The mediating role of health behaviors in the association between depression, anxiety and cancer incidence: an individual participant data meta-analysis.

BACKGROUND: Although behavioral mechanisms in the association among depression, anxiety, and cancer are plausible, few studies have empirically studied mediation by health behaviors. We aimed to examine the mediating role of several health behaviors in the associations among depression, anxiety, and the incidence of various cancer types (overall, breast, prostate, lung, colorectal, smoking-related, and alcohol-related cancers). METHODS: Two-stage individual participant data meta-analyses were performed based on 18 cohorts within the Psychosocial Factors and Cancer Incidence consortium that had a measure of depression or anxiety (N = 319 613, cancer incidence = 25 803). Health behaviors included smoking, physical inactivity, alcohol use, body mass index (BMI), sedentary behavior, and sleep duration and quality. In stage one, path-specific regression estimates were obtained in each cohort. In stage two, cohort-specific estimates were pooled using random-effects multivariate meta-analysis, and natural indirect effects (i.e. mediating effects) were calculated as hazard ratios (HRs). RESULTS: Smoking (HRs range 1.04-1.10) and physical inactivity (HRs range 1.01-1.02) significantly mediated the associations among depression, anxiety, and lung cancer. Smoking was also a mediator for smoking-related cancers (HRs range 1.03-1.06). There was mediation by health behaviors, especially smoking, physical inactivity, alcohol use, and a higher BMI, in the associations among depression, anxiety, and overall cancer or other types of cancer, but effects were small (HRs generally below 1.01). CONCLUSIONS: Smoking constitutes a mediating pathway linking depression and anxiety to lung cancer and smoking-related cancers. Our findings underline the importance of smoking cessation interventions for persons with depression or anxiety.

A novel approach to estimate moderated treatment effects and moderated mediated effects with continuous moderators.

Moderation analysis is used to study under what conditions or for which subgroups of individuals a treatment effect is stronger or weaker. When a moderator variable is categorical, such as assigned sex, treatment effects can be estimated for each group resulting in a treatment effect for males and a treatment effect for females. If a moderator variable is a continuous variable, a strategy for investigating moderated treatment effects is to estimate conditional effects (i.e., simple slopes) via the pick-a-point approach. When conditional effects are estimated using the pick-a-point approach, the conditional effects are often given the interpretation of "the treatment effect for the subgroup of individuals…." However, the interpretation of these conditional effects as subgroup effects is potentially misleading because conditional effects are interpreted at a specific value of the moderator variable (e.g., +1 SD above the mean). We describe a simple solution that resolves this problem using a simulation-based approach. We describe how to apply this simulation-based approach to estimate subgroup effects by defining subgroups using a range of scores on the continuous moderator variable. We apply this method to three empirical examples to demonstrate how to estimate subgroup effects for moderated treatment and moderated mediated effects when the moderator variable is a continuous variable. Finally, we provide researchers with both SAS and R code to implement this method for similar situations described in this paper. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Determinants of informal care time, distress, depression, and quality of life in care partners along the trajectory of Alzheimer's disease.

Introduction: We evaluated determinants associated with care partner outcomes along the Alzheimer's disease (AD) stages. Methods: We included n = 270 care partners of amyloid-positive patients in the pre-dementia and dementia stages of AD. Using linear regression analysis, we examined determinants of four care partner outcomes: informal care time, caregiver distress, depression, and quality of life (QoL). Results: More behavioral symptoms and functional impairment in patients were associated with more informal care time and depressive symptoms in care partners. More behavioral symptoms were related with more caregiver distress. Spouse care partners spent more time on informal care and QoL was lower in female care partners. Behavioral problems and subtle functional impairment of the patient predisposed for worse care partner outcomes already in the pre-dementia stages. Discussion: Both patient and care partner determinants contribute to the care partner outcomes, already in early disease stages. This study provides red flags for high care partner burden.

Misspecification of confounder-exposure and confounder-outcome associations leads to bias in effect estimates.

