Commentary: Weighing up the dead and missing: reflections on inverse-probability weighting and principal stratification to address truncation by death

A comparative effectiveness study of carotid intervention for long-term stroke prevention in patients with severe asymptomatic stenosis from a large integrated health system

OBJECTIVE

The results of current prospective trials comparing the effectiveness of carotid endarterectomy (CEA) vs standard medical therapy for long-term stroke prevention in patients with asymptomatic carotid stenosis (ACS) will not be available for several years. In this study, we compared the observed effectiveness of CEA and standard medical therapy vs standard medical therapy alone to prevent ipsilateral stroke in a contemporary cohort of patients with ACS.

METHODS

This cohort study was conducted in a large integrated health system in adult subjects with 70% to 99% ACS (no neurologic symptom within 6 months) with no prior ipsilateral carotid artery intervention. Causal inference methods were used to emulate a conceptual randomized trial using data from January 1, 2008, through December 31, 2017, for comparing the event-free survival over 96 months between two treatment strategies: (1) CEA within 12 months from cohort entry vs (2) no CEA (standard medical therapy alone). To account for both baseline and time-dependent confounding, inverse probability weighting estimation was used to derive adjusted hazard ratios, and cumulative risk differences were assessed based on two logistic marginal structural models for counterfactual hazards. Propensity scores were data-adaptively estimated using super learning. The primary outcome was ipsilateral anterior ischemic stroke.

RESULTS

The cohort included 3824 eligible patients with ACS (mean age: 73.7 years, 57.9% male, 12.3% active smokers), of whom 1467 underwent CEA in the first year, whereas 2297 never underwent CEA. The median follow-up was 68 months. A total of 1760 participants (46%) died, 445 (12%) were lost to follow-up, and 158 (4%) experienced ipsilateral stroke. The cumulative risk differences for each year of follow-up showed a protective effect of CEA starting in year 2 (risk difference = 1.1%, 95% confidence interval: 0.5%-1.6%) and persisting to year 8 (2.6%, 95% confidence interval: 0.3%-4.8%) compared with patients not receiving CEA.

CONCLUSIONS

In this contemporary cohort study of patients with ACS using rigorous analytic methodology, CEA appears to have a small but statistically significant effect on stroke prevention out to 8 years. Further study is needed to appropriately select the subset of patients most likely to benefit from intervention.

Considering Questions Before Methods in Dementia Research With Competing Events and Causal Goals

Studying causal exposure effects on dementia is challenging when death is a competing event. Researchers often interpret death as a potential source of bias, although bias cannot be defined or assessed if the causal question is not explicitly specified. Here we discuss 2 possible notions of a causal effect on dementia risk: the "controlled direct effect" and the "total effect." We provide definitions and discuss the "censoring" assumptions needed for identification in either case and their link to familiar statistical methods. We illustrate concepts in a hypothetical randomized trial on smoking cessation in late midlife, and emulate such a trial using observational data from the Rotterdam Study, the Netherlands, 1990-2015. We estimated a total effect of smoking cessation (compared with continued smoking) on 20-year dementia risk of 2.1 (95% confidence interval: -0.1, 4.2) percentage points and a controlled direct effect of smoking cessation on 20-year dementia risk had death been prevented of -2.7 (95% confidence interval: -6.1, 0.8) percentage points. Our study highlights how analyses corresponding to different causal questions can have different results, here with point estimates on opposite sides of the null. Having a clear causal question in view of the competing event and transparent and explicit assumptions are essential to interpreting results and potential bias.

Prenatal Exposure to Toxic Metals and Neural Tube Defects: A Systematic Review of the Epidemiologic Evidence

BACKGROUND

Neural tube defects (NTDs) affect > 300,000 pregnancies worldwide annually. Few nongenetic factors, other than folate deficiency, have been identified that may provide intervenable solutions to reduce the burden of NTDs. Prenatal exposure to toxic metals [arsenic (As), cadmium (Cd), mercury (Hg), manganese (Mn) and lead (Pb)] may increase the risk of NTDs. Although a growing epidemiologic literature has examined associations, to our knowledge no systematic review has been conducted to date.

OBJECTIVE

Through adaptation of the Navigation Guide systematic review methodology, we aimed to answer the question "does exposure to As, Cd, Hg, Mn, or Pb during gestation increase the risk of NTDs?" and to assess challenges to evaluating this question given the current evidence.

METHODS

We selected available evidence on prenatal As, Cd, Hg, Mn, or Pb exposure and risk of specific NTDs (e.g., spina bifida, anencephaly) or all NTDs via a comprehensive search across MEDLINE, Embase, Web of Science, and TOXLINE databases and applied inclusion/exclusion criteria. We rated the quality and strength of the evidence for each metal. We applied a customized risk of bias protocol and evaluated the sufficiency of evidence of an effect of each metal on NTDs.

