Handling missing data when estimating causal effects with targeted maximum likelihood estimation.

Targeted maximum likelihood estimation (TMLE) is increasingly used for doubly robust causal inference, but how missing data should be handled when using TMLE with data-adaptive approaches is unclear. Based on data (1992-1998) from the Victorian Adolescent Health Cohort Study, we conducted a simulation study to evaluate 8 missing-data methods in this context: complete-case analysis, extended TMLE incorporating an outcome-missingness model, the missing covariate missing indicator method, and 5 multiple imputation (MI) approaches using parametric or machine-learning models. We considered 6 scenarios that varied in terms of exposure/outcome generation models (presence of confounder-confounder interactions) and missingness mechanisms (whether outcome influenced missingness in other variables and presence of interaction/nonlinear terms in missingness models). Complete-case analysis and extended TMLE had small biases when outcome did not influence missingness in other variables. Parametric MI without interactions had large bias when exposure/outcome generation models included interactions. Parametric MI including interactions performed best in bias and variance reduction across all settings, except when missingness models included a nonlinear term. When choosing a method for handling missing data in the context of TMLE, researchers must consider the missingness mechanism and, for MI, compatibility with the analysis method. In many settings, a parametric MI approach that incorporates interactions and nonlinearities is expected to perform well.

Statistical analysis of observational studies in disability research.

Observational studies have a critical role in disability research, providing the opportunity to address a range of research questions. Over the past decades, there have been substantial shifts and developments in statistical methods for observational studies, most notably for causal inference. In this review, we provide an overview of modern design and analysis concepts critical for observational studies, drawing examples from the field of disability research and highlighting the challenges in this field, to inform the readership on important statistical considerations for their studies.

Recoverability and estimation of causal effects under typical multivariable missingness mechanisms.

In the context of missing data, the identifiability or "recoverability" of the average causal effect (ACE) depends not only on the usual causal assumptions but also on missingness assumptions that can be depicted by adding variable-specific missingness indicators to causal diagrams, creating missingness directed acyclic graphs (m-DAGs). Previous research described canonical m-DAGs, representing typical multivariable missingness mechanisms in epidemiological studies, and examined mathematically the recoverability of the ACE in each case. However, this work assumed no effect modification and did not investigate methods for estimation across such scenarios. Here, we extend this research by determining the recoverability of the ACE in settings with effect modification and conducting a simulation study to evaluate the performance of widely used missing data methods when estimating the ACE using correctly specified g-computation. Methods assessed were complete case analysis (CCA) and various implementations of multiple imputation (MI) with varying degrees of compatibility with the outcome model used in g-computation. Simulations were based on an example from the Victorian Adolescent Health Cohort Study (VAHCS), where interest was in estimating the ACE of adolescent cannabis use on mental health in young adulthood. We found that the ACE is recoverable when no incomplete variable (exposure, outcome, or confounder) causes its own missingness, and nonrecoverable otherwise, in simplified versions of 10 canonical m-DAGs that excluded unmeasured common causes of missingness indicators. Despite this nonrecoverability, simulations showed that MI approaches that are compatible with the outcome model in g-computation may enable approximately unbiased estimation across all canonical m-DAGs considered, except when the outcome causes its own missingness or causes the missingness of a variable that causes its own missingness. In the latter settings, researchers may need to consider sensitivity analysis methods incorporating external information (e.g., delta-adjustment methods). The VAHCS case study illustrates the practical implications of these findings.

A comparison of strategies for selecting auxiliary variables for multiple imputation.

Multiple imputation (MI) is a popular method for handling missing data. Auxiliary variables can be added to the imputation model(s) to improve MI estimates. However, the choice of which auxiliary variables to include is not always straightforward. Several data-driven auxiliary variable selection strategies have been proposed, but there has been limited evaluation of their performance. Using a simulation study we evaluated the performance of eight auxiliary variable selection strategies: (1, 2) two versions of selection based on correlations in the observed data; (3) selection using hypothesis tests of the "missing completely at random" assumption; (4) replacing auxiliary variables with their principal components; (5, 6) forward and forward stepwise selection; (7) forward selection based on the estimated fraction of missing information; and (8) selection via the least absolute shrinkage and selection operator (LASSO). A complete case analysis and an MI analysis using all auxiliary variables (the "full model") were included for comparison. We also applied all strategies to a motivating case study. The full model outperformed all auxiliary variable selection strategies in the simulation study, with the LASSO strategy the best performing auxiliary variable selection strategy overall. All MI analysis strategies that we were able to apply to the case study led to similar estimates, although computational time was substantially reduced when variable selection was employed. This study provides further support for adopting an inclusive auxiliary variable strategy where possible. Auxiliary variable selection using the LASSO may be a promising alternative when the full model fails or is too burdensome.

