Marginal and Conditional Confounding Using Logits.

This article presents two ways of quantifying confounding using logistic response models for binary outcomes. Drawing on the distinction between marginal and conditional odds ratios in statistics, we define two corresponding measures of confounding (marginal and conditional) that can be recovered from a simple standardization approach. We investigate when marginal and conditional confounding may differ, outline why the method by Karlson, Holm, and Breen recovers conditional confounding under a "no interaction"-assumption, and suggest that researchers may measure marginal confounding by using inverse probability weighting. We provide two empirical examples that illustrate our standardization approach.

Effects of increased body mass index on employment status: a Mendelian randomisation study.

BACKGROUND: The obesity epidemic may have substantial implications for the global workforce, including causal effects on employment, but clear evidence is lacking. Obesity may prevent people from being in paid work through poor health or through social discrimination. We studied genetic variants robustly associated with body mass index (BMI) to investigate its causal effects on employment. DATASET/METHODS: White UK ethnicity participants of working age (men 40-64 years, women 40-59 years), with suitable genetic data were selected in the UK Biobank study (N = 230,791). Employment status was categorised in two ways: first, contrasting being in paid employment with any other status; and second, contrasting being in paid employment with sickness/disability, unemployment, early retirement and caring for home/family. Socioeconomic indicators also investigated were hours worked, household income, educational attainment and Townsend deprivation index (TDI). We conducted observational and two-sample Mendelian randomisation (MR) analyses to investigate the effect of increased BMI on employment-related outcomes. RESULTS: Regressions showed BMI associated with all the employment-related outcomes investigated. MR analyses provided evidence for higher BMI causing increased risk of sickness/disability (OR 1.08, 95% CI 1.04, 1.11, per 1 Kg/m2 BMI increase) and decreased caring for home/family (OR 0.96, 95% CI 0.93, 0.99), higher TDI (Beta 0.038, 95% CI 0.018, 0.059), and lower household income (OR 0.98, 95% CI 0.96, 0.99). In contrast, MR provided evidence for no causal effect of BMI on unemployment, early retirement, non-employment, hours worked or educational attainment. There was little evidence for causal effects differing by sex or age. Robustness tests yielded consistent results. DISCUSSION: BMI appears to exert a causal effect on employment status, largely by affecting an individual's health rather than through increased unemployment arising from social discrimination. The obesity epidemic may be contributing to increased worklessness and therefore could impose a substantial societal burden.

Standard multiple imputation of survey data didn't perform better than simple substitution in enhancing an administrative dataset: the example of self-rated health in England.

BACKGROUND: Health surveys provide a rich array of information but on relatively small numbers of individuals and evidence suggests that they are becoming less representative as response levels fall. Routinely collected administrative data offer more extensive population coverage but typically comprise fewer health topics. We explore whether data combination and multiple imputation of health variables from survey data is a simple and robust way of generating these variables in the general population. METHODS: We use the UK Integrated Household Survey and the English 2011 population census both of which included self-rated general health. Setting aside the census self-rated health data we multiply imputed self-rated health responses for the census using the survey data and compared these with the actual census results in 576 unique groups defined by age, sex, housing tenure and geographic region. RESULTS: Compared with original census data across the groups, multiply imputed proportions of bad or very bad self-rated health were not a markedly better fit than those simply derived from the survey proportions. CONCLUSION: While multiple imputation may have the potential to augment population data with information from surveys, further testing and refinement is required.

The long-term health effects of attending a selective school: a natural experiment.