RESULTS

We identified 30 studies that met our criteria. Risk of bias for confounding and selection was high in most studies, but low for missing data. We determined that, although the evidence was limited, the literature supported an association between prenatal exposure to Hg or Mn and increased risk of NTDs. For the remaining metals, the evidence was inadequate to establish or rule out an effect.

CONCLUSION

The role of gestational As, Cd, or Pb exposure in the etiology of NTDs remains unclear and warrants further investigation in high-quality studies, with a particular focus on controlling confounding, mitigating selection bias, and improving exposure assessment. https://doi.org/10.1289/EHP11872.

Long-term Trajectories of Physical Function Decline in Women With and Without Cancer

Importance

Patients with cancer experience acute declines in physical function, hypothesized to reflect accelerated aging driven by cancer-related symptoms and effects of cancer therapies. No study has examined long-term trajectories of physical function by cancer site, stage, or treatment compared with cancer-free controls.

Objective

Examine trajectories of physical function a decade before and after cancer diagnosis among older survivors and cancer-free controls.

Design, Setting, and Participants

This prospective cohort study enrolled patients from 1993 to 1998 and followed up until December 2020. The Women's Health Initiative, a diverse cohort of postmenopausal women, included 9203 incident cancers (5989 breast, 1352 colorectal, 960 endometrial, and 902 lung) matched to up to 5 controls (n = 45 358) on age/year of enrollment and study arm.

Exposures

Cancer diagnosis (site, stage, and treatment) via Medicare and medical records.

Main Outcomes and Measures

Trajectories of self-reported physical function (RAND Short Form 36 [RAND-36] scale; range: 0-100, higher scores indicate superior physical function) estimated from linear mixed effects models with slope changes at diagnosis and 1-year after diagnosis.

Results

This study included 9203 women with cancer and 45 358 matched controls. For the women with cancer, the mean (SD) age at diagnosis was 73.0 (7.6) years. Prediagnosis, physical function declines of survivors with local cancers were similar to controls; after diagnosis, survivors experienced accelerated declines relative to controls, whose scores declined 1 to 2 points per year. Short-term declines in the year following diagnosis were most severe in women with regional disease (eg, -5.3 [95% CI, -6.4 to -4.3] points per year in regional vs -2.8 [95% CI, -3.4 to -2.3] for local breast cancer) or who received systemic therapy (eg, for local endometrial cancer, -7.9 [95% CI, -12.2 to -3.6] points per year with any chemotherapy; -3.1 [95% CI, -6.0 to -0.3] with radiation therapy alone; and -2.6 [95% CI, -4.2 to -1.0] with neither, respectively). While rates of physical function decline slowed in the later postdiagnosis period (eg, women with regional colorectal cancer declined -4.3 [95% CI, -5.9 to -2.6] points per year in the year following diagnosis vs -1.4 [95% CI, -1.7 to -1.0] points per year in the decade thereafter), survivors had estimated physical function significantly below that of age-matched controls 5 years after diagnosis.

Conclusions and Relevance

In this prospective cohort study, survivors of cancer experienced accelerated declines in physical function after diagnosis, and physical function remained below that of age-matched controls even years later. Patients with cancer may benefit from supportive interventions to preserve physical functioning.

Causal Inference in Studying the Long-Term Health Effects of Disasters: Challenges and Potential Solutions

Two frequently encountered but underrecognized challenges for causal inference in studying the long-term health effects of disasters among survivors include 1) time-varying effects of disasters on a time-to-event outcome and 2) selection bias due to selective attrition. In this paper, we review approaches for overcoming these challenges and demonstrate application of the approaches to a real-world longitudinal data set of older adults who were directly affected by the 2011 Great East Japan Earthquake and Tsunami (n = 4,857). To illustrate the problem of time-varying effects of disasters, we examined the association between degree of damage due to the tsunami and all-cause mortality. We compared results from Cox regression analysis assuming proportional hazards with those derived using adjusted parametric survival curves allowing for time-varying hazard ratios. To illustrate the problem of selection bias, we examined the association between proximity to the coast (a proxy for housing damage from the tsunami) and depressive symptoms. We corrected for selection bias due to attrition in the 2 postdisaster follow-up surveys (conducted in 2013 and 2016) using multivariable adjustment, inverse probability of censoring weighting, and survivor average causal effect estimation. Our results demonstrate that analytical approaches which ignore time-varying effects on mortality and selection bias due to selective attrition may underestimate the long-term health effects of disasters.