Confounding-adjustment methods for the causal difference in medians.

BACKGROUND: With continuous outcomes, the average causal effect is typically defined using a contrast of expected potential outcomes. However, in the presence of skewed outcome data, the expectation (population mean) may no longer be meaningful. In practice the typical approach is to continue defining the estimand this way or transform the outcome to obtain a more symmetric distribution, although neither approach may be entirely satisfactory. Alternatively the causal effect can be redefined as a contrast of median potential outcomes, yet discussion of confounding-adjustment methods to estimate the causal difference in medians is limited. In this study we described and compared confounding-adjustment methods to address this gap. METHODS: The methods considered were multivariable quantile regression, an inverse probability weighted (IPW) estimator, weighted quantile regression (another form of IPW) and two little-known implementations of g-computation for this problem. Methods were evaluated within a simulation study under varying degrees of skewness in the outcome and applied to an empirical study using data from the Longitudinal Study of Australian Children. RESULTS: Simulation results indicated the IPW estimator, weighted quantile regression and g-computation implementations minimised bias across all settings when the relevant models were correctly specified, with g-computation additionally minimising the variance. Multivariable quantile regression, which relies on a constant-effect assumption, consistently yielded biased results. Application to the empirical study illustrated the practical value of these methods. CONCLUSION: The presented methods provide appealing avenues for estimating the causal difference in medians.

On the use of multiple imputation to address data missing by design as well as unintended missing data in case-cohort studies with a binary endpoint.

BACKGROUND: Case-cohort studies are conducted within cohort studies, with the defining feature that collection of exposure data is limited to a subset of the cohort, leading to a large proportion of missing data by design. Standard analysis uses inverse probability weighting (IPW) to address this intended missing data, but little research has been conducted into how best to perform analysis when there is also unintended missingness. Multiple imputation (MI) has become a default standard for handling unintended missingness and is typically used in combination with IPW to handle the intended missingness due to the case-control sampling. Alternatively, MI could be used to handle both the intended and unintended missingness. While the performance of an MI-only approach has been investigated in the context of a case-cohort study with a time-to-event outcome, it is unclear how this approach performs with a binary outcome. METHODS: We conducted a simulation study to assess and compare the performance of approaches using only MI, only IPW, and a combination of MI and IPW, for handling intended and unintended missingness in the case-cohort setting. We also applied the approaches to a case study. RESULTS: Our results show that the combined approach is approximately unbiased for estimation of the exposure effect when the sample size is large, and was the least biased with small sample sizes, while MI-only and IPW-only exhibited larger biases in both sample size settings. CONCLUSIONS: These findings suggest that a combined MI/IPW approach should be preferred to handle intended and unintended missing data in case-cohort studies with binary outcomes.

Should multiple imputation be stratified by exposure group when estimating causal effects via outcome regression in observational studies?

BACKGROUND: Despite recent advances in causal inference methods, outcome regression remains the most widely used approach for estimating causal effects in epidemiological studies with a single-point exposure and outcome. Missing data are common in these studies, and complete-case analysis (CCA) and multiple imputation (MI) are two frequently used methods for handling them. In randomised controlled trials (RCTs), it has been shown that MI should be conducted separately by treatment group. In observational studies, causal inference is now understood as the task of emulating an RCT, which raises the question of whether MI should be conducted by exposure group in such studies. METHODS: We addressed this question by evaluating the performance of seven methods for handling missing data when estimating causal effects with outcome regression. We conducted an extensive simulation study based on an illustrative case study from the Victorian Adolescent Health Cohort Study, assessing a range of scenarios, including seven outcome generation models with exposure-confounder interactions of differing strength. RESULTS: The simulation results showed that MI by exposure group led to the least bias when the size of the smallest exposure group was relatively large, followed by MI approaches that included the exposure-confounder interactions. CONCLUSIONS: The findings from our simulation study, which was designed based on a real case study, suggest that current practice for the conduct of MI in causal inference may need to shift to stratifying by exposure group where feasible, or otherwise including exposure-confounder interactions in the imputation model.