BACKGROUND: Education is widely associated with better physical and mental health, but isolating its causal effect is difficult because education is linked with many socioeconomic advantages. One way to isolate education's effect is to consider environments where similar students are assigned to different educational experiences based on objective criteria. Here we measure the health effects of assignment to selective schooling based on test score, a widely debated educational policy. METHODS: In 1960s Britain, children were assigned to secondary schools via a test taken at age 11. We used regression discontinuity analysis to measure health differences in 5039 people who were separated into selective and non-selective schools this way. We measured selective schooling's effect on six outcomes: mid-life self-reports of health, mental health, and life limitation due to health, as well as chronic disease burden derived from hospital records in mid-life and later life, and the likelihood of dying prematurely. The analysis plan was accepted as a registered report while we were blind to the health outcome data. RESULTS: Effect estimates for selective schooling were as follows: self-reported health, 0.1 worse on a 4-point scale (95%CI - 0.2 to 0); mental health, 0.2 worse on a 16-point scale (- 0.5 to 0.1); likelihood of life limitation due to health, 5 percentage points higher (- 1 to 10); mid-life chronic disease diagnoses, 3 fewer/100 people (- 9 to + 4); late-life chronic disease diagnoses, 9 more/100 people (- 3 to + 20); and risk of dying before age 60, no difference (- 2 to 3 percentage points). Extensive sensitivity analyses gave estimates consistent with these results. In summary, effects ranged from 0.10-0.15 standard deviations worse for self-reported health, and from 0.02 standard deviations better to 0.07 worse for records-derived health. However, they were too imprecise to allow the conclusion that selective schooling was detrimental. CONCLUSIONS: We found that people who attended selective secondary school had more advantaged economic backgrounds, higher IQs, higher likelihood of getting a university degree, and better health. However, we did not find that selective schooling itself improved health. This lack of a positive influence of selective secondary schooling on health was consistent despite varying a wide range of model assumptions.

Impact of Political Economy on Population Health: A Systematic Review of Reviews.

Background. Although there is a large literature examining the relationship between a wide range of political economy exposures and health outcomes, the extent to which the different aspects of political economy influence health, and through which mechanisms and in what contexts, is only partially understood. The areas in which there are few high-quality studies are also unclear. Objectives. To systematically review the literature describing the impact of political economy on population health. Search Methods. We undertook a systematic review of reviews, searching MEDLINE, Embase, International Bibliography of the Social Sciences, ProQuest Public Health, Sociological Abstracts, Applied Social Sciences Index and Abstracts, EconLit, SocINDEX, Web of Science, and the gray literature via Google Scholar. Selection Criteria. We included studies that were a review of the literature. Relevant exposures were differences or changes in policy, law, or rules; economic conditions; institutions or social structures; or politics, power, or conflict. Relevant outcomes were any overall measure of population health such as self-assessed health, mortality, life expectancy, survival, morbidity, well-being, illness, ill health, and life span. Two authors independently reviewed all citations for relevance. Data Collection and Analysis. We undertook critical appraisal of all included reviews by using modified Assessing the Methodological Quality of Systematic Reviews (AMSTAR) criteria and then synthesized narratively giving greater weight to the higher-quality reviews. Main Results. From 4912 citations, we included 58 reviews. Both the quality of the reviews and the underlying studies within the reviews were variable. Social democratic welfare states, higher public spending, fair trade policies, extensions to compulsory education provision, microfinance initiatives in low-income countries, health and safety policy, improved access to health care, and high-quality affordable housing have positive impacts on population health. Neoliberal restructuring seems to be associated with increased health inequalities and higher income inequality with lower self-rated health and higher mortality. Authors' Conclusions. Politics, economics, and public policy are important determinants of population health. Countries with social democratic regimes, higher public spending, and lower income inequalities have populations with better health. There are substantial gaps in the synthesized evidence on the relationship between political economy and health, and there is a need for higher-quality reviews and empirical studies in this area. However, there is sufficient evidence in this review, if applied through policy and practice, to have marked beneficial health impacts. Public Health Implications. Policymakers should be aware that social democratic welfare state types, countries that spend more on public services, and countries with lower income inequalities have better self-rated health and lower mortality. Research funders and researchers should be aware that there remain substantial gaps in the available evidence base. One such area concerns the interrelationship between governance, polities, power, macroeconomic policy, public policy, and population health, including how these aspects of political economy generate social class processes and forms of discrimination that have a differential impact across social groups. This includes the influence of patterns of ownership (of land and capital) and tax policies. For some areas, there are many lower-quality reviews, which leave uncertainties in the relationship between political economy and population health, and a high-quality review is needed. There are also areas in which the available reviews have identified primary research gaps such as the impact of changes to housing policy, availability, and tenure.