Using Propensity Scores for Causal Inference: Pitfalls and Tips

Methods based on propensity score (PS) have become increasingly popular as a tool for causal inference. A better understanding of the relative advantages and disadvantages of the alternative analytic approaches can contribute to the optimal choice and use of a specific PS method over other methods. In this article, we provide an accessible overview of causal inference from observational data and two major PS-based methods (matching and inverse probability weighting), focusing on the underlying assumptions and decision-making processes. We then discuss common pitfalls and tips for applying the PS methods to empirical research and compare the conventional multivariable outcome regression and the two alternative PS-based methods (ie, matching and inverse probability weighting) and discuss their similarities and differences. Although we note subtle differences in causal identification assumptions, we highlight that the methods are distinct primarily in terms of the statistical modeling assumptions involved and the target population for which exposure effects are being estimated.

Accounting for selection bias due to death in estimating the effect of wealth shock on cognition for the Health and Retirement Study

The Health and Retirement Study (HRS) is a longitudinal study of U.S. adults enrolled at age 50 and older. We were interested in investigating the effect of a sudden large decline in wealth on the cognitive ability of subjects measured using a dataset provided composite score. However, our analysis was complicated by the lack of randomization, time-dependent confounding, and a substantial fraction of the sample and population will die during follow-up leading to some of our outcomes being censored. The common method to handle this type of problem is marginal structural models (MSM). Although MSM produces valid estimates, this may not be the most appropriate method to reflect a useful real-world situation because MSM upweights subjects who are more likely to die to obtain a hypothetical population that over time, resembles that would have been obtained in the absence of death. A more refined and practical framework, principal stratification (PS), would be to restrict analysis to the strata of the population that would survive regardless of negative wealth shock experience. In this work, we propose a new algorithm for the estimation of the treatment effect under PS by imputing the counterfactual survival status and outcomes. Simulation studies suggest that our algorithm works well in various scenarios. We found no evidence that a negative wealth shock experience would affect the cognitive score of HRS subjects.

The Effect of Prenatal Treatments on Offspring Events in the Presence of Competing Events: An Application to a Randomized Trial of Fertility Therapies

When studying the effect of a prenatal treatment on events in the offspring, failure to produce a live birth is a competing event for events in the offspring. A common approach to handle this competing event is reporting both the treatment-specific probabilities of live births and of the event of interest among live births. However, when the treatment affects the competing event, the latter probability cannot be interpreted as the causal effect among live births. Here we provide guidance for researchers interested in the effects of prenatal treatments on events in the offspring in the presence of the competing event "no live birth." We review the total effect of treatment on a composite event and the total effect of treatment on the event of interest. These causal effects are helpful for decision making but are agnostic about the pathways through which treatment affects the event of interest. Therefore, based on recent work, we also review three causal effects that explicitly consider the pathways through which treatment may affect the event of interest in the presence of competing events: the direct effect of treatment on the event of interest under an intervention to eliminate the competing event, the separable direct and indirect effects of treatment on the event of interest, and the effect of treatment in the principal stratum of those who would have had a live birth irrespective of treatment choice. As an illustrative example, we use a randomized trial of fertility treatments and risk of neonatal complications.

A comparison of methods to estimate the survivor average causal effect in the presence of missing data: a simulation study

BACKGROUND

Attrition due to death and non-attendance are common sources of bias in studies of age-related diseases. A simulation study is presented to compare two methods for estimating the survivor average causal effect (SACE) of a binary exposure (sex-specific dietary iron intake) on a binary outcome (age-related macular degeneration, AMD) in this setting.

METHODS

A dataset of 10,000 participants was simulated 1200 times under each scenario with outcome data missing dependent on measured and unmeasured covariates and survival. Scenarios differed by the magnitude and direction of effect of an unmeasured confounder on both survival and the outcome, and whether participants who died following a protective exposure would also die if they had not received the exposure (validity of the monotonicity assumption). The performance of a marginal structural model (MSM, weighting for exposure, survival and missing data) was compared to a sensitivity approach for estimating the SACE. As an illustrative example, the SACE of iron intake on AMD was estimated using data from 39,918 participants of the Melbourne Collaborative Cohort Study.

RESULTS

The MSM approach tended to underestimate the true magnitude of effect when the unmeasured confounder had opposing directions of effect on survival and the outcome. Overestimation was observed when the unmeasured confounder had the same direction of effect on survival and the outcome. Violation of the monotonicity assumption did not increase bias. The estimates were similar between the MSM approach and the sensitivity approach assessed at the sensitivity parameter of 1 (assuming no survival bias). In the illustrative example, high iron intake was found to be protective of AMD (adjusted OR 0.57, 95% CI 0.40-0.82) using complete case analysis via traditional logistic regression. The adjusted SACE odds ratio did not differ substantially from the complete case estimate, ranging from 0.54 to 0.58 for each of the SACE methods.

