association that is in most cases robust to con- is to inform policy then our questions should be motivated bytrols for additional socioeconomic, behavioral, and biologic identifying causal relationships that can be of the most use tofactors.25 These observational studies ask whether more edu- policymakers. It goes without saying that in this short essaycated individuals have lower mortality, but generally do not we have painted social epidemiology with perhaps too broadask what policymakers want to know: what would the health a brush—many social epidemiologists undoubtedly produceeffects be of intervening to increase a person’s amount of high-quality evidence on the impact of social exposures oneducation, or their achievement of a certain educational qual- health. Yet, our concern here is that much of social epidemiol-ification? Most observational study designs cannot answer ogy is providing questionable answers to important questions,this question, because it is very difficult to control for the rather than focusing on answerable questions. Batty35 recentlymultitude of factors that jointly determine both education and lamented the lack of interest among the research communityhealth.26 However, more recent studies have used changes in in generating experimental or quasi-experimental evidence onschooling policies that are more plausibly independent of social determinants of health. To focus on answerable ques-unobserved individual characteristics and their health out- tions, social epidemiologists may need to sharpen their focuscomes (eg, minimum age required for leaving school) to on specific questions that are more narrow in scope, but forminimize unmeasured confounding. These studies bring us which experimental or quasi-experimental data exist.closer to answering the question of whether intervening to The major strength of experimental and quasi-experi-increase educational achievement is likely to have effects on mental designs lies in their robustness to unobserved con-mortality, and the results are decidedly more mixed.25 Several founding. Such designs also have conceptual strengths thatstudies provide evidence that exogenous changes in school- come from asking specific causal questions. Trials have theiring resulting from policies reduce mortality,27–29 but several own limitations and will never answer some social epidemi-other studies do not, even for different analyses of the same ologic hypotheses, but attempting to mimic the randomizedcountry.29–32 The contrast between the consistency of studies trial one would hypothetically conduct to answer an importantanswering descriptive questions and the mixed evidence for question would seem to be a good place to start.36,37 Finally,(more approximately) causal questions suggests that this dis- we would also argue that greater transparency in methodstinction matters. and reporting would increase the value of replication stud- A final example comes from our own work, but again ies38 that can directly address the question of how sensitivereflects the importance of focusing on a well-specified causal research findings in social epidemiology may be to measure-question. A recent article found that U.S. states with medi- ment error, model selection, unmeasured confounding, andcal marijuana laws had higher adolescent marijuana use.33 other potential biases.39,40 Greater attention to both researchThe authors focused on two comparisons: (1) the difference design and rigorous analysis can help social epidemiologistsin marijuana use between states passing laws and states with answer important questions rather than continuing to generateno laws and no history of such laws and (2) the difference in questionable answers.marijuana use between states that had already passed laws andstates that had never passed a law. Neither of these contrasts ABOUT THE AUTHORSrepresent the causal question facing policymakers in a state Sam Harper and Erin Strumpf are Assistant Professors incurrently without a medical marijuana law; namely, what is the Department of Epidemiology, Biostatistics & Occupationallikely to happen to