Interpreting mutual adjustment for multiple indicators of socioeconomic position without committing mutual adjustment fallacies.

Research into the effects of Socioeconomic Position (SEP) on health will sometimes compare effects from multiple, different measures of SEP in "mutually adjusted" regression models. Interpreting each effect estimate from such models equivalently as the "independent" effect of each measure may be misleading, a mutual adjustment (or Table 2) fallacy. We use directed acyclic graphs (DAGs) to explain how interpretation of such models rests on assumptions about the causal relationships between those various SEP measures. We use an example DAG whereby education leads to occupation and both determine income, and explain implications for the interpretation of mutually adjusted coefficients for these three SEP indicators. Under this DAG, the mutually adjusted coefficient for education will represent the direct effect of education, not mediated via occupation or income. The coefficient for occupation represents the direct effect of occupation, not mediated via income, or confounded by education. The coefficient for income represents the effect of income, after adjusting for confounding by education and occupation. Direct comparisons of mutually adjusted coefficients are not comparing like with like. A theoretical understanding of how SEP measures relate to each other can influence conclusions as to which measures of SEP are most important. Additionally, in some situations adjustment for confounding from more distal SEP measures (like education and occupation) may be sufficient to block unmeasured socioeconomic confounding, allowing for greater causal confidence in adjusted effect estimates for more proximal measures of SEP (like income).

How Healthy Are Survey Respondents Compared with the General Population?: Using Survey-linked Death Records to Compare Mortality Outcomes.

BACKGROUND: National surveys are used to capture US health trends and set clinical guidelines, yet the sampling frame often includes those in noninstitutional households, potentially missing those most vulnerable for poor health. Declining response rates in national surveys also represent a challenge, and existing inputs to survey weights have limitations. We compared mortality rates between those who respond to surveys and the general population over time. METHODS: Survey respondents from 20 waves of the National Health Interview Survey from 1990 to 2009 who have been linked to death records through 31 December 2011 were included. For each cohort in the survey, we estimated their mortality rates along with that cohort's mortality rate in the census population using vital statistics records, and differences were examined using Poisson models. RESULTS: In all years, survey respondents had lower mortality rates compared with the general population when data were both weighted and unweighted. Among men, survey respondents in the weighted sample had 0.86 (95% confidence interval = 0.853, 0.868) times the mortality rate of the general population (among women, RR = 0.887; 95% confidence interval, 0.879, 0.895). Differences in mortality are evident along all points of the life course. Differences have remained relatively stable over time. CONCLUSION: Survey respondents have lower death rates than the general US population, suggesting that they are a systematically healthier source population. Incorporating nonhousehold samples and revised weighting strategies to account for sample frame exclusion and nonresponse may allow for more rigorous estimation of the US population's health.

Making the most of natural experiments: What can studies of the withdrawal of public health interventions offer?

Many interventions that may have large impacts on health and health inequalities, such as social and public health policies and health system reforms, are not amenable to evaluation using randomised controlled trials. The United Kingdom Medical Research Council's guidance on the evaluation of natural experiments draws attention to the need for ingenuity to identify interventions which can be robustly studied as they occur, and without experimental manipulation. Studies of intervention withdrawal may usefully widen the range of interventions that can be evaluated, allowing some interventions and policies, such as those that have developed piecemeal over a long period, to be evaluated for the first time. In particular, sudden removal may allow a more robust assessment of an intervention's long-term impact by minimising 'learning effects'. Interpreting changes that follow withdrawal as evidence of the impact of an intervention assumes that the effect is reversible and this assumption must be carefully justified. Otherwise, withdrawal-based studies suffer similar threats to validity as intervention studies. These threats should be addressed using recognised approaches, including appropriate choice of comparators, detailed understanding of the change processes at work, careful specification of research questions, and the use of falsification tests and other methods for strengthening causal attribution. Evaluating intervention withdrawal provides opportunities to answer important questions about effectiveness of population health interventions, and to study the social determinants of health. Researchers, policymakers and practitioners should be alert to the opportunities provided by the withdrawal of interventions, but also aware of the pitfalls.