CONCLUSIONS

On average, MSMs with weighting for exposure, missing data and survival produced biased estimates of the SACE in the presence of an unmeasured survival-outcome confounder. The direction and magnitude of effect of unmeasured survival-outcome confounders should be considered when assessing exposure-outcome associations in the presence of attrition due to death.

Investigating and Remediating Selection Bias in Geriatrics Research: The Selection Bias Toolkit

OBJECTIVES

Selection bias is a well-known concern in research on older adults. We discuss two common forms of selection bias in aging research: (1) survivor bias and (2) bias due to loss to follow-up. Our objective was to review these two forms of selection bias in geriatrics research. In clinical aging research, selection bias is a particular concern because all participants must have survived to old age, and be healthy enough, to take part in a research study in geriatrics.

DESIGN

We demonstrate the key issues related to selection bias using three case studies focused on obesity, a common clinical risk factor in older adults. We also created a Selection Bias Toolkit that includes strategies to prevent selection bias when designing a research study in older adults and analytic techniques that can be used to examine, and correct for, the influence of selection bias in geriatrics research.

RESULTS

Survivor bias and bias due to loss to follow-up can distort study results in geriatric populations. Key steps to avoid selection bias at the study design stage include creating causal diagrams, minimizing barriers to participation, and measuring variables that predict loss to follow-up. The Selection Bias Toolkit details several analytic strategies available to geriatrics researchers to examine and correct for selection bias (eg, regression modeling and sensitivity analysis).

CONCLUSION

The toolkit is designed to provide a broad overview of methods available to examine and correct for selection bias. It is specifically intended for use in the context of aging research. J Am Geriatr Soc 67:1970-1976, 2019.

Intersectional cohort change: Disparities in mobility limitations among older Singaporeans

Mobility is fundamental to independent living, but past research on physical function and mobility in older adults has not considered both intersectional social identities and cohort change in tandem. This paper utilizes data on mobility limitations from older adults in multi-ethnic Singapore to test whether cohort change varies simultaneously by gender and ethnicity. Panel data (n = 9334 person-years) collected over six years (2009-2015) were used to estimate aging vector models. Findings show that after adjusting for all covariates, Malay and Indian males in later-born cohorts have an increased number of mobility limitations compared to earlier-born cohorts. While a similar trend was also found for Chinese males and females in unconditional models, these were fully mediated by sociodemographic and health variables. These results highlight the importance of considering cohort change at the intersection of gender and ethnicity, bringing attention to possible inequities between ethnic groups.

Targeted learning with daily EHR data

Electronic health records (EHR) data provide a cost- and time-effective opportunity to conduct cohort studies of the effects of multiple time-point interventions in the diverse patient population found in real-world clinical settings. Because the computational cost of analyzing EHR data at daily (or more granular) scale can be quite high, a pragmatic approach has been to partition the follow-up into coarser intervals of pre-specified length (eg, quarterly or monthly intervals). The feasibility and practical impact of analyzing EHR data at a granular scale has not been previously evaluated. We start filling these gaps by leveraging large-scale EHR data from a diabetes study to develop a scalable targeted learning approach that allows analyses with small intervals. We then study the practical effects of selecting different coarsening intervals on inferences by reanalyzing data from the same large-scale pool of patients. Specifically, we map daily EHR data into four analytic datasets using 90-, 30-, 15-, and 5-day intervals. We apply a semiparametric and doubly robust estimation approach, the longitudinal Targeted Minimum Loss-Based Estimation (TMLE), to estimate the causal effects of four dynamic treatment rules with each dataset, and compare the resulting inferences. To overcome the computational challenges presented by the size of these data, we propose a novel TMLE implementation, the "long-format TMLE," and rely on the latest advances in scalable data-adaptive machine-learning software, xgboost and h2o, for estimation of the TMLE nuisance parameters.