Learning outcomes in primary school children with emotional problems: a prospective cohort study.

BACKGROUND: Academic difficulties are common in adolescents with mental health problems. Although earlier childhood emotional problems, characterised by heightened anxiety and depressive symptoms are common forerunners to adolescent mental health problems, the degree to which mental health problems in childhood may contribute independently to academic difficulties has been little explored. METHODS: Data were drawn from a prospective cohort study of students in Melbourne, Australia (N = 1239). Data were linked with a standardised national assessment of academic performance at baseline (9 years) and wave three (11 years). Depressive and anxiety symptoms were assessed at baseline and wave two (10 years). Regression analyses estimated the association between emotional problems (9 and/or 10 years) and academic performance at 11 years, adjusting for baseline academic performance, sex, age and socioeconomic status, and hyperactivity/inattention symptoms. RESULTS: Students with depressive symptoms at 9 years of age had lost nearly 4 months of numeracy learning two years later after controlling for baseline academic performance and confounders. Results were similar for anxiety symptoms. Regardless of when depressive symptoms occurred there were consistent associations with poorer numeracy performance at 11 years. The association of depressive symptoms with reading performance was weaker than for numeracy if they were present at wave two. Persistent anxiety symptoms across two waves led to nearly a 4 month loss of numeracy learning at 11 years, but the difference was not meaningful for reading. Findings were similar when including hyperactivity/inattention symptoms. CONCLUSIONS: Childhood anxiety and depression are not only forerunners of later mental health problems but predict academic achievement. Partnerships between education and health systems have the potential to not only improve childhood emotional problems but also improve learning.

Impact of early intervention on the population prevalence of common mental disorders: 20-year prospective study.

BACKGROUND: The potential for early interventions to reduce the later prevalence of common mental disorders (CMD) first experienced in adolescence is unclear. AIMS: To examine the course of CMD and evaluate the extent to which the prevalence of CMD could be reduced by preventing adolescent CMD, or by intervening to change four young adult processes, between the ages of 20 and 29 years, that could be mediating the link between adolescent and adult disorder. METHOD: This was a prospective cohort study of 1923 Australian participants assessed repeatedly from adolescence (wave 1, mean age 14 years) to adulthood (wave 10, mean age 35 years). Causal mediation analysis was undertaken to evaluate the extent to which the prevalence of CMD at age 35 years in those with adolescent CMD could be reduced by either preventing adolescent CMD, or by intervening on four young adult mediating processes: the occurrence of young adult CMD, frequent cannabis use, parenting a child by age 24 years, and engagement in higher education and employment. RESULTS: At age 35, 19.2% of participants reported CMD; a quarter of these participants experienced CMD during both adolescence and young adulthood. In total, 49% of those with CMD during both adolescence and young adulthood went on to report CMD at age 35 years. Preventing adolescent CMD reduced the population prevalence at age 35 years by 3.9%. Intervening on all four young adult processes among those with adolescent CMD, reduced this prevalence by 1.6%. CONCLUSIONS: In this Australian cohort, a large proportion of adolescent CMD resolved by adulthood, and by age 35 years, the largest proportion of CMD emerged among individuals without prior CMD. Time-limited, early intervention in those with earlier adolescent disorder is unlikely to substantially reduce the prevalence of CMD in midlife.

Multiple imputation approaches for handling incomplete three-level data with time-varying cluster-memberships.

Three-level data arising from repeated measures on individuals clustered within higher-level units are common in medical research. A complexity arises when individuals change clusters over time, resulting in a cross-classified data structure. Missing values in these studies are commonly handled via multiple imputation (MI). If the three-level, cross-classified structure is modeled in the analysis, it also needs to be accommodated in the imputation model to ensure valid results. While incomplete three-level data can be handled using various approaches within MI, the performance of these in the cross-classified data setting remains unclear. We conducted simulations under a range of scenarios to compare these approaches in the context of an acute-effects cross-classified random effects substantive model, which models the time-varying cluster membership via simple additive random effects. The simulation study was based on a case study in a longitudinal cohort of students clustered within schools. We evaluated methods that ignore the time-varying cluster memberships by taking the first or most common cluster for each individual; pragmatic extensions of single- and two-level MI approaches within the joint modeling (JM) and the fully conditional specification (FCS) frameworks, using dummy indicators (DI) and/or imputing repeated measures in wide format to account for the cross-classified structure; and a three-level FCS MI approach developed specifically for cross-classified data. Results indicated that the FCS implementations performed well in terms of bias and precision while JM approaches performed poorly. Under both frameworks approaches using the DI extension should be used with caution in the presence of sparse data.