marijuana use among adolescents if we Health at McGill University, where Erin Strumpf also holdslegalize medical marijuana? Yet the data do permit asking an appointment in the Department of Economics.precisely this question. Using difference-in-differences esti- Sam Harper works on measuring social inequalities inmation, we compared the change in marijuana use among health and how they change over time. Erin Strumpfs researchadolescents after a medical marijuana law is passed to cor- in health economics focuses on measuring the causal impactsresponding changes over the same period in states that do not of health and health care policies on spending and healthpass a law and found little evidence of any effect.34 Although outcomes overall, and in inequalities across groups.this strategy obviously does not approach randomization toa medical marijuana law, it provides better control for con- REFERENCESfounding by unmeasured state characteristics and overall 1. Kaufman JS. Social epidemiology. In: Rothman KJ, Greenland S, Lash TL, eds. Modern Epidemiology. 3rd ed. Philadelphia: Wolters Kluwersecular trends. More importantly, the analytic strategy follows Health/Lippincott Williams & Wilkins; 2008:533–548.directly from asking a specific question about social policy. 2. WHO Commission on Social Determinants of Health. Closing the gap in a generation: health equity through action on the social determinants of health. Final Report of the Commission on the Social Determinants of CONCLUSIONS Health. Geneva, Switzerland: World Health Organization, Commission The above examples illustrate that the manner in which on Social Determinants of Health; epidemiology asks questions has important implica- 3. Townsend P, Davidson N. Inequalities in Health: The Black Report. Har- mondsworth, UK: Penguin Books; 1982.tions for the evidence it generates. Different questions gener- 4. Acheson D. Independent Inquiry Into Inequalities in Health Report. Lon-ate different answers, even with the same data, and if the goal don: Stationery Office; 1998.© 2012 Lippincott Williams & Wilkins  |  797
  4. 4. Harper and Strumpf Epidemiology  •  Volume 23, Number 6, November 2012 5. Marmot M. Fair society, healthy lives: The Marmot Review. Strategic 2 3. Ludwig J, Liebman JB, Kling JR, et al. What can we learn about neigh- review of health inequalities in England post-2010. 2010; Available at: borhood effects from the moving to opportunity experiment? Am J Sociol. 2008;114:144–188. HealthyLives.pdf. Accessed May 5, 2011. 2 4. Sampson RJ, Sharkey P, Raudenbush SW. Durable effects of concentrated 6. Mackenbach JP. Has the English strategy to reduce health inequalities disadvantage on verbal ability among African-American children. Proc failed? Soc Sci Med. 2010;71:1249–1253; discussion 1254–1248. Natl Acad Sci USA. 2008;105:845–852. 7. Bambra C, Smith KE, Garthwaite K, Joyce KE, Hunter DJ. A labour of 2 5. Kawachi I, Adler NE, Dow WH. Money, schooling, and health: Mecha- Sisyphus? Public policy and health inequalities research from the Black nisms and causal evidence. Ann N Y Acad Sci. 2010;1186:56–68. and Acheson Reports to the Marmot Review. J Epidemiol Community 2 6. Cutler DM, Lleras-Muney A. Education and health: evaluating theories Health. 2011;65:399–406. and evidence. In: Schoeni RF, House JS, Kaplan GA, Pollack H, eds. 8. Petticrew M, Whitehead M, Macintyre SJ, Graham H, Egan M. Evidence Making Americans Healthier: Social and Economic Policy as Health for public health policy on inequalities: 1: the reality according to policy- Policy. New York: Russell Sage Foundation; 2008:29–60. makers. J Epidemiol Community Health. 2004;58:811–816. 2 7. Lleras-Muney A. The relationship between education and adult mortality 9. Whitehead M, Popay J. Swimming upstream? Taking action on the social in the United States. Rev Econ Stat. 2005;72:189–221. determinants of health inequalities. Soc Sci Med. 2010;71:1234–1236; 2 8. van Kippersluis H, O’Donnell O, van Doorslaer E. Long-run returns to discussion 1254–1238. education does schooling lead to an extended old age? J Hum Resources.1 0. Pickett KE, Dorling D. Against the organization of misery? The Marmot 2011;46:695–721. Review of health inequalities. Soc Sci Med . 