An Investigation of Selection Bias in Estimating Racial Disparity in Stroke Risk Factors

Selection due to survival or attrition might bias estimates of racial disparities in health, but few studies quantify the likely magnitude of such bias. In a large national cohort with moderate loss to follow-up, we contrasted racial differences in 2 stroke risk factors, incident hypertension and incident left ventricular hypertrophy, estimated by complete-case analyses, inverse probability of attrition weighting, and the survivor average causal effect. We used data on 12,497 black and 17,660 white participants enrolled in the United States (2003-2007) and collected incident risk factor data approximately 10 years after baseline. At follow-up, 21.0% of white participants and 23.0% of black participants had died; additionally 22.0% of white participants and 28.4% of black participants had withdrawn. Individual probabilities of completing the follow-up visit were estimated using baseline demographic and health characteristics. Adjusted risk ratio estimates of racial disparities from complete-case analyses in both incident hypertension (1.11, 95% confidence interval: 1.02, 1.21) and incident left ventricular hypertrophy (1.02, 95% confidence interval: 0.84, 1.24) were virtually identical to estimates from inverse probability of attrition weighting and survivor average causal effect. Despite racial differences in mortality and attrition, we found little evidence of selection bias in the estimation of racial differences for these incident risk factors.

Survival-related Selection Bias in Studies of Racial Health Disparities: The Importance of the Target Population and Study Design

The impact of survival-related selection bias has not always been discussed in relevant studies of racial health disparities. Moreover, the analytic approaches most frequently employed in the epidemiologic literature to minimize selection bias are difficult to implement appropriately in racial disparities research. This difficulty stems from the fact that frequently employed analytic techniques require that common causes of survival and the outcome are accurately measured. Unfortunately, such common causes are often unmeasured or poorly measured in racial health disparities studies. In the absence of accurate measures of the aforementioned common causes, redefining the target population or changing the study design represents a useful approach for reducing the extent of survival-related selection bias. To help researchers recognize and minimize survival-related selection bias in racial health disparities studies, we illustrate the aforementioned selection bias and how redefining the target population or changing the study design can be useful.

Left Truncation Bias to Explain the Protective Effect of Smoking on Preeclampsia: Potential, But How Plausible?

BACKGROUND

An inverse association between maternal smoking and preeclampsia has been frequently observed in epidemiologic studies for several decades. In the May 2015 issue of this journal, Lisonkova and Joseph described a simulation study suggesting that bias from left truncation might explain the inverse association. The simulations were based on strong assumptions regarding the underlying mechanisms through which bias might occur.

METHODS

To examine the sensitivity of the previous authors' conclusions to these assumptions, we constructed a new Monte Carlo simulation using published estimates to frame our data-generating parameters. We estimated the association between smoking and preeclampsia across a range of scenarios that incorporated abnormal placentation and early pregnancy loss.

RESULTS

Our results confirmed that the previous authors' findings are highly dependent on assumptions regarding the strength of association between abnormal placentation and preeclampsia. Thus, the bias they described may be less pronounced than was suggested.

CONCLUSIONS

Under empirically derived constraints of these critical assumptions, left truncation does not appear to fully explain the inverse association between smoking and preeclampsia. Furthermore, when considering processes in which left truncation may result from the exposure, it is important to precisely describe the target population and parameter of interest before assessing potential bias. We comment on the specification of a meaningful target population when assessing maternal smoking and preeclampsia as a public health issue. We describe considerations for defining a target population in studies of perinatal exposures when those exposures cause competing events (e.g., early pregnancy loss) for primary outcomes of interest.

Multiple imputation of cognitive performance as a repeatedly measured outcome

Longitudinal studies of cognitive performance are sensitive to dropout, as participants experiencing cognitive deficits are less likely to attend study visits, which may bias estimated associations between exposures of interest and cognitive decline. Multiple imputation is a powerful tool for handling missing data, however its use for missing cognitive outcome measures in longitudinal analyses remains limited. We use multiple imputation by chained equations (MICE) to impute cognitive performance scores of participants who did not attend the 2011-2013 exam of the Atherosclerosis Risk in Communities Study. We examined the validity of imputed scores using observed and simulated data under varying assumptions. We examined differences in the estimated association between diabetes at baseline and 20-year cognitive decline with and without imputed values. Lastly, we discuss how different analytic methods (mixed models and models fit using generalized estimate equations) and choice of for whom to impute result in different estimands. Validation using observed data showed MICE produced unbiased imputations. Simulations showed a substantial reduction in the bias of the 20-year association between diabetes and cognitive decline comparing MICE (3-4 % bias) to analyses of available data only (16-23 % bias) in a construct where missingness was strongly informative but realistic. Associations between diabetes and 20-year cognitive decline were substantially stronger with MICE than in available-case analyses. Our study suggests when informative data are available for non-examined participants, MICE can be an effective tool for imputing cognitive performance and improving assessment of cognitive decline, though careful thought should be given to target imputation population and analytic model chosen, as they may yield different estimands.