Evaluation of multiple imputation approaches for handling missing covariate information in a case-cohort study with a binary outcome.

BACKGROUND: In case-cohort studies a random subcohort is selected from the inception cohort and acts as the sample of controls for several outcome investigations. Analysis is conducted using only the cases and the subcohort, with inverse probability weighting (IPW) used to account for the unequal sampling probabilities resulting from the study design. Like all epidemiological studies, case-cohort studies are susceptible to missing data. Multiple imputation (MI) has become increasingly popular for addressing missing data in epidemiological studies. It is currently unclear how best to incorporate the weights from a case-cohort analysis in MI procedures used to address missing covariate data. METHOD: A simulation study was conducted with missingness in two covariates, motivated by a case study within the Barwon Infant Study. MI methods considered were: using the outcome, a proxy for weights in the simple case-cohort design considered, as a predictor in the imputation model, with and without exposure and covariate interactions; imputing separately within each weight category; and using a weighted imputation model. These methods were compared to a complete case analysis (CCA) within the context of a standard IPW analysis model estimating either the risk or odds ratio. The strength of associations, missing data mechanism, proportion of observations with incomplete covariate data, and subcohort selection probability varied across the simulation scenarios. Methods were also applied to the case study. RESULTS: There was similar performance in terms of relative bias and precision with all MI methods across the scenarios considered, with expected improvements compared with the CCA. Slight underestimation of the standard error was seen throughout but the nominal level of coverage (95%) was generally achieved. All MI methods showed a similar increase in precision as the subcohort selection probability increased, irrespective of the scenario. A similar pattern of results was seen in the case study. CONCLUSIONS: How weights were incorporated into the imputation model had minimal effect on the performance of MI; this may be due to case-cohort studies only having two weight categories. In this context, inclusion of the outcome in the imputation model was sufficient to account for the unequal sampling probabilities in the analysis model.

Association Between Earlier Introduction of Peanut and Prevalence of Peanut Allergy in Infants in Australia.

Importance: Randomized clinical trials showed that earlier peanut introduction can prevent peanut allergy in select high-risk populations. This led to changes in infant feeding guidelines in 2016 to recommend early peanut introduction for all infants to reduce the risk of peanut allergy. Objective: To measure the change in population prevalence of peanut allergy in infants after the introduction of these new guidelines and evaluate the association between early peanut introduction and peanut allergy. Design: Two population-based cross-sectional samples of infants aged 12 months were recruited 10 years apart using the same sampling frame and methods to allow comparison of changes over time. Infants were recruited from immunization centers around Melbourne, Australia. Infants attending their 12-month immunization visit were eligible to participate (eligible age range, 11-15 months), regardless of history of peanut exposure or allergy history. Exposures: Questionnaires collected data on demographics, food allergy risk factors, peanut introduction, and reactions. Main Outcome and Measures: All infants underwent skin prick tests to peanut and those with positive results underwent oral food challenges. Prevalence estimates were standardized to account for changes in population demographics over time. Results: This study included 7209 infants (1933 in 2018-2019 and 5276 in 2007-2011). Of the participants in the older vs more recent cohort, 51.8% vs 50.8% were male; median (IQR) ages were 12.5 (12.2-13.0) months vs 12.4 (12.2-12.9) months. There was an increase in infants of East Asian ancestry over time (16.5% in 2018-2019 vs 10.5% in 2007-2011), which is a food allergy risk factor. After standardizing for infant ancestry and other demographics changes, peanut allergy prevalence was 2.6% (95% CI, 1.8%-3.4%) in 2018-2019, compared with 3.1% in 2007-2011 (difference, -0.5% [95% CI, -1.4% to 0.4%]; P = .26). Earlier age of peanut introduction was significantly associated with a lower risk of peanut allergy among infants of Australian ancestry in 2018-2019 (age 12 months compared with age 6 months or younger: adjusted odds ratio, 0.08 [05% CI, 0.02-0.36]; age 12 months compared with 7 to less than 10 months: adjusted odds ratio, 0.09 [95% CI, 0.02-0.53]), but not significant among infants of East Asian ancestry (P for interaction = .002). Conclusions and Relevance: In cross-sectional analyses, introduction of a guideline recommending early peanut introduction in Australia was not associated with a statistically significant lower or higher prevalence of peanut allergy across the population.

