Invited commentary: structural equation models and epidemiologic analysis

Evaluation of direct and indirect effects of seasonal malaria chemoprevention in Mali

Randomized controlled trials have established that seasonal malaria chemoprevention (SMC) in children is a promising strategy to reduce malaria transmission in Sahelian West Africa. This strategy was recently introduced in a dozen countries, and about 12 million children received SMC in 2016. However, evidence on SMC effectiveness under routine programme conditions is sparse. We aim to measure the effects of the nationwide SMC programme in Mali on the prevalence of malaria and anemia in children 6-59 months. We used data from the 2015 nationally representative malaria indicator survey. A post-test only with non-randomized control group study was designed. We fitted a generalized structural equation model that controlled for potential bias on observed and non-observed variables (endogenous treatment effect model). Having received SMC reduced by 44% (95% CI [0.39-0.49]) the risk of having a positive rapid diagnostic test for malaria. In addition, the programme indirectly reduced by 18% the risk of moderate-to-severe anemia (95% CI [0.15-0.21]). SMC in Mali has substantial protective effects under routine nationwide programme conditions. Endogenous treatment effects analyses can contribute to rigorously measuring the effectiveness of health programmes and to bridging a widening gap in evaluation methods to measure progress towards achieving malaria elimination.

Evaluating effects of prenatal exposure to phthalate mixtures on birth weight: A comparison of three statistical approaches

OBJECTIVES

We applied three statistical approaches for evaluating associations between prenatal urinary concentrations of a mixture of phthalate metabolites and birth weight.

METHODS

We included 300 women who provided 732 urine samples during pregnancy and delivered a singleton infant. We measured urinary concentrations of metabolites of di(2-ethylhexyl)-phthalate, di-isobutyl-, di-n-butyl-, butylbenzyl-, and diethyl phthalates. We applied 1) linear regressions; 2) classification methods [principal component analysis (PCA) and structural equation models (SEM)]; and 3) Bayesian kernel machine regression (BKMR), to evaluate associations between phthalate metabolite mixtures and birth weight adjusting for potential confounders. Data were presented as mean differences (95% CI) in birth weight (grams) as each phthalate increased from the 10th to the 90th percentile.

RESULTS

When analyzing individual phthalate metabolites using linear regressions, each metabolite demonstrated a modest inverse association with birth weight [from -93 (-206, 21) to -49 (-164, 65)]. When simultaneously including all metabolites in a multivariable model, inflation of the estimates and standard errors were noted. PCA identified two principal components, both inversely associated with birth weight [-23 (-68, 22), -27 (-71, 17), respectively]. These inverse associations were confirmed when applying SEM. BKMR further identified that monoethyl and mono(2-ethylhexyl) phthalate and phthalate concentrations were linearly related to lower birth weight [-51(-164, 63) and -122 (-311, 67), respectively], and suggested no evidence of interaction between metabolites.

CONCLUSIONS

While none of the methods produced significant results, we demonstrated the potential issues arising using linear regression models in the context of correlated exposures. Among the other selected approaches, classification techniques identified common sources of exposures with implications for interventions, while BKMR further identified specific contributions of individual metabolites.

A structural equation modeling approach to understanding pathways that connect socioeconomic status and smoking

The inverse association between socioeconomic status and smoking is well established, yet the mechanisms that drive this relationship are unclear. We developed and tested four theoretical models of the pathways that link socioeconomic status to current smoking prevalence using a structural equation modeling (SEM) approach. Using data from the 2013 National Health Interview Survey, we selected four indicator variables (poverty ratio, personal earnings, educational attainment, and employment status) that we hypothesize underlie a latent variable, socioeconomic status. We measured direct, indirect, and total effects of socioeconomic status on smoking on four pathways through four latent variables representing social cohesion, financial strain, sleep disturbance, and psychological distress. Results of the model indicated that the probability of being a smoker decreased by 26% of a standard deviation for every one standard deviation increase in socioeconomic status. The direct effects of socioeconomic status on smoking accounted for the majority of the total effects, but the overall model also included significant indirect effects. Of the four mediators, sleep disturbance and psychological distress had the largest total effects on current smoking. We explored the use of structural equation modeling in epidemiology to quantify effects of socioeconomic status on smoking through four social and psychological factors to identify potential targets for interventions. A better understanding of the complex relationship between socioeconomic status and smoking is critical as we continue to reduce the burden of tobacco and eliminate health disparities related to smoking.

Dynamic Effects of Depressive Symptoms on Osteoarthritis Knee Pain

OBJECTIVE

To estimate the dynamic causal effects of depressive symptoms on osteoarthritis (OA) knee pain.

METHODS

Marginal structural models were used to examine dynamic associations between depressive symptoms and pain over 48 months among older adults (n = 2,287) with radiographic knee OA (Kellgren/Lawrence grade 2 or 3) in the Osteoarthritis Initiative. Depressive symptoms at each annual visit were assessed (threshold ≥16) using the Center for Epidemiologic Studies Depression Scale. OA knee pain was measured using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain subscale, rescaled to range from 0 to 100.

RESULTS

Depressive symptoms at each visit were generally not associated with greater OA knee pain at subsequent time points. Causal mean differences in WOMAC pain score comparing depressed to nondepressed patients ranged from 1.78 (95% confidence interval [95% CI] -0.73, 4.30) to 2.58 (95% CI 0.23, 4.93) within the first and fourth years, and the depressive symptoms by time interaction were not statistically significant (P = 0.94). However, there was a statistically significant dose-response relationship between the persistence of depressive symptoms and OA knee pain severity (P = 0.002). Causal mean differences in WOMAC pain score comparing depressed to nondepressed patients were 0.89 (95% CI -0.17, 1.96) for 1 visit with depressive symptoms, 2.35 (95% CI 0.64, 4.06) for 2 visits with depressive symptoms, and 3.57 (95% CI 0.43, 6.71) for 3 visits with depressive symptoms.

CONCLUSION

The causal effect of depressive symptoms on OA knee pain does not change over time, but pain severity significantly increases with the persistence of depressed mood.

Childhood Maltreatment and BMI Trajectory: The Mediating Role of Depression

INTRODUCTION

Childhood maltreatment is associated with later obesity, but the underlying mechanisms are unknown. The objective of this study was to estimate the extent to which depression mediates the associations between childhood maltreatment and BMI in adolescence through adulthood.

METHODS

Data on a cohort of 13,362 adolescents in the National Longitudinal Study of Adolescent to Adult Health (Wave I [1994-1995] to Wave IV [2008-2009]) were analyzed in 2015-2016. Classes of maltreatment experienced prior to age 12 years were statistically identified using latent class analysis. Gender-stratified latent growth curve analysis was used to estimate total effects of maltreatment classes on latent BMI trajectory (aged 13-31 years) and indirect effects of maltreatment classes that occurred through latent depression trajectory (aged 12-31 years).

RESULTS

Four latent maltreatment classes were identified: high abuse and neglect; physical abuse dominant; supervisory neglect dominant; and no/low maltreatment. In girls, compared with no/low maltreatment, supervisory neglect dominant (coefficient=0.3, 95% CI=0.0, 0.7) and physical abuse dominant (coefficient=0.6, 95% CI=0.1, 1.2) maltreatment were associated with faster gain in BMI. Change in depression over time fully mediated the association of BMI slope with physical abuse dominant maltreatment, but not with supervisory neglect dominant maltreatment. In boys, high abuse and neglect maltreatment was associated with marginally greater BMI at baseline (coefficient=0.7, 95% CI= -0.1, 1.5); this association was not mediated by depression.