Natural Experiments: An Overview of Methods, Approaches, and Contributions to Public Health Intervention Research.

Population health interventions are essential to reduce health inequalities and tackle other public health priorities, but they are not always amenable to experimental manipulation. Natural experiment (NE) approaches are attracting growing interest as a way of providing evidence in such circumstances. One key challenge in evaluating NEs is selective exposure to the intervention. Studies should be based on a clear theoretical understanding of the processes that determine exposure. Even if the observed effects are large and rapidly follow implementation, confidence in attributing these effects to the intervention can be improved by carefully considering alternative explanations. Causal inference can be strengthened by including additional design features alongside the principal method of effect estimation. NE studies often rely on existing (including routinely collected) data. Investment in such data sources and the infrastructure for linking exposure and outcome data is essential if the potential for such studies to inform decision making is to be realized.

Leaving the labour market later in life: how does it impact on mechanisms for health?

OBJECTIVES: Negative associations between non-employment and health among older people are well established and are potentially important for successful ageing. However, opportunities to improve health through re-employment or extending working lives are limited as later-life exits from employment are often unwanted and permanent. We aim to establish a greater understanding of the psychosocial mechanisms underlying non-employment and health associations in older people to identify modifiable pathways through which the negative impact of non-employment can be ameliorated. METHODS: Using multilevel analysis of four waves of repeated panel data from a representative sample of 1551 older men and women reaching state retirement age in the West of Scotland from 1987/1988 to 2000/2004, we explored respondents' strength of agreement with 20 statements relating to their self-defined employment status, covering themes of functioning, social engagement, self-esteem, mental engagement, stress, and control and autonomy. RESULTS: Compared with those in employment, respondents who were retired, unemployed, sick/disabled and home makers were more likely to agree that this resulted in poor social engagement, low self-esteem and, with the possible exception of retirees, reduced mental engagement. Associations were particularly marked among unemployed and sick/disabled respondents who also agreed that their status was a source of worry and prevented them from feeling in control. CONCLUSION: Older people who are not in employment are at higher risk of poor physical and mental health. Interventions targeting psychosocial mechanisms such as social and mental engagement and self-esteem offer potentially valuable opportunities to improve health outcomes and promote successful ageing.

Insomnia symptoms as a cause of type 2 diabetes Incidence: a 20 year cohort study.

BACKGROUND: Insomnia symptoms are associated with type 2 diabetes incidence but are also associated with a range of potential time-varying covariates which may confound and/or mediate associations. We aimed to assess whether cumulative exposure to insomnia symptoms has a causal effect on type 2 diabetes incidence. METHODS: A prospective cohort study in the West of Scotland, following respondents for 20 years from age 36. 996 respondents were free of diabetes at baseline and had valid data from up to four follow-up visits. Type 2 diabetes was assessed at the final visit by self-report, taking diabetic medication, or blood-test (HbA1c ≥ 6.5% or 48 mmol/mol). Effects of cumulative insomnia exposure on type 2 diabetes incidence were estimated with traditional regression and marginal structural models, adjusting for time-dependent confounding (smoking, diet, physical inactivity, obesity, heavy drinking, psychiatric distress) as well as for gender and baseline occupational class. RESULTS: Traditional regression yielded an odds ratio (OR) of 1.34 (95% CI: 1.06-1.70) for type 2 diabetes incidence for each additional survey wave in which insomnia was reported. Marginal structural models adjusted for prior covariates (assuming concurrently measured covariates were potential mediators), reduced this OR to 1.20 (95% CI: 0.98-1.46), and when concurrent covariates were also included (viewing them as potential confounders) this dropped further to 1.08 (95% CI: 0.85-1.37). CONCLUSIONS: The association between cumulative experience of insomnia and type 2 diabetes incidence appeared confounded. Evidence for a residual causal effect depended on assumptions as to whether concurrently measured covariates were confounders or mediators.

