Current Epidemiology Reports

, Volume 2, Issue 1, pp 61–70 | Cite as

How Consequential Is Social Epidemiology? A Review of Recent Evidence

  • Arijit Nandi
  • Sam Harper
Social Epidemiology (JM Oakes, Section Editor)
Part of the following topical collections:
  1. Topical Collection on Social Epidemiology


A number of authors have suggested that epidemiology generally, and social epidemiology in particular, should be producing more consequential research to inform specific interventions for improving population health. However, the extent to which current research in social epidemiology prioritizes consequentialist questions is unknown. To provide some quantitative evidence on this question, we collected all abstracts accepted for either an oral or poster presentation at the annual meetings of the Society for Epidemiologic Research between 2009 and 2013. We identified all abstracts relating to the study of social determinants of health and classified each abstract as consequentialist if it evaluated the effect of a specific intervention. Among 619 abstracts examining social determinants of health, we classified 41 studies (6.6 % of total) as consequentialist, with minimal year-to-year variation. Little of social epidemiology appears to be focused on consequential research. Changes in pedagogy and greater involvement with the policy community may help.


Social epidemiology Consequentialism Neighborhood effects Intervention studies Causal inference Public policy 


For more than three decades, there has been an ongoing discourse about the fundamental role of epidemiologic science and the extent to which our research activities should directly inform efforts to prevent disease and improve population health [1]. While acknowledging the importance of describing the distributions of disease in populations and clarifying disease etiologies and mechanisms, a common thread in this discussion is that academic epidemiology spends most of its time on disease etiology and “far less of its time and imagination asking how we might improve population health” [2•]. Prominent epidemiologists, including William Foege [1], Willard Cates [3], and, most recently, Sandro Galea, have called for a more “consequentialist” epidemiology that prioritizes the “assessment of the potential contribution to population health of particular interventions implemented” [2•]. In the domain of social epidemiology, Patricia O’Campo recently questioned whether we are producing actionable evidence to address the social determinants of health and health inequality [4].

Consequentialism and Social Epidemiology

Social epidemiology, which is chiefly concerned with examining the role that social factors play in generating and sustaining health [5], has clear relevance to the discussion of consequentialism in epidemiologic research. Because the vision of consequential epidemiology described above is squarely focused on taking action to improve health, and such actions not only involve social planning but also have broad social consequences, social epidemiologists should be well-positioned to inform this kind of work. Social epidemiology often comes in two varieties—descriptive and etiologic. Whereas the former tracks the social patterning of health outcomes, the latter focuses on understanding the causal effects of social exposures such as education, for example, on health outcomes. Generating evidence on the causal effects of modifiable social exposures naturally leads to questions about how to intervene to alter exposure distributions.

However, even if its efforts to uncover causal effects are successful, it is not clear that choosing a course of action for intervention based on rigorous evidence is well informed by current practice in social epidemiology. Take the example of the multitude of studies linking maternal education and infant mortality within and between countries [6, 7, 8, 9, 10]. Prominent social epidemiologists suggest that “the knowledge and the means are at hand” [11] to ameliorate such inequalities, but what is to be done? What course of action should “consequentialist” epidemiologists advocate for in order to address this persistent inequality? Should we take a “social determinants of health” approach and intervene directly on maternal education? If so, how? By making secondary education free? Making it compulsory? Perhaps we should focus on improving school quality? Or, absent interventions on maternal education, should we try to influence the mechanisms linking it to infant mortality, such as access to maternal care, targeted immunization coverage among kids of less-educated mothers, increasing control over household resources among less-educated mothers, or increasing access to family planning? It could be a combination of several interventions and, of course, it is important to know which are most cost-effective.

But it is unclear whether social epidemiology as currently practiced is focusing much attention on these questions of action for specific interventions. A number of recent reflections on the current state of social epidemiology bemoan the discipline’s lack of engagement with interventions to change health and reduce health inequalities [12•, 13, 14, 15]. For example, O’Campo et al. [15] recently argued that social epidemiologists should be pursuing evaluations of macro-level policies and programs in order to provide evidence on solutions to the complex problems that inhere in the production and maintenance of health inequalities. Moreover, though there is enormous heterogeneity across outcomes and environments, many social epidemiologists have argued that social inequalities in health are increasing, and that our efforts to reduce them, even as the specific target of policies, have not been successful [16]. Others argue that social epidemiology has failed to properly engage with policymakers, who suggest that the discipline is chiefly focused on finding associations rather than providing evidence on questions that are crucial for designing policy interventions [17].

There thus seems to be some consensus that social epidemiology “should” be producing evidence that would fall under the rubric of “consequentialist.” Given the ongoing demands that social epidemiology produce research that is relevant for policy, and mostly qualitative arguments that the field has not made good on this agenda, we undertook a structured review of recent evidence presented at a high-profile conference to assess the extent to which social epidemiology is currently producing evidence that would contribute to a more “consequentialist” epidemiology.


Search Strategy

We reviewed abstracts accepted to annual meetings of the Society for Epidemiologic Research (SER) between 2009 and 2013 and evaluated trends in consequentialist research in the study of social determinants of health. The SER, founded in 1968, is one of the preeminent meetings of epidemiologic researchers in North America, and a review of accepted SER abstracts has been used previously to comment on the general direction of the field [18]. SER receives abstract submissions from members and non-members to be considered for presentation at the annual meeting. Submitted abstracts are reviewed by members of the Society, based on their areas of expertise. Each abstract is reviewed by a minimum of three individuals and is scored on a scale of 1 to 10. Reviewer scores are averaged to determine the overall abstract score. Higher-ranking abstracts are then made available to organizers to arrange the Contributed Concurrent and Poster Sessions held at the annual meeting. Accepted abstracts are published in a special supplement to the American Journal of Epidemiology. We collected all abstracts accepted for either an oral or poster presentation at SER between 2009 and 2013.