CONCLUSIONS

Although maltreatment was associated with depression and BMI trajectories from adolescence to adulthood, depression only mediated associations with physical abuse dominant maltreatment in girls.

Circulating Cholesterol Levels May Link to the Factors Influencing Parkinson's Risk

OBJECTIVES

A growing literature suggests that circulating cholesterol levels have been associated with Parkinson's disease (PD). In this study, we investigated a possible causal basis for the cholesterol-PD link.

METHODS

Fasting plasma cholesterol levels were obtained from 91 PD and 70 age- and gender-matched controls from an NINDS PD Biomarkers Program cohort at the Pennsylvania State University College of Medicine. Based on the literature, genetic polymorphisms in selected cholesterol management genes (APOE, LDLR, LRP1, and LRPAP1) were chosen as confounding variables because they may influence both cholesterol levels and PD risk. First, the marginal structure model was applied, where the associations of total- and LDL-cholesterol levels with genetic polymorphisms, statin usage, and smoking history were estimated using linear regression. Then, potential causal influences of total- and LDL-cholesterol on PD occurrence were investigated using a generalized propensity score approach in the second step.

RESULTS

Both statins (p < 0.001) and LRP1 (p < 0.03) influenced total- and LDL-cholesterol levels. There also was a trend for APOE to affect total- and LDL-cholesterol (p = 0.08 for both), and for LRPAR1 to affect LDL-cholesterol (p = 0.05). Conversely, LDLR did not influence plasma cholesterol levels (p > 0.19). Based on propensity score methods, lower total- and LDL-cholesterol were significantly linked to PD (p < 0.001 and p = 0.04, respectively).

CONCLUSION

The current study suggests that circulating total- and LDL-cholesterol levels potentially may be linked to the factor(s) influencing PD risk. Further studies to validate these results would impact our understanding of the role of cholesterol as a risk factor in PD, and its relationship to recent public health controversies.

Pathways from maternal depression to young adult offspring depression: an exploratory longitudinal mediation analysis

Maternal depression in the peri-natal period is associated with increased risk for young adult depression in offspring. This study explored mediation of these links via trajectories of child conduct and emotional problems (Strengths and Difficulties Questionnaire) from ages 4-16 years old in data from the Avon Longitudinal Study of Parents and Children cohort (n = 13373). Through gender-specific structural equation models, a composite measure of exposure to early maternal depression (Edinburgh Postnatal Depression Scale), predicted young adult depression at age 18 (Revised Clinical Interview Schedule - distal outcome). Mediational effects were then estimated by testing which parts of joint piecewise latent trajectory models for child/adolescent conduct and emotional problems were associated with both exposure and distal outcome. For girls, only conduct problems in early childhood were consistently indicated to mediate effects of early maternal depression on risk of young adulthood depression. Some evidence for a pathway via changing levels of childhood and adolescent emotional difficulties was also suggested. For boys, by contrast, the differing models gave less consistent findings providing some evidence for a small time-specific indirect effect via early childhood conduct problems. In addition to its practice implications the current methodological application offers considerable potential in exploratory longitudinal developmental mediation studies. © 2016 The Authors International Journal of Methods in Psychiatric Research Published by John Wiley & Sons Ltd.

Association between polygenic risk scores for attention-deficit hyperactivity disorder and educational and cognitive outcomes in the general population

Background

Children with a diagnosis of attention-deficit hyperactivity disorder ADHD) have lower cognitive ability and are at risk of adverse educational outcomes; ADHD genetic risks have been found to predict childhood cognitive ability and other neurodevelopmental traits in the general population; thus genetic risks might plausibly also contribute to cognitive ability later in development and to educational underachievement.

Methods

We generated ADHD polygenic risk scores in the Avon Longitudinal Study of Parents and Children participants (maximum N : 6928 children and 7280 mothers) based on the results of a discovery clinical sample, a genome-wide association study of 727 cases with ADHD diagnosis and 5081 controls. We tested if ADHD polygenic risk scores were associated with educational outcomes and IQ in adolescents and their mothers.

Results

High ADHD polygenic scores in adolescents were associated with worse educational outcomes at Key Stage 3 [national tests conducted at age 13-14 years; β = -1.4 (-2.0 to -0.8), P  = 2.3 × 10 -6 ), at General Certificate of Secondary Education exams at age 15-16 years (β = -4.0 (-6.1 to -1.9), P  = 1.8 × 10 -4 ], reduced odds of sitting Key Stage 5 examinations at age 16-18 years [odds ratio (OR) = 0.90 (0.88 to 0.97), P  = 0.001] and lower IQ scores at age 15.5 [β = -0.8 (-1.2 to -0.4), P  = 2.4 × 10 -4 ]. Moreover, maternal ADHD polygenic scores were associated with lower maternal educational achievement [β = -0.09 (-0.10 to -0.06), P  = 0.005] and lower maternal IQ [β = -0.6 (-1.2 to -0.1), P  = 0.03].

Conclusions

ADHD diagnosis risk alleles impact on functional outcomes in two generations (mother and child) and likely have intergenerational environmental effects.

Examining the relationship between socio-economic status, WASH practices and wasting

Childhood wasting is a global problem and is significantly more pronounced in low and middle income countries like Bangladesh. Socio Economic Status (SES) and Water, Sanitation and Hygiene (WASH) practices may be significantly associated with wasting. Most previous research is consistent about the role of SES, but the significance of WASH in the context of wasting remains ambiguous. The effect of SES and WASH on weight for length (WHZ) is examined using a Structural Equation Model (SEM) to explicitly describe the direct and indirect role of WASH in the context of SES.A nationally representative survey of 10,478 Bangladeshi children under 5 were examined. An expert defined SEM was used to construct latent variables for SES and WASH. The SEM included a direct pathway from SES to WHZ and an indirect pathway from SES to WHZ via WASH along with regression of relevant covariates on the outcome WHZ and the latent variables. Both SES (p<0.01) and WASH (p<0.05) significantly affect WHZ. SES (p<0.01) also significantly affects WASH. Other structural components showed that child's age (p<0.01) affects WHZ and types of residence (p<0.01) affects SES. WASH practices at least partially mediate the association between SES and wasting status. WASH and SES are both significantly associated with WHZ.

Improved parental dietary quality is associated with children's dietary intake through the home environment

BACKGROUND

Improving access to supermarkets has been shown to improve some dietary outcomes, yet there is little evidence for such effects on children. Relatedly, there is a dearth of research assessing the impact of a structural change (i.e. supermarket in a former food desert) on the home environment and its relationship with children's diet.

OBJECTIVE

Assess the relative impact of the home environment on children's diet after the introduction of a new supermarket in a food desert.

METHODS

Among a randomly selected cohort of households living in a food desert, parental diet was assessed before and after the opening of a full-service supermarket. The home environment and children's intake of fruits and vegetables was measured at one point - after the store's opening. Structural equation models were used to estimate the pathways between changes in parental dietary quality at follow-up and children's dietary intake through the home environment.

RESULTS

Parental dietary improvement after the supermarket opened was associated with having a better home environment (β = 0.45, p = 0.001) and with healthier children's dietary intake (β = 0.46, p < 0.001) through higher family nutrition and physical activity scores (β = 0.25, p = 0.02).

CONCLUSIONS

Policy solutions designed to improve diet among low-resource communities should take into account the importance of the home environment.