Employment status and income as potential mediators of educational inequalities in population mental health.

We assessed whether educational inequalities in mental health may be mediated by employment status and household income. Poor mental health was assessed using General Health Questionnaire 'caseness' in working age adult participants (N = 48 654) of the Health Survey for England (2001-10). Relative indices of inequality by education level were calculated. Substantial inequalities were apparent, with adjustment for employment status and household income markedly reducing their magnitude. Educational inequalities in mental health were attenuated by employment status. Policy responses to economic recession (such as active labour market interventions) might reduce mental health inequalities but longitudinal research is needed to exclude reverse causation.

The role of material, psychosocial and behavioral factors in mediating the association between socioeconomic position and allostatic load (measured by cardiovascular, metabolic and inflammatory markers).

Lower socioeconomic position (SEP), both accumulated across the life course and at different life-stages, has been found to be associated with higher cumulative physiological burden, as measured by allostatic load. This study aimed to identify what factors mediate the association between SEP and allostatic load, as measured through combining cardiovascular, metabolic and inflammatory markers. We explored the role of material, psychological and behavioral factors, accumulated across two periods in time, in mediating the association between SEP and allostatic load. Data are from the West of Scotland Twenty-07 Study, with respondents followed over five waves of data collection from ages 35 to 55 (n=999). Allostatic load was measured by summing nine binary biomarker scores ('1'=in the highest-risk quartile) measured when respondents were 55years old (wave 5). SEP was measured by a person's accumulated social class over two periods All mediators and SEP were measured at baseline in 1987 and 20years later and combined to form accumulated measures of risk. Material mediators included car and home ownership, and having low income. The General Health Questionnaire (GHQ-12) was used as the psychosocial mediator. Behavioral mediators included smoking, alcohol consumption, physical activity and diet. Path analysis using linear regressions adjusting for sex were performed for each of the potential mediators to assess evidence of attenuation in the association between lower SEP and higher allostatic load. Analyses by mediator type revealed that renting one's home (approximately 78% attenuation) and having low income (approx. 62% attenuation) largely attenuated the SEP-allostatic load association. GHQ did not attenuate the association. Smoking had the strongest attenuating effect of all health behaviors (by 33%) with no other health behaviors attenuating the association substantially. Material factors, namely home tenure and income status, and smoking have important roles in explaining socioeconomic disparities in allostatic load, particularly when accumulated over time.

What is the effect of unemployment on all-cause mortality? A cohort study using propensity score matching.

BACKGROUND: There is a strong association between unemployment and mortality, but whether this relationship is causal remains debated. This study utilizes population-level administrative data from Scotland within a propensity score framework to explore whether the association between unemployment and mortality may be causal. METHODS: The study examined a sample of working men and women aged 25-54 in 1991. Subsequent employment status in 2001 was observed (in work or unemployed) and the relative all-cause mortality risk of unemployment between 2001 and 2010 was estimated. To account for potential selection into unemployment of those in poor health, a propensity score matching approach was used. Matching variables were observed prior to unemployment and included health status up to the year of unemployment (hospital admissions and self-reported limiting long-term illness), as well as measures of socioeconomic position. RESULTS: Unemployment was associated with a significant all-cause mortality risk relative to employment for men (hazard ratio [HR] 1.85; 95% confidence interval [CI] 1.33-2.55). This effect was robust to controlling for prior health and sociodemographic characteristics. Effects for women were smaller and statistically insignificant (HR 1.51; 95% CI 0.68-3.37). CONCLUSION: For men, the findings support the notion that the often-observed association between unemployment and mortality may contain a significant causal component; although for women, there is less support for this conclusion. However, female employment status, as recorded in the census, is more complex than for men and may have served to underestimate any mortality effect of unemployment. Future work should examine this issue further.