We identified all abstracts relating to the study of social determinants of health. Posters sessions are organized by theme. For example, in 2013, the themes were aging, behavioral, cancer, cardiovascular, diabetes, environmental, genetics, global health, health services, history, HIV, infection, injury, latebreakers, men’s health, methods, neuro epi, nutrition, obesity, occupational, pediatrics, perinatal/reproductive, psychiatric, renal, respiratory, screening, social, substance use, and women’s health. Concurrent Contributed Sessions in 2013 are collections of abstracts that are selected by researchers appointed by SER for oral presentations concerning a specific theme. From poster and spotlight sessions, we automatically selected all abstracts presented in thematic areas related to the study of social determinants and then reviewed all other topic areas for additional studies. We adopted a sensitive definition of social determinants, erring on the side of including studies for review.

We next reviewed the text of all abstracts related to the study of social determinants of health, among all abstracts accepted for presentation. From each abstract, we collected the following information: author affiliation; title; whether the abstract related to the study of social determinants of health; the setting, including the country or countries where the work was conducted; the main exposure(s); and the outcome(s). We then independently reviewed and judged whether each abstract was “consequentialist” or not. We defined a study as consequentialist if it evaluated (or simulated) the effect of a specific intervention. After we completed our assessments, we examined the agreement, measured by the Kappa statistic. We reconciled any disagreements in our ratings and identified a final set of abstracts deemed consequentialist.


Between 2009 and 2013, a total of 619 abstracts examining social determinants of health were accepted for a poster or spotlight presentation. The overall agreement between our evaluations was excellent (Kappa = 0.88). After reconciling disagreements in our assessments, 41 studies (6.6 % of total) were deemed consequentialist. The proportion of abstracts that we classified as consequentialist was 5.8 % in 2009, 4.3 % in 2010, 5.2 % in 2011, 6.3 % in 2012, and 12.7 % in 2013 (Fig. 1). We identified two general streams of consequentialist research: (1) observational studies where the investigators had no control over the assignment of the treatment (e.g., a policy) and (2) experimental studies where they did. That is, we only considered as “experimental” studies for which the manipulation of the exposure or intervention was under the control of the researcher (including simulation studies). A summary of the abstracts identified as consequentialist is provided in Table 1.
Fig. 1

Consequentialism in social epidemiology, SER annual meetings, 2009–13

Table 1

Summary of consequentialist studies in social epidemiology among abstracts accepted to SER annual meetings, 2009–2013





Primary exposure

Main outcome(s)

Observational studies



The Impact Of A Baby Bonus Payment On Birth Rates In Australia

Australia: Australian Bureau of Statistics 1997–2006

Baby bonus (policy)

Birth rate trends



Income Shocks And Very Low Weight Births Among Black Mothers In California


Earned Income Tax Credit (policy/shock)

Very low birth weight



A Multilevel-Based Study Of School Policy For Tobacco Control In Relation To Cigarette Smoking Among Children

Taiwan: school-based sample

School tobacco policy

Student smoking



The Impact Of Tobacco Control Policies On Disparities In Children’s Secondhand Smoke Exposure

United States: National Survey of Children’s Health (2003–2007)

Smoke-free legislation & cigarette excise taxes

Household tobacco use



Can Neighborhood Policy Interventions Improve Health? Quasi-Experimental Evidence From Medellín, Colombia

Medellın, Colombia

Construction of a gondola

Alcohol use, perceived violence, homicide, behavioral outcomes, illicit drug use



School Desegregation And Associated Decreases in Female Adolescent Births

United States Census Data (1970/1980)

School desegregation policy

Prevalence of births to adolescents



A Multilevel-Based Study Of Clustering In Cigarette Smoking Behavior In Relation To Individual- And School-Level Characteristics Among Schoolchildren

Taiwan: 52 schools

School-level (smoking policies, etc.) & individual-level (alcohol use, family smoking, etc.) characteristics




Causal Inferences From Longitudinal Marginal Structural Models: Effect Of 10 Years Participation In A Conditional Cash Transfer Program On Children’s Social-Behavioral Development

Mexico: Oportunidades (1997–2007)

Conditional cash transfer

Child’s social-behavioral development



Public Housing Policy Change And Residential Mobility Predict Low Birthweight-Preterm Birth

Atlanta, GA: cohort of women who lived in public housing from 1994 to 2007

Housing changes (moving from projects to private/mixed, moving as a result of housing policy change)

Low birth weight-preterm birth



The Impact Of Poverty Reduction Policy On Child And Adolescent Overweight: A Quasi- Experimental Analysis Of The Earned Income Tax Credit

United States: Children and Young Adults of the National Longitudinal Survey of Youth 1979 (1986–2006)

EITC-related increase in family income

Child overweight



The Supplemental Nutrition Assistance Program: Differences In Dietary Intake Between Participants And Low-Income Nonparticipants

United States: NHANES subset (1999–2008)

Supplemental Nutrition Assistance Program (SNAP) participation

Dietary intake patterns/quality



Decrease Trend In Common-Cause Adolescent Mortality Rates Following School Desegregation Legislation, Us 1968–1988