A Simulation Platform for Quantifying Survival Bias: An Application to Research on Determinants of Cognitive Decline

Bias due to selective mortality is a potential concern in many studies and is especially relevant in cognitive aging research because cognitive impairment strongly predicts subsequent mortality. Biased estimation of the effect of an exposure on rate of cognitive decline can occur when mortality is a common effect of exposure and an unmeasured determinant of cognitive decline and in similar settings. This potential is often represented as collider-stratification bias in directed acyclic graphs, but it is difficult to anticipate the magnitude of bias. In this paper, we present a flexible simulation platform with which to quantify the expected bias in longitudinal studies of determinants of cognitive decline. We evaluated potential survival bias in naive analyses under several selective survival scenarios, assuming that exposure had no effect on cognitive decline for anyone in the population. Compared with the situation with no collider bias, the magnitude of bias was higher when exposure and an unmeasured determinant of cognitive decline interacted on the hazard ratio scale to influence mortality or when both exposure and rate of cognitive decline influenced mortality. Bias was, as expected, larger in high-mortality situations. This simulation platform provides a flexible tool for evaluating biases in studies with high mortality, as is common in cognitive aging research.

Lung function decline over 25 years of follow-up among black and white adults in the ARIC study cohort

BACKGROUND

Interpretation of longitudinal information about lung function decline from middle to older age has been limited by loss to follow-up that may be correlated with baseline lung function or the rate of decline. We conducted these analyses to estimate age-related decline in lung function across groups of race, sex, and smoking status while accounting for dropout from the Atherosclerosis Risk in Communities Study.

METHODS

We analyzed data from 13,896 black and white participants, aged 45-64 years at the 1987-1989 baseline clinical examination. Using spirometry data collected at baseline and two follow-up visits, we estimated annual population-averaged mean changes in forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) by race, sex, and smoking status using inverse-probability-weighted independence estimating equations conditioning-on-being-alive.

RESULTS

Estimated rates of FEV1 decline estimated using inverse-probability-weighted independence estimating equations conditioning on being alive were higher among white than black participants at age 45 years (e.g., male never smokers: black: -29.5 ml/year; white: -51.9 ml/year), but higher among black than white participants by age 75 (black: -51.2 ml/year; white: -26). Observed differences by race were more pronounced among men than among women. By smoking status, FEV1 declines were larger among current than former or never smokers at age 45 across all categories of race and sex. By age 60, FEV1 decline was larger among former and never than current smokers. Estimated annual declines generated using unweighted generalized estimating equations were smaller for current smokers at younger ages in all four groups of race and sex compared with results from weighted analyses that accounted for attrition.

CONCLUSIONS

Using methods accounting for dropout from an approximately 25-year health study, estimated rates of lung function decline varied by age, race, sex, and smoking status, with largest declines observed among current smokers at younger ages.

Biased Exposure-Health Effect Estimates from Selection in Cohort Studies: Are Environmental Studies at Particular Risk?

BACKGROUND

The process of creating a cohort or cohort substudy may induce misleading exposure-health effect associations through collider stratification bias (i.e., selection bias) or bias due to conditioning on an intermediate. Studies of environmental risk factors may be at particular risk.

OBJECTIVES

We aimed to demonstrate how such biases of the exposure-health effect association arise and how one may mitigate them.

METHODS

We used directed acyclic graphs and the example of bone lead and mortality (all-cause, cardiovascular, and ischemic heart disease) among 835 white men in the Normative Aging Study (NAS) to illustrate potential bias related to recruitment into the NAS and the bone lead substudy. We then applied methods (adjustment, restriction, and inverse probability of attrition weighting) to mitigate these biases in analyses using Cox proportional hazards models to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs).

RESULTS

Analyses adjusted for age at bone lead measurement, smoking, and education among all men found HRs (95% CI) for the highest versus lowest tertile of patella lead of 1.34 (0.90, 2.00), 1.46 (0.86, 2.48), and 2.01 (0.86, 4.68) for all-cause, cardiovascular, and ischemic heart disease mortality, respectively. After applying methods to mitigate the biases, the HR (95% CI) among the 637 men analyzed were 1.86 (1.12, 3.09), 2.47 (1.23, 4.96), and 5.20 (1.61, 16.8), respectively.

CONCLUSIONS

Careful attention to the underlying structure of the observed data is critical to identifying potential biases and methods to mitigate them. Understanding factors that influence initial study participation and study loss to follow-up is critical. Recruitment of population-based samples and enrolling participants at a younger age, before the potential onset of exposure-related health effects, can help reduce these potential pitfalls.

CITATION

Weisskopf MG, Sparrow D, Hu H, Power MC. 2015. Biased exposure-health effect estimates from selection in cohort studies: are environmental studies at particular risk? Environ Health Perspect 123:1113-1122; http://dx.doi.org/10.1289/ehp.1408888.