Causal inference and longitudinal data: a case study of religion and mental health

PURPOSE

We provide an introduction to causal inference with longitudinal data and discuss the complexities of analysis and interpretation when exposures can vary over time.

METHODS

We consider what types of causal questions can be addressed with the standard regression-based analyses and what types of covariate control and control for the prior values of outcome and exposure must be made to reason about causal effects. We also consider newer classes of causal models, including marginal structural models, that can assess questions of the joint effects of time-varying exposures and can take into account feedback between the exposure and outcome over time. Such feedback renders cross-sectional data ineffective for drawing inferences about causation.

RESULTS

The challenges are illustrated by analyses concerning potential effects of religious service attendance on depression, in which there may in fact be effects in both directions with service attendance preventing the subsequent depression, but depression itself leading to lower levels of the subsequent religious service attendance.

CONCLUSIONS

Longitudinal designs, with careful control for prior exposures, outcomes, and confounders, and suitable methodology, will strengthen research on mental health, religion and health, and in the biomedical and social sciences generally.

The indirect association of job strain with long-term sickness absence through bullying: a mediation analysis using structural equation modeling

BACKGROUND

In this longitudinal study the complex interplay between both job strain and bullying in relation to sickness absence was investigated. Following the "work environment hypothesis", which establishes several work characteristics as antecedents of bullying, we assumed that job strain, conceptualized by the Job-Demand-Control model, has an indirect relation with long-term sickness absence through bullying.

METHODS

The sample consisted of 2983 Belgian workers, aged 30 to 55 years, who participated in the Belstress III study. They completed a survey, including the Job Content Questionnaire and a bullying inventory, at baseline. Their sickness absence figures were registered during 1 year follow-up. Long-term sickness absence was defined as at least 15 consecutive days. A mediation analysis, using structural equation modeling, was performed to examine the indirect association of job strain through bullying with long-term sickness absence. The full structural model was adjusted for several possible confounders: age, gender, occupational group, educational level, company, smoking habits, alcohol use, body mass index, self-rated health, baseline long-term sickness absence and neuroticism.

RESULTS

The results support the hypothesis: a significant indirect association of job strain with long-term sickness absence through bullying was observed, suggesting that bullying is an intermediate variable between job strain and long-term sickness absence. No evidence for the reversed pathway of an indirect association of bullying through job strain was found.

CONCLUSIONS

Bullying was observed as a mediating variable in the relation between job strain and sickness absence. The results suggest that exposure to job strain may create circumstances in which a worker risks to become a target of bullying. Our findings are generally in line with the work environment hypothesis, which emphasizes the importance of organizational work factors in the origin of bullying. This study highlights that remodeling jobs to reduce job strain may be important in the prevention of bullying and subsequent sickness absence.

The Modified Reasons for Smoking Scale: factorial structure, validity and reliability in pregnant smokers

RATIONALE, AIMS AND OBJECTIVES

Smoking during pregnancy can cause several maternal and neonatal health risks, yet a considerable number of pregnant women continue to smoke. The objectives of this study were to test the factorial structure, validity and reliability of the Dutch version of the Modified Reasons for Smoking Scale (MRSS) in a sample of smoking pregnant women and to understand reasons for continued smoking during pregnancy.

METHODS

A longitudinal design was performed. Data of 97 pregnant smokers were collected during prenatal consultation. Structural equation modelling was performed to assess the construct validity of the MRSS: an exploratory factor analysis was conducted, followed by a confirmatory factor analysis.Test-retest reliability (<16 weeks and 32-34 weeks pregnancy) and internal consistency were assessed using the intraclass correlation coefficient and the Cronbach's alpha, respectively. To verify concurrent validity, Mann-Whitney U-tests were performed examining associations between the MRSS subscales and nicotine dependence, daily consumption, depressive symptoms and intention to quit.

RESULTS

We found a factorial structure for the MRSS of 11 items within five subscales in order of importance: tension reduction, addiction, pleasure, habit and social function. Results for internal consistency and test-retest reliability were good to acceptable. There were significant associations of nicotine dependence with tension reduction and addiction and of daily consumption with addiction and habit.

CONCLUSIONS

Validity and reliability of the MRSS were shown in a sample of pregnant smokers. Tension reduction was the most important reason for continued smoking, followed by pleasure and addiction. Although the score for nicotine dependence was low, addiction was an important reason for continued smoking during pregnancy; therefore, nicotine replacement therapy could be considered. Half of the respondents experienced depressive symptoms. Hence, it is important to identify those women who need more specialized care, which can include not only smoking cessation counselling but also treatment for depression.

Best (but oft-forgotten) practices: the design, analysis, and interpretation of Mendelian randomization studies

Mendelian randomization (MR) is an increasingly important tool for appraising causality in observational epidemiology. The technique exploits the principle that genotypes are not generally susceptible to reverse causation bias and confounding, reflecting their fixed nature and Mendel’s first and second laws of inheritance. The approach is, however, subject to important limitations and assumptions that, if unaddressed or compounded by poor study design, can lead to erroneous conclusions. Nevertheless, the advent of 2-sample approaches (in which exposure and outcome are measured in separate samples) and the increasing availability of open-access data from large consortia of genome-wide association studies and population biobanks mean that the approach is likely to become routine practice in evidence synthesis and causal inference research. In this article we provide an overview of the design, analysis, and interpretation of MR studies, with a special emphasis on assumptions and limitations. We also consider different analytic strategies for strengthening causal inference. Although impossible to prove causality with any single approach, MR is a highly cost-effective strategy for prioritizing intervention targets for disease prevention and for strengthening the evidence base for public health policy.

Mediation analysis to estimate direct and indirect milk losses associated with bacterial load in bovine subclinical mammary infections

Milk losses associated with mastitis can be attributed to either effects of pathogens per se (i.e. direct losses) or to effects of the immune response triggered by the presence of mammary pathogens (i.e. indirect losses). Test-day milk somatic cell counts (SCC) and number of bacterial colony forming units (CFU) found in milk samples are putative measures of the level of immune response and of the bacterial load, respectively. Mediation models, in which one independent variable affects a second variable which, in turn, affects a third one, are conceivable models to estimate direct and indirect losses. Here, we evaluated the feasibility of a mediation model in which test-day SCC and milk were regressed toward bacterial CFU measured at three selected sampling dates, 1 week apart. We applied this method on cows free of clinical signs and with records on up to 3 test-days before and after the date of the first bacteriological samples. Most bacteriological cultures were negative (52.38%), others contained either staphylococci (23.08%), streptococci (9.16%), mixed bacteria (8.79%) or were contaminated (6.59%). Only losses mediated by an increase in SCC were significantly different from null. In cows with three consecutive bacteriological positive results, we estimated a decreased milk yield of 0.28 kg per day for each unit increase in log2-transformed CFU that elicited one unit increase in log2-transformed SCC. In cows with one or two bacteriological positive results, indirect milk loss was not significantly different from null although test-day milk decreased by 0.74 kg per day for each unit increase of log2-transformed SCC. These results highlight the importance of milk losses that are mediated by an increase in SCC during mammary infection and the feasibility of decomposing total milk loss into its direct and indirect components.