Evaluation of approaches for accommodating interactions and non-linear terms in multiple imputation of incomplete three-level data.

Three-level data structures arising from repeated measures on individuals clustered within larger units are common in health research studies. Missing data are prominent in such studies and are often handled via multiple imputation (MI). Although several MI approaches can be used to account for the three-level structure, including adaptations to single- and two-level approaches, when the substantive analysis model includes interactions or quadratic effects, these too need to be accommodated in the imputation model. In such analyses, substantive model compatible (SMC) MI has shown great promise in the context of single-level data. Although there have been recent developments in multilevel SMC MI, to date only one approach that explicitly handles incomplete three-level data is available. Alternatively, researchers can use pragmatic adaptations to single- and two-level MI approaches, or two-level SMC-MI approaches. We describe the available approaches and evaluate them via simulations in the context of three three-level random effects analysis models involving an interaction between the incomplete time-varying exposure and time, an interaction between the time-varying exposure and an incomplete time-fixed confounder, or a quadratic effect of the exposure. Results showed that all approaches considered performed well in terms of bias and precision when the target analysis involved an interaction with time, but the three-level SMC MI approach performed best when the target analysis involved an interaction between the time-varying exposure and an incomplete time-fixed confounder, or a quadratic effect of the exposure. We illustrate the methods using data from the Childhood to Adolescence Transition Study.

Multiple imputation for handling missing outcome data in randomized trials involving a mixture of independent and paired data.

Randomized trials involving independent and paired observations occur in many areas of health research, for example in paediatrics, where studies can include infants from both single and twin births. Multiple imputation (MI) is often used to address missing outcome data in randomized trials, yet its performance in trials with independent and paired observations, where design effects can be less than or greater than one, remains to be explored. Using simulated data and through application to a trial dataset, we investigated the performance of different methods of MI for a continuous or binary outcome when followed by analysis using generalized estimating equations to account for clustering due to the pairs. We found that imputing data separately for independent and paired data, with paired data imputed in wide format, was the best performing MI method, producing unbiased point and standard error estimates for the treatment effect throughout. Ignoring clustering in the imputation model performed well in settings where the design effect due to the inclusion of paired data was close to one, but otherwise led to moderately biased variance estimates. Including a random cluster effect in the imputation model led to slightly biased point estimates for binary outcome data and variance estimates that were too small in some settings. Based on these results, we recommend researchers impute independent and paired data separately where feasible to do so. The exception is if the design effect due to the inclusion of paired data is close to one, where ignoring clustering may be appropriate.

Multiple imputation of semi-continuous exposure variables that are categorized for analysis.

Semi-continuous variables are characterized by a point mass at one value and a continuous range of values for remaining observations. An example is alcohol consumption quantity, with a spike of zeros representing non-drinkers and positive values for drinkers. If multiple imputation is used to handle missing values for semi-continuous variables, it is unclear how this should be implemented within the standard approaches of fully conditional specification (FCS) and multivariate normal imputation (MVNI). This question is brought into focus by the use of categorized versions of semi-continuous exposure variables in analyses (eg, no drinking, drinking below binge level, binge drinking, heavy binge drinking), raising the question of how best to achieve congeniality between imputation and analysis models. We performed a simulation study comparing nine approaches for imputing semi-continuous exposures requiring categorization for analysis. Three methods imputed the categories directly: ordinal logistic regression, and imputation of binary indicator variables representing the categories using MVNI (with two variants). Six methods (predictive mean matching, zero-inflated binomial imputation, and two-part imputation methods with variants in FCS and MVNI) imputed the semi-continuous variable, with categories derived after imputation. The ordinal and zero-inflated binomial methods had good performance across most scenarios, while MVNI methods requiring rounding after imputation did not perform well. There were mixed results for predictive mean matching and the two-part methods, depending on whether the estimands were proportions or regression coefficients. The results highlight the need to consider the parameter of interest when selecting an imputation procedure.