United States, 123 counties: Vital Statistics mortality records (1968–1988)

Changes in school desegregation policy

Childhood/adolescent mortality



Effects Of State-Level Policies On Risk Behavior In Men Who Have Sex With Men

United States: men recruited from sex-seeking websites

State-level gay rights policies

Sexual & alcohol risk behaviors



The Effect Of The Smoke-Free Ordinance On Acute Myocardial Infarction In South Carolina

Greenville, Charleston, & Spartanburg, SC: health claims data

Smoke-free ordinances (city level)

Acute myocardial infarction hospitalizations



Difference In Difference Estimates Of The Effect Of An Income Benefit Policy On Food Insecurity In Families With Young Children

Canada: cross-sectional Canadian health surveys (2000–2009)

Universal Child Benefit (policy)

Self-reported food insecurity



Use Of Propensity Score Matching To Assess The Effect Of A State Policy On Treatment Outcomes In U.S. HIV-Infected Persons

United States: NA-ACCORD (2001–2007)

Use of waiting lists for AIDS Drug Assistance Program enrollment in a participant’s state of residence

Time to ART initiation, HIV viral load suppression



Effects Of Minimum Legal Drinking Age On Alcohol And Marijuana Use: Evidence From Toxicological Testing Data For Fatally Injured Drivers Aged 16 To 25 Years

13 US States

Minimum legal drinking age

Alcohol and marijuana use



The Impact Of Tobacco Control Policies On Disparities In Maternal Smoking During Pregnancy.

29 US States

Tobacco control policies

Disparities in maternal smoking during pregnancy



Income As An Effect Modifier In The Relationship Between Smoke-Free Car Rules And Childhood Asthma

United States; BRFSS

Smoke-free car rules

Childhood asthma



Has Massachusetts Health Care Reform Worked For The Working Poor?

Massachusetts; Boston Area Community Health Survey

Massachusetts health care reform

Health insurance coverage



What Does The Massachusetts Experience Tell Us About The Potential Impact Of The Affordable Care Act On The Use Of Clinical Preventive Services?

Massachusetts; BRFSS

Massachusetts health care reform

Clinical preventive service use



Evaluating The Effects Of The Introduction Of Off- Sale Alcohol Outlets On Violent Crime

Lubbock, Texas

Policy that ended prohibition of off-sale alcohol outlets

Violent crime



Public Smoking Legislation And Hospitalizations For Ischemic Heart Disease In Atlantic Canada, 1996–2010: Mixed-Effects Modeling Of Multilevel Population Health Interventions


Public smoking legislation

Hospitalizations for ischemic heart disease



The Effect Of A Delivery Fee Exemption Policy On Socioeconomic Inequalities In Health Facility Deliveries: A Difference-In-Differences Analysis From Five West African Countries

West Africa

Fee exemption policy

Socioeconomic inequalities in health facility deliveries



Social Participation And Drug Use In A Cohort Of Brazilian Sex Workers

Corumbá, Brazil

An intervention that increased participation by drug users in external social groups

Drug use

Experimental studies



Evaluation Of A Community-Based Intervention To Increase Breast Cancer Screening And Early Detection Among Low-Income, African American Women

Savannah & Macon, Georgia

Breast cancer survivor testimonials on radio stations with wide African American listenership

Region-specific calls to the Cancer Information Service’s hotline



Modeling The Implications Of Non-Health Educational Policies For Cardiovascular Disease Incidence And Mortality

Not indicated

Sustained school funding

Cardiovascular disease



Long-Term Impact Of A School-Based Intervention On Knowledge, Dietary Intakes And Physical Activity Among Primary School Children

Not indicated (possibly Trinidad & Tobago)

School-based nutrition intervention program

Knowledge, dietary intakes and physical activity among children attending primary school



How Long Does It Take To Reduce Inequalities In Lung Cancer Incidence Through Smoking Policies?

Not applicable

Projected increases in cigarette prices & health advertisements

Lung cancer & gap by SES



Evaluation Of A Community-Based Intervention To Increase Breast Cancer Screening And Early Detection Among Low-Income, African American Women

Savannah & Macon, GA: National Breast and Cervical Cancer Early Detection Program

Multimedia breast cancer awareness campaign

Mammogram receipt



Performance Of Mobile Units Of The Quebec Breast Cancer Screening Program, 2002–2009

Quebec: Breast Cancer Screening Program

Mobile screening units

Screening participation & performance



Financial Incentive For Promoting Good-Health Behaviors In The Workplace

A large agribusiness (location not indicated)

Financial incentive worksite wellness program

General health & health behaviors, including BMI, lipids, fat intake, mental well-being



Mental Health Effects Of A Housing Mobility Program By Vulnerable Subgroups: Who Benefits From Moves To Low-Poverty Neighborhoods?

United States: Moving to Opportunity study

MTO voucher receipt (by gender & baseline health vulnerability)

Psychological distress



The Effect Of Older Education On Knowledge About Healthy Lifestyles. A Randomized Controlled Trial

Arak, Iran

Educational intervention

Knowledge re: healthy lifestyles



Did Adolescents In Higher SES Families Differentially Benefit From A Housing Mobility Intervention?