Mediation Analysis: A Practitioner's Guide

This article provides an overview of recent developments in mediation analysis, that is, analyses used to assess the relative magnitude of different pathways and mechanisms by which an exposure may affect an outcome. Traditional approaches to mediation in the biomedical and social sciences are described. Attention is given to the confounding assumptions required for a causal interpretation of direct and indirect effect estimates. Methods from the causal inference literature to conduct mediation in the presence of exposure-mediator interactions, binary outcomes, binary mediators, and case-control study designs are presented. Sensitivity analysis techniques for unmeasured confounding and measurement error are introduced. Discussion is given to extensions to time-to-event outcomes and multiple mediators. Further flexible modeling strategies arising from the precise counterfactual definitions of direct and indirect effects are also described. The focus throughout is on methodology that is easily implementable in practice across a broad range of potential applications.

Multiple pathways from the neighborhood food environment to increased body mass index through dietary behaviors: A structural equation-based analysis in the CARDIA study

OBJECTIVES

To examine longitudinal pathways from multiple types of neighborhood restaurants and food stores to BMI, through dietary behaviors.

METHODS

We used data from participants (n=5114) in the United States-based Coronary Artery Risk Development in Young Adults study and a structural equation model to estimate longitudinal (1985-86 to 2005-06) pathways simultaneously from neighborhood fast food restaurants, sit-down restaurants, supermarkets, and convenience stores to BMI through dietary behaviors, controlling for socioeconomic status (SES) and physical activity.

RESULTS

Higher numbers of neighborhood fast food restaurants and lower numbers of sit-down restaurants were associated with higher consumption of an obesogenic fast food-type diet. The pathways from food stores to BMI through diet were inconsistent in magnitude and statistical significance.

CONCLUSIONS

Efforts to decrease the numbers of neighborhood fast food restaurants and to increase the numbers of sit-down restaurant options could influence diet behaviors. Availability of neighborhood fast food and sit-down restaurants may play comparatively stronger roles than food stores in shaping dietary behaviors and BMI.

Are Early-Life Socioeconomic Conditions Directly Related to Birth Outcomes? Grandmaternal Education, Grandchild Birth Weight, and Associated Bias Analyses

Grandmaternal education may be related to grandchild birth weight (GBW) through maternal early-life development; however, conventional regression models may be endogenously confounded. Alternative models employing explicit structural assumptions may provide incrementally clearer evidence. We used data from the US National Longitudinal Study of Adolescent to Adult Health (1995-2009; 1,681 mother-child pairs) to estimate "direct effects" of grandmaternal educational level (less than high school, high school diploma or equivalent, or college degree) at the time of the mother's birth on GBW, adjusted for maternal life-course factors: maltreatment as a child, education and income as an adult, prepregnancy overweight, and prenatal smoking. Using conventional and marginal structural model (MSM) approaches, we estimated 54-g (95% confidence interval: -14.0, 122.1) and 87-g (95% confidence interval: 10.9, 162.5) higher GBWs per increase in educational level, respectively. The MSM allowed simultaneous mediation by and adjustment for prepregnancy overweight. Estimates were insensitive to alternate structural assumptions and mediator parameterizations. Bias analysis suggested that a single unmeasured confounder would have to have a strong influence on GBW (approximately 150 g) or be greatly imbalanced across exposure groups (approximately 25%) to completely explain the findings. Coupling an MSM with sensitivity analyses provides some evidence that maternal early-life socioeconomic environment is directly associated with offspring birth weight.

Mendelian Randomization versus Path Models: Making Causal Inferences in Genetic Epidemiology

OBJECTIVE

The term Mendelian randomization is popular in the current literature. The first aim of this work is to describe the idea of Mendelian randomization studies and the assumptions required for drawing valid conclusions. The second aim is to contrast Mendelian randomization and path modeling when different 'omics' levels are considered jointly.

METHODS

We define Mendelian randomization as introduced by Katan in 1986, and review its crucial assumptions. We introduce path models as the relevant additional component to the current use of Mendelian randomization studies in 'omics'. Real data examples for the association between lipid levels and coronary artery disease illustrate the use of path models.

RESULTS

Numerous assumptions underlie Mendelian randomization, and they are difficult to be fulfilled in applications. Path models are suitable for investigating causality, and they should not be mixed up with the term Mendelian randomization. In many applications, path modeling would be the appropriate analysis in addition to a simple Mendelian randomization analysis.

CONCLUSIONS

Mendelian randomization and path models use different concepts for causal inference. Path modeling but not simple Mendelian randomization analysis is well suited to study causality with different levels of 'omics' data.

Comorbid physical and mental illnesses among pathological gamblers: Results from a population based study in Singapore

The aim of the current study was to examine the comorbidity of pathological gambling with other mental and physical disorders as well as to examine health related quality of life perceived by those with pathological gambling using data from a community survey in Singapore. All respondents were administered the South Oaks Gambling Screen to screen for pathological gambling. The diagnosis of mental disorders was established using the Composite International Diagnostic Interview; while chronic physical conditions were established using a checklist. The weighted lifetime prevalence of pathological gambling was 2.7%. After multiple logistic regression, age 18-34 years (OR=5.3, 95% CI=1.6-17.4), male gender (OR=7.8, CI=3.8-16.2), widowhood (OR=4.2, 95% CI=1.02-17.5), and those with pre-primary (OR=17.1, CI=4.9-59.1), primary (OR=5.3, CI=1.7-16.6), and secondary education (OR=6, CI=2.5-14.7) had significantly higher odds of having pathological gambling. Those of Malay (OR=0.1, 95% CI=0.07-0.2) and Indian ethnicity (OR=0.2, 95% CI=0.1-0.3) had significantly lower odds of having pathological gambling compared to those of Chinese ethnicity. Pathological gamblers had significantly higher odds of having comorbid mental and physical disorders than non-gamblers/non-problem gamblers. The significant association of comorbid mental and physical disorders among those with pathological gambling indicates a need to screen for these disorders and for their subsequent treatment.

Structural equation modeling of immunotoxicity associated with exposure to perfluorinated alkylates

BACKGROUND

Exposure to perfluorinated alkylate substances (PFASs) is associated with immune suppression in animal models, and serum concentrations of specific antibodies against certain childhood vaccines tend to decrease at higher exposures. As such, we investigated the immunotoxic impacts of the three major PFASs in a Faroese birth cohort.

METHODS

A total of 464 children contributed blood samples collected at age 7 years. PFAS concentrations and concentrations of antibodies against diphtheria and tetanus were assessed in serum at age 7 years, and results were available from samples collected at age 5. In addition to standard regressions, structural equation models were generated to determine the association between three major PFASs measured at the two points in time and the two antibody concentrations.

RESULTS

Concentrations of all three 7-year PFAS concentrations were individually associated with a decrease in concentrations of antibodies, however, it was not possible to attribute causality to any single PFAS concentration. Hence, the three 7-year concentrations were combined and showed that a 2-fold increase in PFAS was associated with a decrease by 54.4% (95% CI: 22.0%, 73.3%) in the antibody concentration. If considering both the age-5 and age-7 concentrations of the three major PFASs, the exposure showed a slightly greater loss.

CONCLUSIONS

These analyses strengthen the evidence of human PFAS immunotoxicity at current exposure levels and reflect the usefulness of structural equation models to adjust for imprecision in the exposure variables.