BACKGROUND: Confounding is a common issue in epidemiological research. Commonly used confounder-adjustment methods include multivariable regression analysis and propensity score methods. Although it is common practice to assess the linearity assumption for the exposure-outcome effect, most researchers do not assess linearity of the relationship between the confounder and the exposure and between the confounder and the outcome before adjusting for the confounder in the analysis. Failing to take the true non-linear functional form of the confounder-exposure and confounder-outcome associations into account may result in an under- or overestimation of the true exposure effect. Therefore, this paper aims to demonstrate the importance of assessing the linearity assumption for confounder-exposure and confounder-outcome associations and the importance of correctly specifying these associations when the linearity assumption is violated. METHODS: A Monte Carlo simulation study was used to assess and compare the performance of confounder-adjustment methods when the functional form of the confounder-exposure and confounder-outcome associations were misspecified (i.e., linearity was wrongly assumed) and correctly specified (i.e., linearity was rightly assumed) under multiple sample sizes. An empirical data example was used to illustrate that the misspecification of confounder-exposure and confounder-outcome associations leads to bias. RESULTS: The simulation study illustrated that the exposure effect estimate will be biased when for propensity score (PS) methods the confounder-exposure association is misspecified. For methods in which the outcome is regressed on the confounder or the PS, the exposure effect estimate will be biased if the confounder-outcome association is misspecified. In the empirical data example, correct specification of the confounder-exposure and confounder-outcome associations resulted in smaller exposure effect estimates. CONCLUSION: When attempting to remove bias by adjusting for confounding, misspecification of the confounder-exposure and confounder-outcome associations might actually introduce bias. It is therefore important that researchers not only assess the linearity of the exposure-outcome effect, but also of the confounder-exposure or confounder-outcome associations depending on the confounder-adjustment method used.

Statistical Mediation Analysis for Models with a Binary Mediator and a Binary Outcome: the Differences Between Causal and Traditional Mediation Analysis.

Mediation analysis is an important statistical method in prevention research, as it can be used to determine effective intervention components. Traditional mediation analysis defines direct and indirect effects in terms of linear regression coefficients. It is unclear how these traditional effects are estimated in settings with binary variables. An important recent methodological advancement in the mediation analysis literature is the development of the causal mediation analysis framework. Causal mediation analysis defines causal effects as the difference between two potential outcomes. These definitions can be applied to any mediation model to estimate natural direct and indirect effects, including models with binary variables and an exposure-mediator interaction. This paper aims to clarify the similarities and differences between the causal and traditional effect estimates for mediation models with a binary mediator and a binary outcome. Causal and traditional mediation analyses were applied to an empirical example to demonstrate these similarities and differences. Causal and traditional mediation analysis provided similar controlled direct effect estimates, but different estimates of the natural direct effects, natural indirect effects, and total effect. Traditional mediation analysis methods do not generalize well to mediation models with binary variables, while the natural effect definitions can be applied to any mediation model. Causal mediation analysis is therefore the preferred method for the analysis of mediation models with binary variables.

Causal mediation effects in single case experimental designs.

Single case experimental designs (SCEDs) are used to test treatment effects in a wide range of fields and consist of repeated measurements for a single case throughout one or more baseline phases and throughout one or more treatment phases. Recently, mediation analysis has been applied to SCEDs. Mediation analysis decomposes the total treatment-outcome effect into a direct and indirect effect, and therefore aims to unravel the causal processes underlying treatment-outcome effects. The most recent methodological advancement for mediation analysis is the development of causal mediation analysis methodology which clarifies the necessary causal assumptions for mediation analysis. The goal of this article is to derive the causal mediation effects and corresponding standard errors based on piecewise linear regression models for the mediator and outcome and to evaluate the performance of these regression estimators and standard errors. Whereas previous studies estimated the direct and indirect effects as either the change in level or change in trend, we showed that the causal direct and indirect effects incorporate both the change in level and change in trend. Based on our simulation study we showed that for the causal indirect effects, Monte Carlo confidence intervals provided accurate (i.e., p = .05) Type I error rates and higher statistical power than normal theory confidence intervals. For the causal direct effects and total effect, normal theory confidence intervals provided accurate Type I error rates and higher statistical power than the Monte Carlo confidence intervals. Limitations and future directions are discussed. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Life satisfaction across the entire trajectory of Alzheimer's disease: A mediation analysis.