United States: Moving to Opportunity study

Moving voucher receipt family SES at baseline

Psychological distress & behavioral problems



Effects Of Housing Mobility And Lower Poverty Neighborhoods On Adolescent Asthma: The Moving To Opportunity Experiment

United States: Moving to Opportunity study

Moving voucher receipt

Youth asthma



The Microclinic Health Program: A Sowrk- Based Intervention For Weight Loss And Diabetes Risk Management

Amman, Jordan

4 month diabetes education program

Clinical markers (weight, Body Mass Index [BMI], and Hemoglobin A1c [HbA1c])



The Moving To Opportunity Experiment And Adolescent Psychological Distress: Were Effects Mediated By Neighborhood Characteristics?

United States; Moving to Opportunity study

Randomized section 8 vouchers or public housing (control)

Adolescent psychological distress



Effects Of A Housing Mobility Experiment On Neighborhood Quality.

United States; Moving to Opportunity study

Randomized section 8 vouchers or public housing (control)

Neighborhood economic conditions, including neighborhood disorder, collective efficacy, and violent crime



A Social Network Intervention For Reducing Influenza- Like Illness Transmission: The Ex-Flu Cluster Randomized Trial

United States: university students living in residence halls

Randomized to 3-day or 6-day sequestration vs. none (control)

Influenza-like illness



Gender Differences In Effect Of Reduction in Alcohol Use On HIV Risk Behaviors In Tomsk, Russian Federation

Tomsk, Russian Federation

Randomized alcohol treatment

HIV risk behavior

Observational Studies

Among the 41 papers that evaluated the effect of a specific intervention, 25 (61 %) were observational studies. In most cases, these studies evaluated the effects of changes in (1) social policies and programs (abstracts 1, 2, 5, 6, 8–10, 12, 13, 15, 25) and (2) public health and health services interventions (abstracts 11, 16, 20, 21, 24), including several policies targeting alcohol (abstracts 17, 22) and tobacco (abstracts 3, 4, 7, 14, 18, 19, 23) use.

Experimental Studies

A total of 16 of the 41 (39 %) consequentialist papers used an experimental design. Several studies evaluated the effects of randomized interventions (abstracts 34, 40, 41), including the Moving to Opportunity (MTO) experiment (abstracts 33, 35, 36, 38, 39). Other studies evaluated the non-random assignment of participants to various school, workplace, and community-based health interventions (abstracts 26, 28, 30–32, 37). Two studies used simulated data to evaluate the effect of specific interventions, including an educational policy that provided sustained school funding (abstract 27) and a smoking policy that increased cigarette prices and health advertisements (abstract 29).


We evaluated the extent to which recent epidemiologic research on the study of social determinants of health can be considered consequentialist. We identified few studies that evaluated specific social interventions for improving health or reducing health inequalities (including our own work). This is not a novel observation. Prior work has hinted at imbalances between social epidemiology’s research goals and its research outputs. Bonneux noted in 2007 that, while our field faces “a feast of descriptive studies of socio-economic causes of ill health, we still face a famine of evaluative intervention studies” [19]. Our informal review of the current state of the science suggests that the scarcity of evaluative intervention studies has persisted since this observation.

Whether the predominance of descriptive over consequentialist work in social epidemiology impedes our ability to act on the social determinants of health is unclear. But if we assume there is at least some utility for policy-related evidence demonstrating how social interventions can be used to improve health and reduce social inequalities in health, the current shortage of consequentialist studies certainly does not help. Reiterating a decade-old argument for incorporating “experimental and quasi-experimental designs evaluating the potential health impacts of social and economic policies” [20], we suggest (again) that a greater emphasis on evaluations could make social epidemiology more policy relevant. Before underlining some of the barriers that may contribute to the underutilization of evaluation studies and recommending some practical solutions, we illustrate the urgency for a more consequentialist social epidemiology using a case study on neighborhood effects research.

Neighborhood Effects on Health

There has been an explosion of interest in neighborhood health effects over the past two decades, due in large part to the proliferation of methods for multilevel statistical analysis [21, 22]. This was evident in our review—nearly one-quarter of social epidemiology abstracts presented at SER over the past 5 years could be characterized as neighborhood effects studies. Like the study of social determinants more generally, much of this work has been observational, including some descriptive research (e.g., estimating social or racial inequalities in health across neighborhood boundaries) as well as “associational” studies that have frequently examined whether characteristics of neighborhoods—including their social conditions, physical environments, and racial composition—influence health [21, 22, 23, 24]. Despite occupying a substantial amount of social epidemiology’s research activity, we categorized few (n = 7) neighborhood effects studies as consequentialist; important methodological and substantive challenges might explain this unevenness.

An established methodological critique of research into neighborhood effects is that the influence of neighborhood conditions on health can seldom be characterized as causal [25, 26]. Substantial attention has been given to the non-exchangeability of individuals living in different types of neighborhoods owing to neighborhood selection, i.e., the non-random migration of individuals into and out of particular neighborhoods based on characteristics that may be related to their health. By limiting the exchangeability of neighborhood contrasts, neighborhood selection presents a “methodological wall” that usually limits causal inference in observational studies of neighborhood effects and, as others have noted [25, 27, 28], challenges the policy relevance of research findings. Moreover, varying definitions of what constitutes a neighborhood—often (though not always) a consequence of reliance on administrative data—can lead to heterogeneity for the same exposure-outcome association [29, 30].