The role of the working alliance in treatment for alcohol problems

Little research has been done on the role of the therapeutic working alliance in treatment for alcohol problems. This longitudinal study’s objectives were (a) to identify predictors of working alliance and (b) to investigate whether client and/or therapist reports of the working alliance predicted posttreatment motivation and then later treatment outcome. Client and therapist perceptions of the working alliance were assessed after the first treatment session using a short form of the Working Alliance Inventory (WAI) among 173 clients taking part in the United Kingdom Alcohol Treatment Trial (UKATT) and randomized to motivational enhancement therapy (MET) or social behavior and network therapy (SBNT) with complete data on all measures of interest. Structural equation models were fitted to identify predictors of WAI scores and investigate the relationships between WAI and measures of drinking during treatment, posttreatment motivation, and successful treatment outcome (abstinent or nonproblem drinker), and measures of drinks per drinking day and nondrinking days, assessed 9 months after the conclusion of treatment. Motivation to change drinking when treatment began was a strong predictor of client—adjusted coefficient = 2.21 (95% confidence interval [CI] [0.36, 4.06]—but not therapist WAI. Client WAI predicted successful treatment outcome—adjusted odds ratios (OR) = 1.09 (95% CI [1.02, 1.17])—and had effects on drinking during treatment, and on posttreatment motivation to change. There was evidence for effect modification by treatment, with strong associations between WAI and posttreatment motivation, and evidence of WAI prediction of treatment outcomes in the MET group, but no evidence of associations for SBNT. Therapist WAI was not strongly associated with treatment outcome (adjusted OR = 1.05; 95% CI [0.99, 1.10]). The working alliance is important to treatment outcomes for alcohol problems, with client evaluation of the alliance strongly related to motivation to change drinking throughout treatment for MET. It was also much more important than therapist-rated alliance in this study.

Modeling contextual effects using individual-level data and without aggregation: an illustration of multilevel factor analysis (MLFA) with collective efficacy

Population health scientists increasingly study how contextual-level attributes affect individual health. A major challenge in this domain relates to measurement, i.e., how best to measure and create variables that capture characteristics of individuals and their embedded contexts. This paper presents an illustration of multilevel factor analysis (MLFA), an analytic method that enables researchers to model contextual effects using individual-level data without using derived variables. MLFA uses the shared variance in sets of observed items among individuals within the same context to estimate a measurement model for latent constructs; it does this by decomposing the total sample variance-covariance matrix into within-group (e.g., individual-level) and between-group (e.g., contextual-level) matrices and simultaneously modeling distinct latent factor structures at each level. We illustrate the MLFA method using items capturing collective efficacy, which were self-reported by 2,599 adults in 65 census tracts from the Los Angeles Family and Neighborhood Survey (LAFANS). MLFA identified two latent factors at the individual level and one factor at the neighborhood level. Indicators of collective efficacy performed differently at each level. The ability of MLFA to identify different latent factor structures at each level underscores the utility of this analytic tool to model and identify attributes of contexts relevant to health.

The impact of neighborhood quality, perceived stress, and social support on depressive symptoms during pregnancy in African American women

Living in a lower-quality neighborhood is associated with higher levels of depressive symptoms in the general population as well as among pregnant and postpartum women. However, little is known of the important pathways by which this association occurs. We proposed a model in which perceived stress and social support mediated the effects of neighborhood quality on depressive symptoms during pregnancy (measured by the 20-item Center for Epidemiologic Studies-Depression, CES-D, scale) in a sample of 1383 African American women from the Detroit metropolitan area interviewed during their delivery hospitalization. Using structural equation modeling (SEM), we built a latent variable of neighborhood quality using 4 measures (neighborhood disorder, neighborhood safety/danger, walking environment, overall rating). We then tested two SEM mediation models. We found that lower neighborhood quality was associated with higher prevalence of depressive symptoms during pregnancy (standardized total effect = .16, p = .011). We found that perceived stress partially mediated the neighborhood quality association with depressive symptoms. Although the association of social support with depressive symptoms was negligible, social support mediated associations of neighborhood quality with perceived stress [standardized path coefficient = .38 (.02), p = .009]. Our results point to the need for public health, health care, as well as non-health related interventions (e.g. crime prevention programs) to decrease overall exposure to stressors, as well as stress levels of women living in poor quality neighborhoods. Interventions that increase the levels of social support of women during pregnancy are also needed for their potential to decrease stress and ultimately improve mental health at this important time in the life course.

Mediation analysis to estimate direct and indirect milk losses due to clinical mastitis in dairy cattle

Milk losses associated with mastitis can be attributed to either effects of pathogens per se (i.e., direct losses) or effects of the immune response triggered by intramammary infection (indirect losses). The distinction is important in terms of mastitis prevention and treatment. Regardless, the number of pathogens is often unknown (particularly in field studies), making it difficult to estimate direct losses, whereas indirect losses can be approximated by measuring the association between increased somatic cell count (SCC) and milk production. An alternative is to perform a mediation analysis in which changes in milk yield are allocated into their direct and indirect components. We applied this method on data for clinical mastitis, milk and SCC test-day recordings, results of bacteriological cultures (Escherichia coli, Staphylococcus aureus, Streptococcus uberis, coagulase-negative staphylococci, Streptococcus dysgalactiae, and streptococci other than Strep. dysgalactiae and Strep. uberis), and cow characteristics. Following a diagnosis of clinical mastitis, the cow was treated and changes (increase or decrease) in milk production before and after a diagnosis were interpreted counterfactually. On a daily basis, indirect changes, mediated by SCC increase, were significantly different from zero for all bacterial species, with a milk yield decrease (ranging among species from 4 to 33g and mediated by an increase of 1000 SCC/mL/day) before and a daily milk increase (ranging among species from 2 to 12g and mediated by a decrease of 1000 SCC/mL/day) after detection. Direct changes, not mediated by SCC, were only different from zero for coagulase-negative staphylococci before diagnosis (72g per day). We concluded that mixed structural equation models were useful to estimate direct and indirect effects of the presence of clinical mastitis on milk yield.

Mediation analysis with intermediate confounding: structural equation modeling viewed through the causal inference lens

The study of mediation has a long tradition in the social sciences and a relatively more recent one in epidemiology. The first school is linked to path analysis and structural equation models (SEMs), while the second is related mostly to methods developed within the potential outcomes approach to causal inference. By giving model-free definitions of direct and indirect effects and clear assumptions for their identification, the latter school has formalized notions intuitively developed in the former and has greatly increased the flexibility of the models involved. However, through its predominant focus on nonparametric identification, the causal inference approach to effect decomposition via natural effects is limited to settings that exclude intermediate confounders. Such confounders are naturally dealt with (albeit with the caveats of informality and modeling inflexibility) in the SEM framework. Therefore, it seems pertinent to revisit SEMs with intermediate confounders, armed with the formal definitions and (parametric) identification assumptions from causal inference. Here we investigate: 1) how identification assumptions affect the specification of SEMs, 2) whether the more restrictive SEM assumptions can be relaxed, and 3) whether existing sensitivity analyses can be extended to this setting. Data from the Avon Longitudinal Study of Parents and Children (1990-2005) are used for illustration.

Do genetic risk scores for body mass index predict risk of phobic anxiety? Evidence for a shared genetic risk factor

BACKGROUND

Obesity and anxiety are often linked but the direction of effects is not clear.

METHOD

Using genetic instrumental variable (IV) analyses in 5911 female participants from the Nurses' Health Study (NHS, initiated 1976) and 3697 male participants from the Health Professional Follow-up Study (HPFS, initiated 1986), we aimed to determine whether obesity increases symptoms of phobic anxiety. As instrumental variables we used the fat mass and obesity-associated (FTO) gene, the melanocortin 4 receptor (MC4R) gene and a genetic risk score (GRS) based on 32 single nucleotide polymorphisms (SNPs) that significantly predict body mass index (BMI). 'Functional' GRSs corresponding with specific biological pathways that shape BMI (adipogenesis, appetite and cardiopulmonary) were considered. The main outcome was phobic anxiety measured by the Crown Crisp Index (CCI) in 2004 in the NHS and in 2000 in the HPFS.