Guidelines for reporting methodological challenges and evaluating potential bias in dementia research

Clinical and population research on dementia and related neurologic conditions, including Alzheimer's disease, faces several unique methodological challenges. Progress to identify preventive and therapeutic strategies rests on valid and rigorous analytic approaches, but the research literature reflects little consensus on "best practices." We present findings from a large scientific working group on research methods for clinical and population studies of dementia, which identified five categories of methodological challenges as follows: (1) attrition/sample selection, including selective survival; (2) measurement, including uncertainty in diagnostic criteria, measurement error in neuropsychological assessments, and practice or retest effects; (3) specification of longitudinal models when participants are followed for months, years, or even decades; (4) time-varying measurements; and (5) high-dimensional data. We explain why each challenge is important in dementia research and how it could compromise the translation of research findings into effective prevention or care strategies. We advance a checklist of potential sources of bias that should be routinely addressed when reporting dementia research.

Are all biases missing data problems?

Estimating causal effects is a frequent goal of epidemiologic studies. Traditionally, there have been three established systematic threats to consistent estimation of causal effects. These three threats are bias due to confounders, selection, and measurement error. Confounding, selection, and measurement bias have typically been characterized as distinct types of biases. However, each of these biases can also be characterized as missing data problems that can be addressed with missing data solutions. Here we describe how the aforementioned systematic threats arise from missing data as well as review methods and their related assumptions for reducing each bias type. We also link the assumptions made by the reviewed methods to the missing completely at random (MCAR) and missing at random (MAR) assumptions made in the missing data framework that allow for valid inferences to be made based on the observed, incomplete data.

Confirmatory Factor Analysis Compared with Principal Component Analysis to Derive Dietary Patterns: A Longitudinal Study in Adult Women

BACKGROUND

Principal component analysis (PCA) has been used extensively to derive dietary patterns. We proposed to use confirmatory factor analysis (CFA) in the same context as PCA--as a one-step approach--to derive dietary patterns.

OBJECTIVES

The first aim of this study was methodologic and was to compare dietary patterns derived with the use of PCA and CFA, used as equivalent one-step approaches. The second aim of this study was to study these patterns in association with individual characteristics and new adult-onset asthma.

METHODS

We included 30,589 French women from the E3N (epidemiologic prospective cohort study of women from the MGEN national insurance plan) with 1177 reported cases of adult-onset asthma between 1993 and 2005. PCA and CFA were used in the same context, on 27 food groups, to derive dietary patterns. Associations between dietary patterns and adult-onset asthma were assessed by Cox proportional hazards models.

RESULTS

Whether we used PCA or CFA, 3 similar factors were found and labeled "Prudent," "Western," and "Aperitif." Correlations between patterns derived with the use of PCA and CFA were high. For the "Prudent" and "Aperitif" patterns, we observed comparable patterns in terms of associations with food groups, individual characteristics, and the onset of asthma. For the "Western" patterns, the one derived with the use of CFA was more related to an unhealthy diet than the one derived with the use of PCA, with higher correlations with the food groups "processed meat" (0.73 vs. 0.51) and "dough and pastry" (0.63 vs. 0.40), and negative associations with physical activity and with having parents who were farmers. Regarding associations with adult-onset asthma, a significant positive association was observed for the "Western" pattern derived with the use of CFA [multivariate RR for highest vs. lowest quintile: 1.30 (1.02, 1.67), P-trend: 0.03], whereas no association was reported when using PCA [RR: 1.14 (0.89, 1.47), P-trend: 0.40].

CONCLUSION

Although quite similar dietary patterns were derived with the use of PCA and CFA, this study supports the alternative use of CFA to PCA for the identification of dietary patterns in epidemiologic studies.

Analytical results in longitudinal studies depended on target of inference and assumed mechanism of attrition

OBJECTIVES

To compare methods for analysis of longitudinal studies with missing data due to participant dropout and follow-up truncated by death.

STUDY DESIGN AND SETTING

We analyzed physical functioning in an Australian longitudinal study of elderly women where the missing data mechanism could either be missing at random (MAR) or missing not at random (MNAR). We assumed either an immortal cohort where deceased participants are implicitly included after death or a mortal cohort where the target of inference is surviving participants at each survey wave. To illustrate the methods a covariate was included. Simulation was used to assess the effect of the assumptions.

RESULTS

Ignoring attrition or restricting analysis to participants with complete follow up led to biased estimates. Linear mixed model was appropriate for an immortal cohort under MAR but not MNAR. Linear increment model and joint modeling of longitudinal outcome and time to death were the most robust to MNAR. For a mortal cohort, inverse probability weighting and multiple imputation could be used, but care is needed in specifying dropout and imputation models, respectively.