Our review, which attempted to apply a consequentialist lens to the literature, identifies an additional challenge to the policy relevance of most existing neighborhood effects research. Consider the following thought experiment: were we to live in a counterfactual world where neighborhood selection was not a pervasive methodological challenge, and could conclude that neighborhood conditions have a causal impact on health, would we be able to recommend evidence-based interventions for neighborhood interventions to improve health? In most cases, these interventions would be difficult to articulate, since neighborhood effects studies have not sought to identify specific interventions for manipulating features of the neighborhood environment that influence health [31]. In many cases, neighborhood exposures are defined based on summary indexes (such as deprivation scores) derived from data reduction techniques [32], which are useful in some respects but tell us little about which specific exposures matter for health. But even isolating simple, modifiable neighborhood factors tells us little about whether or how to change them. Even if we knew, for example, that neighborhood poverty, one of the most frequently assessed neighborhood-level exposures, exerted a causal effect on health, it would be difficult given the current state of the science to identify a real-world intervention to influence distributions of poverty across neighborhoods.

The distinction between the methodological challenge posed by neighborhood selection and the challenge of identifying interventions for improving health associated with adverse neighborhood conditions is illustrated by the Moving to Opportunity (MTO) study. MTO was a housing relocation experiment conducted in five US cities that provided families living in public housing in high poverty neighborhoods with the opportunity to move to lower poverty neighborhoods [33]. The opportunity to move was randomly assigned through the use of housing vouchers, thereby solving the selection issue. However, MTO does not provide a clear answer to the consequentialist question of how neighborhood environments could be changed to improve population health because it examined the effect of moving families to new neighborhoods rather than the effect of a policy-relevant neighborhood-level intervention on current neighborhoods [21]. Five of the seven papers (abstracts 33, 35, 36, 38, 39) that analyzed the effect of a specific neighborhood intervention were evaluations of the MTO housing mobility experiment.

The challenges to a more consequentialist neighborhood effects epidemiology are not insurmountable. However, given the increasing frequency of studies evaluating neighborhood effects [34], a course correction is sorely needed. Within the domain of observational research, quasi-experimental evaluations can be used to estimate the effects of specific policy-relevant changes in neighborhood environments, while simultaneously addressing the issue of neighborhood selection. Identifying quasi-experiments is, however, difficult and published examples in the neighborhood effects literature are sparse. One study identified by our review (abstract 5) evaluated the effect of an intervention in neighborhood physical infrastructure, specifically the construction of a public transit system in Medellín, Colombia that connected isolated low-income neighborhoods to the city center, on levels of reported violence using a propensity score-matched difference-in-differences design [35•]. This study, by exploiting arguably exogenous variation in neighborhood conditions, avoided the potential self-selection of individuals into different neighborhoods. Moreover, by examining the effect of a “natural experiment,” the study identified a specific intervention for influencing health (or in this case violence) by changing the neighborhood environment, although the generalizability of this intervention to other contexts might be limited. Another study used quasi-experimental [36] and experimental [37] methods to estimate the effects of a neighborhood greening initiative on health and safety in Philadelphia. Though experimental studies for estimating policy-relevant neighborhood effects are rare, they may pay large dividends in terms of strengthening the evidence base for neighborhood effects. According to Oakes, “for those wishing to stay focused on neighborhood effects, (experimental) research into specific policy-relevant changes of neighborhood environments would be most helpful” [38].

Different Approaches to Consequentialism

Although a small proportion of studies in our sample were classified as consequentialist, these studies adopted a variety of approaches, which suggests that consequential research in social epidemiology can be delivered through a diversity of research designs. We do not need to rehearse here the ample arguments for not adhering rigidly to hierarchies of evidence [39], but different design and analysis strategies are likely to have strengths and weaknesses for a given research question. Randomized interventions have clear advantages with respect to internal validity, but social interventions often lead to lots of non-compliance (as in the MTO example above), and even trials with good compliance may not be generalizable to other contexts. Recent work has formalized the assumptions and outlined methods for generalizing RCT evidence to other populations [40], and this could perhaps be applied more widely. For non-randomized designs, prioritizing high-quality natural experiments seems another promising strategy. Natural experiments where treatment assignment is not under researcher control—but is truly random (e.g., lottery) or as good as random—maintain some of the advantages of RCTs, but may also be more generalizable since they occur in less selected populations. But natural experiments are often hard to come by, and require solid combinations of quantitative and qualitative knowledge about the intervention to be credible [41]. In other cases, there may exist no reasonable combination of available data and identification strategy to estimate the impact of a given intervention. A variety of simulation models have become popular for answering questions about the impact of interventions on population health [42, 43, 44], and can be extended to assess impact on health inequalities [45, 46]. Clearly, they may have benefits with respect to adaptability and control of the researcher over the system parameters. However, convincing simulation models need validation [47] and depend heavily on the quality of the input parameters. Given the small fraction of articles in our sample that provided any evidence on the impact of interventions, this may make developing and applying simulation models for consequential social epidemiology challenging.


There were several limitations to our approach. First, the definition of consequentialism is itself subjective. Our characterization of consequentialist research as research that evaluates or simulates a specific intervention follows from our interpretation of the prior literature [1, 2•, 3, 4]; however, reasonable social epidemiologists can disagree about the policy relevance of a given study (though we rarely did). This caveat considered a broader definition would be unlikely to qualitatively change our main conclusions regarding the extent of consequentialism in social epidemiology. Second, our review was based on abstracts accepted to the SER annual meeting, and this selection of work is an imperfect proxy for the current state of social epidemiology. SER draws submissions primarily from North America, and the USA in particular, and work from other regions is likely underrepresented. Third, our review was based on accepted abstracts, not published peer-reviewed papers; however, this might be considered a strength and more accurately reflect current research priorities given pervasive publication bias. Fourth, the abstracts included in 2011 were submitted to the World Congress of Epidemiology, a joint meeting of professional organizations including SER; the abstracts selected from this year represent a broader catchment area.