RESULTS

In observational analysis, a 1-unit higher BMI was associated with higher phobic anxiety symptoms [women: β = 0.05, 95% confidence interval (CI) 0.030-0.068; men: β = 0.04, 95% CI 0.016-0.071). IV analyses showed that BMI was associated with higher phobic anxiety symptoms in the FTO-instrumented analysis (p = 0.005) but not in the GRS-instrumented analysis (p = 0.256). Functional GRSs showed heterogeneous, non-significant effects of BMI on phobic anxiety symptoms.

CONCLUSIONS

Our findings do not provide conclusive evidence in favor of the hypothesis that higher BMI leads to higher levels of phobic anxiety, but rather suggest that genes that influence obesity, in particular FTO, may have direct effects on phobic anxiety, and hence that obesity and phobic anxiety may share common genetic determinants.

Self-efficacy beliefs are associated with visual height intolerance: a cross-sectional survey

BACKGROUND

Responses to height may range from indifference to minor distress to severe symptoms of fear of heights (acrophobia); visual height intolerance (vHI) denotes the whole spectrum of symptoms. Although there are options to manage vHI, only a small part of persons affected by vHI are willing to seek professional help or confront their problem. Purpose of this study was to determine if persons with vHI, specifically those who show avoidant behavior towards heights (avoiders), score lower in their general self-efficacy (GSE) than those who confront vHI (confronters).

METHOD

Cross-sectional survey in 607 individuals living in the urban region of Munich, Germany, using a mailed questionnaire on presence or absence of vHI, confronting or avoiding behaviour, and GSE.

RESULTS

Of all participants (mean age 53.9, 50.3% female), 407 reported life-time presence of vHI. Participants with vHI had a mean GSE score of 31.8 (SD 4.3) points (participants without vHI: 32.5, SD 4.3, p = 0.008 for difference). Among individuals with vHI, 23% reported confronting behavior. Confronters were significantly younger (p<.0001, 50.2 vs. 55.7 years), more likely to be female (p = 0.0039, 64.3% female), and had a higher GSE score (p = 0.0049, 32.5 vs. 31.1). Associations remained significant after multiple adjustment.

CONCLUSIONS

Our study provides evidence for the association of GSE and vHI. These findings may have consequences for strategies of alleviation and therapy of vHI.

Population-based assessment of cancer survivors' financial burden and quality of life: a prospective cohort study

PURPOSE

The impact of financial burden among patients with cancer has not yet been measured in a way that accounts for inter-relationships between quality of life, perceived quality of care, disease status, and sociodemographic characteristics.

PATIENTS AND METHODS

In a national, prospective, observational, population- and health care systems-based cohort study, patients with colorectal or lung cancer were enrolled from 2003 to 2006 within 3 months of diagnosis. For this analysis, surviving patients who were either disease free or had advanced disease were resurveyed a median 7.3 years from diagnosis. Structural equation modeling was used to investigate relationships between financial burden, quality of life, perceived quality of care, and sociodemographic characteristics.

RESULTS

Among 1,000 participants enrolled from five geographic regions, five integrated health care systems, or 15 Veterans Administration Hospitals, 89% (n = 889) were cancer free, and 11% (n = 111) had advanced cancer. Overall, 48% (n = 482) reported difficulties living on their household income, and 41% (n = 396) believed their health care to be "excellent." High financial burden was associated with lower household income (adjusted odds ratio [OR] = 0.61 per $20k per year, P < .001) and younger age (adjusted OR = 0.63 per 10 years; P < .001). High financial burden was also associated with poorer quality of life (adjusted beta = -0.06 per burden category; P < .001). Better quality of life was associated with fewer perceptions of poorer quality of care (adjusted OR = 0.85 per 0.10 EuroQol units; P < .001).

CONCLUSION

Financial burden is prevalent among cancer survivors and is related to patients' health-related quality of life. Future studies should consider interventions to improve patient education and engagement with regard to financial burden.

The role of sleep disturbances in the longitudinal relationship between psychosocial working conditions, measured by work demands and support, and depression

STUDY OBJECTIVES

Because work demands and lack of social support seem to be prospectively linked to sleep problems, and sleep problems are linked to depression, sleep problems may play a role in the relationship between these work characteristics and depressive symptoms. In order to shed more light on this relationship, the current study investigated whether disturbed sleep is a mediator in the longitudinal relationships between work demands, social support, and depression.

DESIGN

Longitudinal cohort study with repeated survey measures on four occasions.

SETTING

Swedish workforce.

PARTICIPANTS

2,017 working participants from the Swedish Longitudinal Occupational Survey of Health in 2006, 2008, 2010, and 2012.

MEASUREMENTS AND RESULTS

Work demands (four items) and social support (six items) were assessed with the Demand Control Questionnaire, disturbed sleep (four items) with the Karolinska Sleep Questionnaire, and depressive symptoms with a brief subscale (six items) from the Symptom Checklist. Autoregressive longitudinal mediation models using structural equation modeling were tested. The work characteristics, and disturbed sleep, were found to be separately associated with depressive symptoms in subsequent waves. However, only demands were found to be longitudinally related to subsequent disturbed sleep. The longitudinal autoregressive models supported a weak mediating role of disturbed sleep in the relationship between demands and depressive symptoms (standardized beta 0.008, P < 0.001), but not between support and depressive symptoms.

CONCLUSIONS

These results indicate that higher demands at work might cause an increase in depressive symptoms, in part, by increasing disturbed sleep, although the mediated effect was relatively small compared to the total effect.

Social norms and its correlates as a pathway to smoking among young Latino adults

Socially and culturally embedded norms regarding smoking may be one pathway by which individuals adopt smoking behaviors. However, few studies have examined if social norms operate in young adults, a population at high risk of becoming regular smokers. There is also little research examining correlates of social norms in populations with a large immigrant segment, where social norms are likely to differ from the receiving country and could contribute to a better understanding of previously reported acculturation-health associations. Using data from a nationally representative sample of young adults in the United States reached via a novel cell-phone sampling design, we explored the relationships between acculturation proxies (nativity, language spoken and generational status), socioeconomic position (SEP), smoking social norms and current smoking status among Latinos 18-34 years of age (n = 873). Specifically, we examined if a measure of injunctive norms assessed by asking participants about the acceptability of smoking among Latino co-ethnic peers was associated with acculturation proxies and SEP. Results showed a strong gradient in smoking social norms by acculturation proxies, with significantly less acceptance of smoking reported among the foreign-born and increasing acceptance among those speaking only/mostly English at home and third-generation individuals. No consistent and significant pattern in smoking social norms was observed by education, income or employment status, possibly due to the age of the study population. Lastly, those who reported that their Latino peers do not find smoking acceptable were significantly less likely to be current smokers compared to those who said their Latino peers were ambivalent about smoking (do not care either way) in crude models, and in models that adjusted for age, sex, generational status, language spoken, and SEP. This study provides new evidence regarding the role of social norms in shaping smoking behaviors among Latino young adults and suggests distinct influences of acculturation proxies and socioeconomic condition on smoking social norms in this population.