Social networking and symptoms of depression and anxiety in early adolescence.

BACKGROUND: Use of social networking in later childhood and adolescence has risen quickly. The consequences of these changes for mental health are debated but require further empirical evaluation. METHODS: Using data from the Childhood to Adolescence Transition Study (n = 1,156), duration of social networking use was measured annually at four time points from 11.9 to 14.8 years of age (≥1 h/day indicating high use). Cross-sectional and prospective relationships between social networking use and depressive and anxiety symptoms were examined. RESULTS: In adjusted (age, socioeconomic status, prior mental health history) cross-sectional analyses, females with high social networking use had greater odds of depressive (odds ratio [OR]: 2.15; 95% confidence interval [CI]: 1.58-2.91) and anxiety symptoms (OR: 1.99; 95% CI: 1.32-3.00) than those that used a few minutes at most, while males with high social networking use had 1.60 greater odds of reporting depressive symptoms (95% CI: 1.09-2.35). For females, an increased odds of depressive symptoms at age 14.8 was observed for high social networking use at one previous wave and at two or three previous waves, even after adjustment (OR: 1.76; 95% CI: 1.11-2.78; OR: 2.06, 95% CI: 1.27-3.37, respectively) compared to no wave of high use. CONCLUSIONS: Our results suggest weak to moderate increased odds of depression and anxiety in girls and boys with high social networking use versus low/normal use. These findings indicate that prevention programs for early mental health problems might benefit from targeting social networking use in early adolescence.

Preconception depression and anxiety symptoms and maternal-infant bonding: a 20-year intergenerational cohort study.

Early maternal-infant bonding problems are often forerunners of later emotional and behavioural difficulties. Interventions typically target the perinatal period but many risks may be established well before pregnancy. Here we examine the extent to which adolescent and young adult depression and anxiety symptoms predict perinatal maternal-infant bonding difficulties. The Victorian Intergenerational Health Cohort Study (VIHCS, est. 2006) is following offspring born to the Victorian Adolescent Health Cohort Study (VAHCS; est. 1992). VAHCS participants were assessed for depression and anxiety symptoms nine times during adolescence and young adulthood (age 14-29 years), and then contacted bi-annually (from age 29-35 years) to identify pregnancies. The Postpartum Bonding Questionnaire (PBQ) was administered to mothers at 2 and 12 months postpartum. A total of 395 women (606 infants) completed the 2-month and/or 12-month postpartum interviews. For most infants (64%), mothers had experienced depression and/or anxiety before pregnancy. Preconception depression and anxiety symptoms that persisted from adolescence into young adulthood predicted maternal-infant bonding problems at 2 months (ß = 0.30, 95% CI 0.04, 0.55) and 12 months postpartum (ß = 0.40, 95% CI 0.16, 0.63). Depression and anxiety symptoms occurring in young adulthood only, also predicted bonding problems at 12 months postpartum (ß = 0.37, 95% CI 0.02, 0.71). Associations between preconception depression and anxiety symptoms and anxiety-related maternal-infant bonding problems at 12 months postpartum remained after adjustment for antenatal and concurrent postpartum depressive symptoms. This study puts forward a case for extending preconception health care beyond contraception and nutrition to a broader engagement in supporting the mental health of young women from adolescence.

Multiple imputation methods for handling missing values in longitudinal studies with sampling weights: Comparison of methods implemented in Stata.

Many analyses of longitudinal cohorts require incorporating sampling weights to account for unequal sampling probabilities of participants, as well as the use of multiple imputation (MI) for dealing with missing data. However, there is no guidance on how MI and sampling weights should be implemented together. We simulated a target population based on the Australian Bureau of Statistics Estimated Resident Population and drew 1000 random samples dependent on three design variables to mimic the Longitudinal Study of Australian Children. The target analysis was the weighted prevalence of overweight/obesity over childhood. We evaluated the performance of several MI approaches available in Stata, based on multivariate normal imputation (MVNI), fully conditional specification (FCS) and twofold FCS: a weighted imputation model, imputing missing data separately for each quintile sampling weight grouping, including the design stratum indicator in the imputation model, and using sampling weights as a covariate in the imputation model. Approaches based on available cases and inverse probability weighting (IPW), with time-varying weights, were also compared. We observed severe issues of convergence with FCS and twofold FCS. All MVNI-based approaches performed similarly, producing minimal bias and nominal coverage, except for when imputation was conducted separately for each quintile sampling weight group. IPW performed equally as well as MVNI-based approaches in terms of bias, however, was less precise. In similar longitudinal studies, we recommend using MVNI with the design stratum as a covariate in the imputation model. If this is unknown, including the sampling weight as a covariate is an appropriate alternative.