As Galea and Link recently noted, there is much that we can learn from studying interventions [14]. However, “recalibrating” our field and shifting our propensity for descriptive or “etiologic” analyses toward evaluative research will require changing some deeply entrenched behaviors. For one, it will require us to modify our approach to training and integrate interdisciplinary methods for analyzing interventions. Few training programs targeted to social epidemiologists provide an adequate foundation for applying experimental or quasi-experimental methods for evaluating the effects of interventions. Certainly, some of the methods that have been a core component of standard epidemiologic curricula, such as randomized controlled trials, are apposite to the evaluation of social interventions, although more emphasis on treatments defined at the “cluster” level, as well as the challenges this entails, is needed. However, by and large, quasi-experimental methods frequently applied in policy evaluation, including difference-in-differences and instrumental variable approaches, are not part of the social epidemiologist’s armamentarium. They should be. We are starting to see a translation of these methods from the economics literature into epidemiologic parlance [48], which will hopefully stimulate their utilization, but exposing students to interdisciplinary methods for impact evaluation during their training programs can help advance a more consequentialist social epidemiology. A second challenge to producing consequentialist evidence is asking a policy-relevant research question. As we illustrated in our case study, estimating a causal effect is a necessary but not sufficient condition for producing consequentialist evidence because research estimating the causal effects of treatments that do not correspond to some practicable intervention is unlikely to be considered policy-relevant [49, 50]. Involving “knowledge users,” including non-governmental and governmental actors, in the research process and identifying relevant research gaps can help in designing and articulating policy-relevant research questions [17, 51]. Third, periodically and systematically reviewing the literature concerning the health effects of specific social interventions can be useful for synthesizing results, identifying research gaps, understanding heterogeneity, and reducing redundancies [52••, 53].

Concluding Remarks

Social epidemiology has made substantial progress since its inception. The investigation of how multilevel determinants of health and health inequalities, spanning from the molecular to the individual to the macro-level, has distinguished social epidemiology from other branches of epidemiology and formalized its role in the social sciences. However, for it to consolidate its position and maintain its relevance, social epidemiology will have to demonstrate that it cannot only elucidate how social factors influence population health but also prescribe evidence-based interventions for acting on them. Through this review and commentary, we have examined the extent of and challenges to a more consequentialist epidemiology. Overcoming these challenges will require another phase of “intense socialization and learning” [54] that include engaging with key stakeholders, articulating policy-relevant research questions, and applying interdisciplinary research designs and methods for answering them.


Compliance with Ethics Guidelines

Conflict of Interest

A Nandi and S Harper declare that they have no conflict of interest.

Human and Animal Rights and Informed Consent

All studies by the authors involving animal and/or human subjects were performed after approval by the appropriate institutional review boards. When required, written informed consent was obtained from all participants.

Source of Funding

The authors acknowledge funding from the Canadian Institutes of Health Research Operating Grant, “Examining the impact of social policies on health equity” (ROH-115209). AN is supported by the Canada Research Chairs program. SH is supported by a Chercheur-boursier Junior 2 from the Fonds de recherche du Québec–Santé.


Papers of particular interest, published recently, have been highlighted as: • Of importance •• Of major importance