Understanding treatment effect mechanisms of the CAMBRA randomized trial in reducing caries increment

The Caries Management By Risk Assessment (CAMBRA) randomized controlled trial showed that an intervention featuring combined antibacterial and fluoride therapy significantly reduced bacterial load and suggested reduced caries increment in adults with 1 to 7 baseline cavitated teeth. While trial results speak to the overall effectiveness of an intervention, insight can be gained from understanding the mechanism by which an intervention acts on putative intermediate variables (mediators) to affect outcomes. This study conducted mediation analyses on 109 participants who completed the trial to understand whether the intervention reduced caries increment through its action on potential mediators (oral bacterial load, fluoride levels, and overall caries risk based on the composite of bacterial challenge and salivary fluoride) between the intervention and dental outcomes. The primary outcome was the increment from baseline in decayed, missing, and filled permanent surfaces (ΔDMFS) 24 mo after completing restorations for baseline cavitated lesions. Analyses adjusted for baseline overall risk, bacterial challenge, and fluoride values under a potential outcome framework using generalized linear models. Overall, the CAMBRA intervention was suggestive in reducing the 24-mo DMFS increment (reduction in ΔDMFS: -0.96; 95% confidence interval [CI]: -2.01 to 0.08; P = 0.07); the intervention significantly reduced the 12-mo overall risk (reduction in overall risk: -19%; 95% CI, -7 to -41%;], P = 0.005). Individual mediators, salivary log10 mutans streptococci, log10 lactobacilli, and fluoride level, did not represent statistically significant pathways alone through which the intervention effect was transmitted. However, 36% of the intervention effect on 24-mo DMFS increment was through a mediation effect on 12-mo overall risk (P = 0.03). These findings suggest a greater intervention effect carried through the combined action on multiple aspects of the caries process rather than through any single factor. In addition, a substantial portion of the total effect of the CAMBRA intervention may have operated through unanticipated or unmeasured pathways not included among the potential mediators studied.

Individual characteristics associated with mismatches between self-reported and accelerometer-measured physical activity

Background

Accurate assessment tools are required for the surveillance of physical activity (PA) levels and the assessment of the effect of interventions. In addition, increasing awareness of PA is often used as the first step in pragmatic behavioural interventions, as discrepancies between the amount of activity an individual perceives they do and the amount actually undertaken may act as a barrier to change. Previous research has demonstrated differences in the amount of activity individuals report doing, compared to their level of physical activity when measured with an accelerometer. Understanding the characteristics of those whose PA level is ranked differently when measured with either self-report or accelerometry is important as it may inform the choice of instrument for future research. The aim of this project was to determine which individual characteristics are associated with differences between self-reported and accelerometer measured physical activity.

Methods

Participant data from the 2009 wave of the Commuting and Health in Cambridge study were used. Quartiles of self-reported and accelerometer-measured PA were derived by ranking each measure from lowest to highest. These quartiles were compared to determine whether individuals’ physical activity was ranked higher by either method. Multinomial logistic regression models were used to investigate the individual characteristics associated with different categories of mismatch.

Results

Data from 486 participants (70% female) were included in the analysis. In adjusted analyses, the physical activity of overweight or obese individuals was significantly more likely to be ranked higher by self-report than by accelerometer than that of normal-weight individuals (OR = 2.07, 95%CI = 1.28–3.34), particularly among women (OR = 3.97, 95%CI = 2.11–7.47).

Conclusions

There was a greater likelihood of mismatch between self-reported and accelerometer measured physical activity levels in overweight or obese adults. Future studies in overweight or obese adults should consider employing both methods of measurement.

Challenges and future directions to evaluating the association between prenatal exposure to endocrine disrupting chemicals and childhood obesity

Obesity is an increasing public health threat worldwide. However, there has been insufficient research addressing the obesogenic potential of prenatal exposure to environmental endocrine disrupting chemicals, largely due to complexities in the design, analysis, and interpretation of such studies. This review describes relevant biological mechanisms, addresses current challenges for investigators, presents potential strategies for overcoming them, and identifies areas where further development is required to improve future research. Special considerations for exposure assessment, outcome heterogeneity, and complex confounding structures are described.

The association between global DNA methylation and telomere length in a longitudinal study of boilermakers

The objectives of this study were to determine if global DNA methylation, as reflected in LINE-1 and Alu elements, is associated with telomere length and whether it modifies the rate of telomeric change. A repeated-measures longitudinal study was performed with a panel of 87 boilermaker subjects. The follow-up period was 29 months. LINE-1 and Alu methylation was determined using pyrosequencing. Leukocyte relative telomere length was assessed via real-time qPCR. Linear-mixed models were used to estimate the association between DNA methylation and telomere length. A structural equation model (SEM) was used to explore the hypothesized relationship between DNA methylation, proxies of particulate matter exposure, and telomere length at baseline. There appeared to be a positive association between both LINE-1 and Alu methylation levels, and telomere length. For every incremental increase in LINE-1 methylation, there was a statistically significant 1.0 × 10(-1) (95% CI: 4.6 × 10(-2), 1.5 × 10(-1), P < 0.01) unit increase in relative telomere length, controlling for age at baseline, current and past smoking status, work history, BMI (log kg/m(2) ) and leukocyte differentials. Furthermore, for every incremental increase in Alu methylation, there was a statistically significant 6.2 × 10(-2) (95% CI: 1.0 × 10(-2), 1.1 × 10(-1), P = 0.02) unit increase in relative telomere length. The interaction between LINE-1 methylation and follow-up time was statistically significant with an estimate -9.8 × 10(-3) (95% CI: -1.8 × 10(-2), -1.9 × 10(-3), P = 0.02); suggesting that the rate of telomeric change was modified by the degree of LINE-1 methylation. No statistically significant association was found between the cumulative PM exposure construct, with global DNA methylation and telomere length at baseline.

Acute alcohol-related dysfunction as a predictor of employment status in a longitudinal study of working-age men in Izhevsk, Russia

AIMS

To investigate longitudinally the effect of alcohol consumption and related acute alcohol-related dysfunction on employment status.

DESIGN, SETTING AND PARTICIPANTS

A total of 1143 men aged 25-55 years in regular paid employment and resident in the city of Izhevsk, Russia were interviewed between 2003-06 and then re-interviewed (2008-09) and their employment status ascertained.

MEASUREMENTS

Exposures of interest were baseline alcohol intake (yearly total volume of ethanol consumed and non-beverage alcohols) and alcohol-related dysfunction, measured by a latent variable defined in terms of frequency of alcohol-related dysfunctional behaviours and by one or more episodes of zapoi (a period of continuous drunkenness lasting 2 or more days). The outcome of interest was whether or not men were still in regular paid employment at follow-up. The inter-relationship between these variables was investigated using structural equation modelling.

FINDINGS

Total volume of ethanol consumed had no substantive effect on future employment status; however, taking into account education and other socio-demographic factors, there was strong evidence that loss of regular paid employment at follow-up was influenced by non-beverage alcohol consumption [odds ratio = 2.30 for non-beverage drinkers compared with beverage-only drinkers, 95% confidence interval (CI) = 1.21, 4.40)], latent acute alcohol-related dysfunction (odds ratio = 1.50 per standard deviation increase in dysfunction score, 95% CI = 1.20, 1.88) and zapoi (odds ratio = 3.08, 95% CI = 1.71, 5.55). Acute alcohol-related dysfunction was an important mediator of the relationship between non-beverage alcohol use and employment status.

CONCLUSIONS

Acute alcohol-related dysfunction is an important factor in determining whether men remain in employment and an important mediator of the effects of alcohol intake.