Explaining the link between adiposity and colorectal cancer risk in men and postmenopausal women in the UK Biobank: A sequential causal mediation analysis.

Mechanisms underlying adiposity-colorectal cancer (CRC) association are incompletely understood. Using UK Biobank data, we investigated the role of C-reactive protein (CRP), hemoglobin-A1c (HbA1c) and (jointly) sex hormone-binding globulin (SHBG) and testosterone, in explaining this association. Total effect of obesity versus normal-weight (based on waist circumference, body mass index, waist-hip ratio) on CRC risk was decomposed into natural direct (NDE) and indirect (NIE) effects using sequential mediation analysis. After a median follow-up of 7.1 years, 2070 incident CRC cases (men = 1,280; postmenopausal women = 790) were recorded. For men, the adjusted risk ratio (RR) for waist circumference (≥102 vs. ≤94 cm) was 1.37 (95% confidence interval [CI], 1.19-1.58). The RRsNIE were 1.08 (95% CI: 1.01-1.16) through all biomarkers, 1.06 (95% CI: 1.01-1.11) through pathways influenced by CRP, 0.99 (95% CI: 0.97-1.01) through HbA1c beyond (the potential influence of) CRP and 1.03 (95% CI: 0.99-1.08) through SHBG and testosterone combined beyond CRP and HbA1c. The RRNDE was 1.26 (95% CI: 1.09-1.47). For women, the RR for waist circumference (≥88 vs. ≤80 cm) was 1.27 (95% CI: 1.07-1.50). The RRsNIE were 1.08 (95% CI: 0.94-1.22) through all biomarkers, 1.08 (95% CI: 0.99-1.17) through CRP, 1.00 (95% CI: 0.98-1.02) through HbA1c beyond CRP and 1.00 (95% CI: 0.92-1.09) through SHBG and testosterone combined beyond CRP and HbA1c. The RRNDE was 1.18 (95% CI: 0.96-1.45). For men and women, pathways influenced by CRP explained a small proportion of the adiposity-CRC association. Testosterone and SHBG also explained a small proportion of this association in men. These results suggest that pathways marked by these obesity-related factors may not explain a large proportion of the adiposity-CRC association.

Adiposity and estrogen receptor-positive, postmenopausal breast cancer risk: Quantification of the mediating effects of fasting insulin and free estradiol.

Adiposity increases estrogen receptor (ER)-positive postmenopausal breast cancer risk. While mechanisms underlying this relationship are uncertain, dysregulated sex-steroid hormone production and insulin signaling are likely pathways. Our aim was to quantify mediating effects of fasting insulin and free estradiol in the adiposity and ER-positive postmenopausal breast cancer association. We used data from a case-cohort study of sex hormones and insulin signaling nested within the Melbourne Collaborative Cohort Study. Eligible women, at baseline, were not diagnosed with cancer, were postmenopausal, did not use hormone therapy and had no history of diabetes or diabetes medication use. Women with ER-negative disease or breast cancer diagnosis within the first follow-up year were excluded. We analyzed the study as a cumulative sampling case-control study with 149 cases and 1,029 controls. Missing values for insulin and free estradiol were multiply imputed with chained equations. Interventional direct (IDE) and indirect (IIE) effects were estimated using regression-based multiple-mediator approach. For women with body mass index (BMI) >30 kg/m2 compared to women with BMI 18.5-25 kg/m2 , the risk ratio (RR) of breast cancer was 1.75 (95% confidence interval [CI] 1.05-2.91). The estimated IDE (RR) not through the mediators was 1.03 (95% CI 0.43-2.48). Percentage mediated effect through free estradiol was 72% (IIE-RR 1.56; 95% CI 1.11-2.19). There was no evidence for an indirect effect through insulin (IIE-RR 1.12; 95% CI 0.68-1.84; 28% mediated). Our results suggest that circulating free estradiol plays an important mediating role in the adiposity-breast cancer relationship but does not explain all of the association.