  1. 1.
    Marks JS. Epidemiology, public health, and public policy. Prev Chron Dis. 2009;6(4):A134.Google Scholar
  2. 2.•
    Galea S. An argument for a consequentialist epidemiology. Am J Epidemiol. 2013;178(8):1185–91. Provides an overview of what consequentialist program of research would look like in epidemiology and how we might achieve it.PubMedCrossRefGoogle Scholar
  3. 3.
    Cates W. Invited Commentary: Consequential (ist) epidemiology: let’s seize the day. Am J Epidemiol. 2013;178(8):1192–4.PubMedCrossRefGoogle Scholar
  4. 4.
    O’Campo P. Are we producing the right kind of actionable evidence for the social determinants of health? J Urban Health. 2012;89(6):881–93.PubMedCentralPubMedCrossRefGoogle Scholar
  5. 5.
    Kaufman JS. Social epidemiology. In: Rothman KJ, Greenland S, Lash TL, editors. Modern epidemiology. 3rd ed. Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins; 2008. p. 533–48.Google Scholar
  6. 6.
    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 Health. Geneva: World Health Organization, Commission on Social Determinants of Health; 2008.Google Scholar
  7. 7.
    Gakidou E, Cowling K, Lozano R, Murray CJL. Increased educational attainment and its effect on child mortality in 175 countries between 1970 and 2009: a systematic analysis. Lancet. 2010;376(9745):959–74.PubMedCrossRefGoogle Scholar
  8. 8.
    Cleland JG, Van Ginneken JK. Maternal education and child survival in developing countries: the search for pathways of influence. Soc Sci Med. 1988;27(12):1357–68.PubMedCrossRefGoogle Scholar
  9. 9.
    Singh GK, Kogan MD. Persistent socioeconomic disparities in infant, neonatal, and postneonatal mortality rates in the United States, 1969–2001. Pediatrics. 2007;119(4):e928–39.PubMedCrossRefGoogle Scholar
  10. 10.
    Ware H. Effects of maternal education, women’s roles, and child care on child mortality. Pop Dev Rev. 1984;10:191–214.CrossRefGoogle Scholar
  11. 11.
    Marmot M, Friel S, Bell R, Houweling TAJ, Taylor S. Closing the gap in a generation: health equity through action on the social determinants of health. Lancet. 2008;372(9650):1661–9.PubMedCrossRefGoogle Scholar
  12. 12.•
    Oakes JM. Invited commentary: paths and pathologies of social epidemiology. Am J Epidemiol. 2013;178(6):850–1. Describes several concrete steps that social epidemiology could take to make its research more successful.PubMedCrossRefGoogle Scholar
  13. 13.
    Muntaner C. Invited commentary: on the future of social epidemiology—a case for scientific realism. Am J Epidemiol. 2013;178(6):852–7.PubMedCrossRefGoogle Scholar
  14. 14.
    Galea S, Link BG. Six paths for the future of social epidemiology. Am J Epidemiol. 2013;178(6):843–9.PubMedCentralPubMedCrossRefGoogle Scholar
  15. 15.
    O’Campo P. Are we producing the right kind of actionable evidence for the social determinants of health? J Urban Health. 2012;89(6):881–93.PubMedCentralPubMedCrossRefGoogle Scholar
  16. 16.
    Mackenbach JP. Has the English strategy to reduce health inequalities failed? Soc Sci Med. 2010;71(7):1249–53. discussion 54–8.PubMedCrossRefGoogle Scholar
  17. 17.
    Petticrew M, Whitehead M, Macintyre SJ, Graham H, Egan M. Evidence for public health policy on inequalities: 1: the reality according to policymakers. J Epidemiol Community Health. 2004;58(10):811–6.PubMedCentralPubMedCrossRefGoogle Scholar
  18. 18.
    Skrabanek P. The poverty of epidemiology. Perspect Biol Med. 1992;35(2):182–5.PubMedCrossRefGoogle Scholar
  19. 19.
    Bonneux L. From evidence based bioethics to evidence based social policies. Eur J Epidemiol. 2007;22(8):483–5.PubMedCentralPubMedCrossRefGoogle Scholar
  20. 20.
    Berkman LF. Introduction: seeing the forest and the trees—from observation to experiments in social epidemiology. Epidemiol Rev. 2004;26:2–6.PubMedCrossRefGoogle Scholar
  21. 21.
    Diez Roux AV. Neighborhoods and health: where are we and were do we go from here? Rev Epidemiol Sante Publique. 2007;55(1):13–21.PubMedCentralPubMedCrossRefGoogle Scholar
  22. 22.
    Diez Roux AV. Investigating neighborhood and area effects on health. Am J Public Health. 2001;91(11):1783–9.PubMedCentralPubMedCrossRefGoogle Scholar
  23. 23.
    Diez Roux AV, Mair C. Neighborhoods and health. Ann N Y Acad Sci. 2010;1186:125–45.PubMedCrossRefGoogle Scholar
  24. 24.
    Nandi A, Kawachi I. Neighborhood effects on mortality. In: Rogers RG, Crimmins EM, editors. International handbook of adult mortality. Dordrecht: Springer; 2011. p. 413–39.CrossRefGoogle Scholar
  25. 25.
    Oakes JM. The (mis)estimation of neighborhood effects: causal inference for a practicable social epidemiology. Soc Sci Med. 2004;58(10):1929–52.PubMedCrossRefGoogle Scholar
  26. 26.
    Oakes JM. Commentary: advancing neighbourhood-effects research—selection, inferential support, and structural confounding. Int J Epidemiol. 2006;35(3):643–7.PubMedCrossRefGoogle Scholar
  27. 27.
    Diez Roux AV. Estimating neighborhood health effects: the challenges of causal inference in a complex world. Soc Sci Med. 2004;58(10):1953–60.PubMedCrossRefGoogle Scholar
  28. 28.
    Harding DJ. Counterfactual models of neighborhood effects: the effect of neighborhood poverty on dropping out and teenage pregnancy. Am J Sociol. 2003;109(3):676–719.CrossRefGoogle Scholar
  29. 29.
    Openshaw S. The modifiable areal unit problem. Norwich: Geo Books; 1983.Google Scholar
  30. 30.
    Parenteau MP, Sawada MC. The modifiable areal unit problem (MAUP) in the relationship between exposure to NO2 and respiratory health. Int J Health Geogr. 2011;10:58.PubMedCentralPubMedCrossRefGoogle Scholar
  31. 31.