CONCLUSION

Appropriate analysis methodology to deal with attrition in longitudinal studies depends on the target of inference and the missing data mechanism.

Smoking and white matter hyperintensity progression: the ARIC-MRI Study

OBJECTIVE

Our objective was to examine the link between smoking and smoking history, including smoking intensity and cessation, overall and by race, in a biracial prospective cohort study.

METHODS

A subset of Atherosclerosis Risk in Communities Study participants (n = 972, 49% black) completed brain MRI scans twice (1993-1995 and 2004-2006). We defined white matter hyperintensity (WMH) progression as an increase of ≥2 points on the 9-point Cardiovascular Health Study scale across scans. Participants reported information on smoking behavior at the baseline MRI and at 2 prior study visits, approximately 3 and 6 years before baseline. We used adjusted logistic regression to evaluate the association between smoking variables and WMH progression in the total sample and separately by race (black and white).

RESULTS

We found WMH progression in 23% of participants (30% of black participants, 17% of white participants). Overall, being a current smoker 6 years before baseline was associated with WMH progression. In race-stratified analyses, we found adverse associations with smoking status at multiple time points and persistent smoking in white but not in black participants. However, we found no statistical support for effect modification by race for most of these analyses. Increasing pack-years of smoking was associated with greater risk of WMH progression, while time since quitting and age at smoking initiation were not associated with WMH progression, with little indication of differences in these associations by race.

CONCLUSIONS

Our findings concur with previous studies suggesting a relationship between smoking and WMH progression, and further demonstrate a dose-dependent association.

Work economic sectors and cardiovascular risk factors: cross-sectional analysis based on the RECORD Study

Background

Little is known on the comparative effect of work economic sectors on multiple cardiovascular risk factors. Such information may be useful to target Public health interventions, e.g., through the occupational medicine. We investigated whether and how a large panel of cardiovascular risk factors varied between 11 work economic sectors.

Methods

Data on 4360 participants from the French RECORD Study geolocated at their residence were analyzed. Ten outcomes were assessed: body mass index (BMI), waist circumference, systolic and diastolic blood pressure (BP), pulse pressure, total cholesterol, glycaemia, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and resting heart rate. Multilevel linear regression models stratified by sex and adjusted for individual and neighborhood sociodemographic characteristics were estimated.

Results

Among men, the Health and social work sector was found to be the most protective sector for BMI, waist circumference, and glycaemia (while the Construction sector and the Transport and communications sector tended to be unfavorable for these outcomes). The Health and social work sector was also associated with higher HDL cholesterol among men. However, men working in the Health and social work sector showed the highest systolic BP and pulse pressure. Women working in the Health and social work sector had the highest BMI, the largest waist circumference, and the most elevated systolic and diastolic BP. The Commercial and repair of vehicles sector, the Transport and communication sector, and the Collective, social, and personal services sector were associated with a more favorable profile for these risk factors among women.

Conclusion

Work economic sectors contribute to shape metabolic and cardiovascular parameters after adjustment for individual/neighborhood sociodemographic characteristics. However, patterns of associations varied strikingly according to the risk factor examined and between men and women. Such findings may be useful to target interventions for reducing cardiovascular risk, e.g., through the occupational medicine.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2458-14-750) contains supplementary material, which is available to authorized users.

[The first wave of the German Health Interview and Examination Survey for Adults (DEGS1): sample design, response, weighting and representativeness]

The "German Health Interview and Examination Survey for Adults" (DEGS) is part of the health monitoring program of the Robert Koch Institute (RKI) and is designed as a combined cross-sectional and longitudinal survey. The first wave (DEGS1; 2008-2011) comprised interviews and physical examinations. The target population were 18- to 79-year olds living in Germany. The mixed design consisted of a new sample randomly chosen from local population registries which was supplemented by participants from the "German National Health Interview and Examination Survey 1998" (GNHIES98). In total, 8,152 persons took part, among them 4,193 newly invited (response 42%) and 3,959 who had previously taken part in GNHIES98 (response 62%). 7,238 participants visited one of the 180 local study centres, 914 took part in the interview-only programme. The comparison of the net sample with the group of non-participants and with the resident population of Germany suggests a high representativeness regarding various attributes. To account for certain aspects of the population structure cross-sectional, trend and longitudinal analyses are corrected by weighting factors. Furthermore, different participation probabilities of the former participants of GNHIES98 are compensated for. An English full-text version of this article is available at SpringerLink as supplemental.