Introduction: We studied life satisfaction across Alzheimer's disease (AD) stages and studied mobility and meaningful activities as mediators of the associations between these AD stages and life satisfaction. Methods: In this cross-sectional study, we included n = 269 amyloid-positive patients with subjective cognitive decline (SCD), mild cognitive impairment (MCI), and AD dementia from the Amsterdam Dementia Cohort. Life satisfaction was measured with the satisfaction with life scale. The mediating role of transportation, work, sports, and hobbies on life satisfaction was examined in single and multiple mediator models. Results: Patients with dementia are less satisfied with life compared to SCD and MCI. These differences in life satisfaction are explained by reduced participation in meaningful activities, which in turn, was largely attributable to decreased transportation use. Discussion: Our findings suggest that improving access to transportation, therewith allowing participation in meaningful activities help to maintain life satisfaction and may be an important target for intervention.

A longitudinal study on quality of life along the spectrum of Alzheimer's disease.

BACKGROUND: Quality of life (QoL) is an important outcome from the perspective of patients and their caregivers, in both dementia and pre-dementia stages. Yet, little is known about the long-term changes in QoL over time. We aimed to compare the trajectories of QoL between amyloid-positive and amyloid-negative SCD or MCI patients and to evaluate QoL trajectories along the Alzheimer's disease (AD) continuum of cognitively normal to dementia. METHODS: We included longitudinal data of 447 subjective cognitive decline (SCD), 276 mild cognitive impairment (MCI), and 417 AD dementia patients from the Amsterdam Dementia Cohort. We compared QoL trajectories (EQ-5D and visual analog scale (VAS)) between (1) amyloid-positive and amyloid-negative SCD or MCI patients and (2) amyloid-positive SCD, MCI, and dementia patients with linear mixed-effect models. The models were adjusted for age, sex, Charlson Comorbidity Index (CCI), education, and EQ-5D scale (3 or 5 level). RESULTS: In SCD, amyloid-positive participants had a higher VAS at baseline but showed a steeper decline over time in EQ-5D and VAS than amyloid-negative participants. Also, in MCI, amyloid-positive patients had higher QoL at baseline but subsequently showed a steeper decline in QoL over time compared to amyloid-negative patients. When we compared amyloid-positive patients along the Alzheimer continuum, we found no difference between SCD, MCI, or dementia in baseline QoL, but QoL decreased at a faster rate in the dementia stage compared with the of SCD and MCI stages. CONCLUSIONS: QoL decreased at a faster rate over time in amyloid-positive SCD or MCI patients than amyloid-negative patients. QoL decreases over time along the entire AD continuum of SCD, MCI and dementia, with the strongest decrease in dementia patients. Knowledge of QoL trajectories is essential for the future evaluation of treatments in AD.

Time lags and time interactions in mixed effects models impacted longitudinal mediation effect estimates.

OBJECTIVES: Longitudinal mediation effects can be estimated with mixed effects models. Mixed effects models are versatile, as they accommodate the estimation of contemporaneous, lagged, time-independent, and time-dependent effects. However, the inclusion of time lags and time interactions in mixed effects models for longitudinal mediation analysis has received little attention. This article demonstrates how time lags and time interactions in mixed effects models affect the interpretation of longitudinal mediation effect estimates. STUDY DESIGN AND SETTING: We used a data example from the Amsterdam Growth and Health Longitudinal Study to illustrate how the inclusion of time lags and time interactions in mixed effects models may affect the size and interpretation of longitudinal mediation effect estimates. RESULTS: The chosen time lags between the determinant, mediator, and outcome influenced the size and interpretation of the mediation effect estimates. Furthermore, time interactions can be used to model linear or nonlinear development of the mediation effects over time. CONCLUSION: The inclusion of time lags and time interactions should be considered when estimating longitudinal mediation effects based on mixed effects models, as this enables the estimation of lagged and time-dependent effects.