    Morenoff J, Diez Roux A, Hansen B, Osypuk TL. Residential environments and obesity: what can we learn about policy interventions from observational studies? Making Americans healthier: social and economic policy as health policy. New York: Russell Sage; 2008. p. 309–43.Google Scholar
  32. 32.
    Messer LC, Kaufman JS. Measuring health inequalities. In: Oakes JM, Kaufman JS, editors. Using census data to approximate neighborhood effects. San Francisco: Jossey-Bass; 2006. p. 209–36.Google Scholar
  33. 33.
    Leventhal T, Brooks-Gunn J. Moving to opportunity: an experimental study of neighborhood effects on mental health. Am J Public Health. 2003;93(9):1576–82.PubMedCentralPubMedCrossRefGoogle Scholar
  34. 34.
    Oakes JM, Andrade KE, Biyoow I, Cowan LT. Twenty Years of Neighborhood Effect Research: An Assessment. Curr Epidemiol Rep. forthcoming.Google Scholar
  35. 35.•
    Cerdá M, Morenoff JD, Hansen BB, et al. Reducing violence by transforming neighborhoods: a natural experiment in Medellín, Colombia. Am J Epidemiol. 2012;175(10):1045–53. An excellent example of a well-designed neighborhood effects study focused on providing evidence on the health impacts of a specific intervention.PubMedCentralPubMedCrossRefGoogle Scholar
  36. 36.
    Branas CC, Cheney RA, MacDonald JM, Tam VW, Jackson TD, Ten Have TR. A difference-in-differences analysis of health, safety, and greening vacant urban space. Am J Epidemiol. 2011;174:1296–306.PubMedCentralPubMedCrossRefGoogle Scholar
  37. 37.
    Garvin EC, Cannuscio CC, Branas CC. Greening vacant lots to reduce violent crime: a randomised controlled trial. Inj Prev. 2013;19(3):198–203.PubMedCentralPubMedCrossRefGoogle Scholar
  38. 38.
    Oakes JM. Invited commentary: repeated measures, selection bias, and effect identification in neighborhood effect studies. Am J Epidemiol. 2014;180:785–7.PubMedCrossRefGoogle Scholar
  39. 39.
    Glasziou P, Vandenbroucke JP, Chalmers I. Assessing the quality of research. BMJ. 2004;328(7430):39–41.PubMedCentralPubMedCrossRefGoogle Scholar
  40. 40.
    Stuart EA, Bradshaw CP, Leaf PJ. Assessing the generalizability of randomized trial results to target populations. Prev Sci. 2014. doi: 10.1007/s11121-014-0513-z.PubMedGoogle Scholar
  41. 41.
    Dunning T. Natural experiments in the social sciences: a design-based approach. Cambridge: Cambridge University Press; 2012.CrossRefGoogle Scholar
  42. 42.
    Galea S, Riddle M, Kaplan GA. Causal thinking and complex system approaches in epidemiology. Int J Epidemiol. 2010;39(1):97–106.PubMedCentralPubMedCrossRefGoogle Scholar
  43. 43.
    Auchincloss AH, Diez Roux AV. A new tool for epidemiology: the usefulness of dynamic-agent models in understanding place effects on health. Am J Epidemiol. 2008;168(1):1–8.PubMedCrossRefGoogle Scholar
  44. 44.
    Wolfson MC. POHEM—a framework for understanding and modelling the health of human populations. World Health Stat Q. 1997;47:157–76.Google Scholar
  45. 45.
    Smith BT, Smith PM, Harper S, Manuel DG, Mustard CA. Reducing social inequalities in health: the role of simulation modelling in chronic disease epidemiology to evaluate the impact of population health interventions. J Epidemiol Community Health. 2014;68(4):384–9.PubMedCentralPubMedCrossRefGoogle Scholar
  46. 46.
    Goldie SJ, Daniels N. Model-based analyses to compare health and economic outcomes of cancer control: inclusion of disparities. J Natl Cancer Inst. 2011;103(18):1373–86.PubMedCentralPubMedCrossRefGoogle Scholar
  47. 47.
    Kopec JA, Fines P, Manuel DG, et al. Validation of population-based disease simulation models: a review of concepts and methods. BMC Public Health. 2010;10:710.PubMedCentralPubMedCrossRefGoogle Scholar
  48. 48.
    Bor J, Moscoe E, Mutevedzi P, Newell ML, Barnighausen T. Regression discontinuity designs in epidemiology: causal inference without randomized trials. Epidemiology. 2014;25(5):729–37.PubMedCentralPubMedCrossRefGoogle Scholar
  49. 49.
    Harper S, Strumpf EC. Commentary: social epidemiology: questionable answers and answerable questions. Epidemiology. 2012;23(6):795–8.PubMedCrossRefGoogle Scholar
  50. 50.
    Hernán MA, Taubman SL. Does obesity shorten life? The importance of well-defined interventions to answer causal questions. Int J Obes. 2008;32(S3):S8–S14.CrossRefGoogle Scholar
  51. 51.
    Whitehead M, Petticrew M, Graham H, Macintyre SJ, Bambra C, Egan M. Evidence for public health policy on inequalities: 2: assembling the evidence jigsaw. J Epidemiol Community Health. 2004;58(10):817–21.PubMedCentralPubMedCrossRefGoogle Scholar
  52. 52.••
    Glasziou P, Altman DG, Bossuyt P, et al. Reducing waste from incomplete or unusable reports of biomedical research. Lancet. 2014;383(9913):267–76. Makes a compelling argument for how to improve the quality of reporting of biomedical research—and why it matters.PubMedCrossRefGoogle Scholar
  53. 53.
    Tugwell P, Petticrew M, Kristjansson E, et al. Assessing equity in systematic reviews: realising the recommendations of the Commission on Social Determinants of Health. BMJ. 2010;341:c4739.PubMedCrossRefGoogle Scholar
  54. 54.
    Susser M, Susser E. Choosing a future for epidemiology: II. From black box to Chinese boxes and eco-epidemiology. Am J Public Health. 1996;86(5):674–7.PubMedCentralPubMedCrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2014

Authors and Affiliations

  1. 1.Institute for Health and Social PolicyMcGill UniversityMontrealCanada
  2. 2.Department of Epidemiology, Biostatistics & Occupational HealthMcGill UniversityMontrealCanada
  3. 3.Centre on Population DynamicsMcGill UniversityMontrealCanada

Personalised recommendations