2010;71:1231–1233; discus- 2 9. Cutler DM, Lleras-Muney A. Education and health: insights from inter- sion 1254–1238. national comparisons. National Bureau of Economic Research Working1 1. Chandra A, Vogl TS. Rising up with shoe leather? A comment on Fair So- Paper Series. 2012; No. 17738. ciety, Healthy Lives (the Marmot Review). Soc Sci Med. 2010;71:1227– 3 0. Clark D, Royer H. The effect of education on adult health and mortality: 1230; discussion 1254. evidence from Britain. National Bureau of Economic Research Working1 2. Canning D, Bowser D. Investing in health to improve the wellbeing of the Paper Series. 2010;No. 16013. disadvantaged: reversing the argument of Fair Society, Healthy Lives (the 3 1. Albouy V Lequien L. Does compulsory education lower mortality? J , Marmot Review). Soc Sci Med. 2010;71:1223–1226: discussion 1254. Health Econ. 2009;28:155–168.1 3. Marmot MG, Smith GD, Stansfeld S, et al. Health inequalities among Brit- 3 2. Mazumder B. Does education improve health? A reexamination of the ish civil servants: the Whitehall II study. Lancet. 1991;337:1387–1393. evidence from compulsory schooling laws. Federal Reserve Bank of Chi-1 4. Satcher D, Fryer GE, McCann J, Troutman A, Woolf SH, Rust G. What if cago Economic Perspectives. 2008;Q2:2–16. we were equal? A comparison of the black-white mortality gap in 1960 3 3. Wall MM, Poh E, Cerda M, Keyes KM, Galea S, Hasin DS. Adolescent and 2000. Health Aff. 2005;24:459–464. marijuana use from 2002 to 2008: higher in states with medical marijuana1 5. Woolf SH, Johnson RE, Phillips RL Jr, Philipsen M. Giving everyone laws, cause still unclear. Ann Epidemiol. 2011;21:714–716. the health of the educated: an examination of whether social change 3 4. Harper S, Strumpf EC, Kaufman JS. Do medical marijuana laws in- would save more lives than medical advances. Am J Public Health. crease marijuana use? Replication study and extension. Ann Epidemiol. 2007;97:679–683. 2012;22:207–212.1 6. Chandola T, Bartley M, Sacker A, Jenkinson C, Marmot M. Health selec- 3 5. Batty GD. We must move on: taking stock (yet again) of the evidence for tion in the Whitehall II study, UK. Soc Sci Med. 2003;56:2059–2072. socio-economic differentials in health. J Epidemiol Community Health.1 7. Elovainio M, Ferrie JE, Singh-Manoux A, et al. Socioeconomic differ- 2011;65:947–948. ences in cardiometabolic factors: social causation or health-related selec- 3 6. Hernán MA, Alonso A, Logan R, et al. Observational studies analyzed tion? Evidence from the Whitehall II Cohort Study, 1991–2004. Am J like randomized experiments: an application to postmenopausal hor- Epidemiol. 2011;174:779–789. mone therapy and coronary heart disease. Epidemiology. 2008;19:1 8. Case A, Paxson C. The long reach of childhood health and circumstance: evi- 766–779. dence from the Whitehall II Study. Econ J (London). 2011;121:F183–F204. 3 7. Rubin DB. The design versus the analysis of observational studies for1 9. Kaufman JS. Commentary: why are we biased against bias? Int J Epide- causal effects: parallels with the design of randomized trials. Stat Med. miol. 2008;37:624–626. 2007;26:20–36.2 0. Diez Roux AV, Mair C. Neighborhoods and health. Ann N Y Acad Sci. 3 8. Hernán MA, Wilcox AJ. Epidemiology, data sharing, and the challenge of 2010;1186:125–145. scientific replication. Epidemiology. 2009;20:167–168.2 1. Kling JR, Liebman JB, Katz LF. Experimental analysis of neighborhood 3 9. Lash TL, Fox MP, Fink AK. Applying Quantitative Bias Analysis to Epi- effects. Econometrica. 2007;75:83–119. demiologic Data. Dordrecht, NY: Springer; 2009.2 2. Clampet-Lundquist S, Massey DS. Neighborhood effects on eonomic 4 0. Vanderweele TJ, Arah OA. Bias formulas for sensitivity analysis of un- self-sufficiency: a reconsideration of the moving to opportunity experi- measured confounding for general outcomes, treatments, and confounders. ment. Am J Sociol. 2008;114:107–143. Epidemiology. 2011;22:42–52.798  | © 2012 Lippincott Williams & Wilkins