[Causal analysis approaches in epidemiology]

Epidemiological research is mostly based on observational studies. Whether such studies can provide evidence of causation remains discussed. Several causal analysis methods have been developed in epidemiology. This paper aims at presenting an overview of these methods: graphical models, path analysis and its extensions, and models based on the counterfactual approach, with a special emphasis on marginal structural models. Graphical approaches have been developed to allow synthetic representations of supposed causal relationships in a given problem. They serve as qualitative support in the study of causal relationships. The sufficient-component cause model has been developed to deal with the issue of multicausality raised by the emergence of chronic multifactorial diseases. Directed acyclic graphs are mostly used as a visual tool to identify possible confounding sources in a study. Structural equations models, the main extension of path analysis, combine a system of equations and a path diagram, representing a set of possible causal relationships. They allow quantifying direct and indirect effects in a general model in which several relationships can be tested simultaneously. Dynamic path analysis further takes into account the role of time. The counterfactual approach defines causality by comparing the observed event and the counterfactual event (the event that would have been observed if, contrary to the fact, the subject had received a different exposure than the one he actually received). This theoretical approach has shown limits of traditional methods to address some causality questions. In particular, in longitudinal studies, when there is time-varying confounding, classical methods (regressions) may be biased. Marginal structural models have been developed to address this issue. In conclusion, "causal models", though they were developed partly independently, are based on equivalent logical foundations. A crucial step in the application of these models is the formulation of causal hypotheses, which will be a basis for all methodological choices. Beyond this step, statistical analysis tools recently developed offer new possibilities to delineate complex relationships, in particular in life course epidemiology.

Fixed effects analysis of repeated measures data

The analysis of repeated measures or panel data allows control of some of the biases which plague other observational studies, particularly unmeasured confounding. When this bias is suspected, and the research question is: 'Does a change in an exposure cause a change in the outcome?', a fixed effects approach can reduce the impact of confounding by time-invariant factors, such as the unmeasured characteristics of individuals. Epidemiologists familiar with using mixed models may initially presume that specifying a random effect (intercept) for every individual in the study is an appropriate method. However, this method uses information from both the within-individual/unit exposure-outcome association and the between-individual/unit exposure-outcome association. Variation between individuals may introduce confounding bias into mixed model estimates, if unmeasured time-invariant factors are associated with both the exposure and the outcome. Fixed effects estimators rely only on variation within individuals and hence are not affected by confounding from unmeasured time-invariant factors. The reduction in bias using a fixed effects model may come at the expense of precision, particularly if there is little change in exposures over time. Neither fixed effects nor mixed models control for unmeasured time-varying confounding or reverse causation.

Causal inference algorithms can be useful in life course epidemiology

OBJECTIVES

Life course epidemiology attempts to unravel causal relationships between variables observed over time. Causal relationships can be represented as directed acyclic graphs. This article explains the theoretical concepts of the search algorithms used for finding such representations, discusses various types of such algorithms, and exemplifies their use in the context of obesity and insulin resistance.

STUDY DESIGN AND SETTING

We investigated possible causal relations between gender, birth weight, waist circumference, and blood glucose level of 4,081 adult participants of the Prevention of REnal and Vascular ENd-stage Disease study. The latter two variables were measured at three time points at intervals of about 3 years.

RESULTS

We present the resulting causal graphs, estimate parameters of the corresponding structural equation models, and discuss usefulness and limitations of this methodology.

CONCLUSION

As an exploratory method, causal graphs and the associated theory can help construct possible causal models underlying observational data. In this way, the causal search algorithms provide a valuable statistical tool for life course epidemiological research.

Impact on infants' cognitive development of antenatal exposure to iron deficiency disorder and common mental disorders

Objectives

The aim of this study was to examine the effects of antenatal exposure to iron deficiency anemia (IDA) and common mental disorders (CMD) on cognitive development of 6 months old infants in a developing country.

Methods

A prospective population-based study in a rural province in Vietnam, which enrolled pregnant women at 12–20 weeks gestation and followed them up with their infants until six months postpartum. Criteria for IDA were Hb <11 g/dL and serum ferritin <15 ng/mL. CMD symptoms were assessed by the Edinburgh Postnatal Depression Scale-Vietnam validation. Infant cognitive development was assessed by Bayley Scales of Infant and Toddler Development, 3^rd Ed. Path analyses were performed to determine the direct and indirect, partly or fully mediated, causal effects of the antenatal exposures.

Results

A total of 497 pregnant women were recruited, of those 378 women provided complete data which were included in the analyses. Statistically significant direct adverse effects of persistent antenatal IDA (estimated difference of −11.62 points; 95% CI −23.01 to −0.22) and antenatal CMD (−4.80 points; 95% CI: −9.40 to −0.20) on infant Bayley cognitive scores at six months were found. Higher birthweight, household wealth, and self-rated sufficient supply of breastmilk were associated with higher cognitive scores. Maternal age >30 years and primiparity had an indirect adverse effect on infants’ Bayley cognitive scores.

Conclusions

These findings suggest that antenatal IDA and CMD both have adverse effects on child cognitive development, which if unrecognized and unaddressed are likely to be lasting. It is crucial that both these risks are considered by policy makers, clinicians, and researchers seeking to improve child cognitive function in developing countries.

Potential uses of Bayesian networks as tools for synthesis of systematic reviews of complex interventions

Bayesian networks (BNs) are tools for representing expert knowledge or evidence. They are especially useful for synthesising evidence or belief concerning a complex intervention, assessing the sensitivity of outcomes to different situations or contextual frameworks and framing decision problems that involve alternative types of intervention. Bayesian networks are useful extensions to logic maps when initiating a review or to facilitate synthesis and bridge the gap between evidence acquisition and decision-making. Formal elicitation techniques allow development of BNs on the basis of expert opinion. Such applications are useful alternatives to 'empty' reviews, which identify knowledge gaps but fail to support decision-making. Where review evidence exists, it can inform the development of a BN. We illustrate the construction of a BN using a motivating example that demonstrates how BNs can ensure coherence, transparently structure the problem addressed by a complex intervention and assess sensitivity to context, all of which are critical components of robust reviews of complex interventions. We suggest that BNs should be utilised to routinely synthesise reviews of complex interventions or empty reviews where decisions must be made despite poor evidence.

On the reciprocal association between loneliness and subjective well-being

Loneliness has been shown to longitudinally predict subjective well-being. The authors used data from a longitudinal population-based study (2002-2006) of non-Hispanic white, African-American, and nonblack Latino-American persons born between 1935 and 1952 and living in Cook County, Illinois. They applied marginal structural models for time-varying exposures to examine the magnitude and persistence of the effects of loneliness on subjective well-being and of subjective well-being on loneliness. Their results indicate that, if interventions on loneliness were made 1 and 2 years prior to assessing final subjective well-being, then only the intervention 1 year prior would have an effect (standardized effect = -0.29). In contrast, increases in subjective well-being 1 year prior (standardized effect = -0.26) and 2 years prior (standardized effect = -0.13) to assessing final loneliness would both have an effect on an individual's final loneliness. These effects persist even after control is made for depressive symptoms, social support, and psychiatric conditions and medications as time-varying confounders. Results from this study indicate an asymmetrical and persistent feedback of fairly substantial magnitude between loneliness and subjective well-being. Mechanisms responsible for the asymmetry are discussed. Developing interventions for loneliness and subjective well-being could have substantial psychological and health benefits.