The role of health behaviours across the life course in the socioeconomic patterning of all-cause mortality: the west of Scotland twenty-07 prospective cohort study.

BACKGROUND: Socioeconomic differentials in mortality are increasing in many industrialised countries. PURPOSE: This study aims to examine the role of behaviours (smoking, alcohol, exercise, and diet) in explaining socioeconomic differentials in mortality and whether this varies over the life course, between cohorts and by gender. METHODS: Analysis of two representative population cohorts of men and women, born in the 1950s and 1930s, were performed. Health behaviours were assessed on five occasions over 20 years. RESULTS: Health behaviours explained a substantial part of the socioeconomic differentials in mortality. Cumulative behaviours and those that were more strongly associated with socioeconomic status had the greatest impact. For example, in the 1950s cohort, the age-sex adjusted hazard ratio comparing respondents with manual versus non-manual occupational status was 1.80 (1.25, 2.58); adjustment for cumulative smoking over 20 years attenuated the association by 49 %, diet by 43 %, drinking by 13 % and inactivity by only 1%. CONCLUSIONS: Health behaviours have an important role in explaining socioeconomic differentials in mortality.

Socioeconomic position across the lifecourse & allostatic load: data from the West of Scotland Twenty-07 cohort study.

BACKGROUND: We examined how socioeconomic position (SEP) across the lifecourse (three critical periods, social mobility and accumulated over time) is associated with allostatic load (a measure of cumulative physiological burden). METHODS: Data are from the West of Scotland Twenty-07 Study, with respondents aged 35 (n = 740), 55 (n = 817) and 75 (n = 483). SEP measures representing childhood, the transition to adulthood and adulthood SEP were used. Allostatic load was produced by summing nine binary biomarker scores (1 = in the highest-risk quartile). Linear regressions were used for each of the lifecourse models; with model fits compared using partial F-tests. RESULTS: For those aged 35 and 55, higher SEP was associated with lower allostatic load (no association in the 75-year-olds). The accumulation model (more time spent with higher SEP) had the best model fit in those aged 35 (b = -0.50, 95%CI = -0.68, -0.32, P = 0.002) and 55 (b = -0.31, 95%CI = -0.49, -0.12, P < 0.001). However, the relative contributions of each life-stage differed, with adulthood SEP less strongly associated with allostatic load. CONCLUSIONS: Long-term, accumulated higher SEP has been shown to be associated with lower allostatic load (less physiological burden). However, the transition to adulthood may represent a particularly sensitive period for SEP to impact on allostatic load.

Changing gender roles and attitudes and their implications for well-being around the new millennium.

PURPOSE: Given evidence that gender role attitudes (GRAs) and actual gender roles impact on well-being, we examine associations between GRAs, three roles (marital status, household chore division, couple employment) and psychological distress in working-age men and women. We investigate time-trends reflecting broader social and economic changes, by focusing on three age groups at two dates. METHODS: We used British Household Panel Survey data from 20- to 64-year-olds in heterosexual couple households in 1991 (N = 5,302) and 2007 (N = 6,621). We examined: levels of traditional GRAs according to gender, age, date, household and employment roles; associations which GRAs and roles had with psychological distress (measured via the GHQ-12); whether psychological distress increased when GRAs conflicted with actual roles; and whether any of these associations differed according to gender, age or date. RESULTS: Gender traditionalism was lower among women, younger people, those participating in 2007 and in 'less traditional' relationships and households. Psychological distress was higher among those with more traditional GRAs and, particularly among men, for those not employed, and there was some evidence of different patterns of association according to age-group. There was limited evidence, among women only, of increased psychological distress when GRAs and actual roles conflicted and/or reductions when GRAs and roles agreed, particularly in respect of household chores and paid employment. CONCLUSIONS: Although some aspects of gender roles and attitudes (traditionalism and paid employment) are associated with well-being, others (marital status and household chores), and attitude-role consistency, may have little impact on the well-being of contemporary UK adults.