Development of multivariable prediction models for institutionalization and mortality in the full spectrum of Alzheimer's disease.

BACKGROUND: Patients and caregivers express a desire for accurate prognostic information about time to institutionalization and mortality. Previous studies predicting institutionalization and mortality focused on the dementia stage. However, Alzheimer's disease (AD) is characterized by a long pre-dementia stage. Therefore, we developed prediction models to predict institutionalization and mortality along the AD continuum of cognitively normal to dementia. METHODS: This study included SCD/MCI patients (subjective cognitive decline (SCD) or mild cognitive impairment (MCI)) and patients with AD dementia from the Amsterdam Dementia Cohort. We developed internally and externally validated prediction models with biomarkers and without biomarkers, stratified by dementia status. Determinants were selected using backward selection (p<0.10). All models included age and sex. Discriminative performance of the models was assessed with Harrell's C statistics. RESULTS: We included n=1418 SCD/MCI patients (n=123 died, n=74 were institutionalized) and n=1179 patients with AD dementia (n=413 died, n=453 were institutionalized). For both SCD/MCI and dementia stages, the models for institutionalization and mortality included after backward selection clinical characteristics, imaging, and cerebrospinal fluid (CSF) biomarkers. In SCD/MCI, the Harrell's C-statistics of the models were 0.81 (model without biomarkers: 0.76) for institutionalization and 0.79 (model without biomarker: 0.76) for mortality. In AD-dementia, the Harrell's C-statistics of the models were 0.68 (model without biomarkers: 0.67) for institutionalization and 0.65 (model without biomarker: 0.65) for mortality. Models based on data from amyloid-positive patients only had similar discrimination. CONCLUSIONS: We constructed prediction models to predict institutionalization and mortality with good accuracy for SCD/MCI patients and moderate accuracy for patients with AD dementia. The developed prediction models can be used to provide patients and their caregivers with prognostic information on time to institutionalization and mortality along the cognitive continuum of AD.

Association of a Zero-Separation Neonatal Care Model With Stress in Mothers of Preterm Infants.

Importance: Active participation in care by parents and zero separation between parents and their newborns is highly recommended during infant hospitalization in the neonatal intensive care unit (NICU). Objective: To study the association of a family integrated care (FICare) model with maternal mental health at hospital discharge of their preterm newborn compared with standard neonatal care (SNC). Design, Setting, and Participants: This prospective, multicenter cohort study included mothers with infants born preterm treated in level-2 neonatal units in the Netherlands (1 unit with single family rooms [the FICare model] and 2 control sites with standard care in open bay units) between May 2017 and January 2020 as part of the AMICA study (fAMily Integrated CAre in the neonatal ward). Participants included mothers of preterm newborns admitted to participating units. Data analysis was performed from January to April 2021. Exposures: FICare model in single family rooms with complete couplet-care for the mother-newborn dyad during maternity and/or neonatal care. Main Outcomes and Measures: Maternal mental health, measured using the Parental Stress Scale: NICU (PSS-NICU). Secondary outcomes included survey scores on the Hospital Anxiety and Depression Scale, Postpartum Bonding Questionnaire, Perceived Maternal Parenting Self-efficacy Scale, and satisfaction with care (using EMPATHIC-N). Parent participation (using the CO-PARTNER tool) was assessed as a potential mediator of the association of the FICare model on outcomes with mediation analyses. Results: A total of 296 mothers were included; 124 of 141 mothers (87.9%) in the FICare model and 115 of 155 (74.2%) mothers in SNC responded to questionnaires (mean [SD] age: FICare, 33.3 [4.0] years; SNC, 33.3 [4.1] years). Mothers in the FICare model had lower total PSS-NICU stress scores at discharge (adjusted mean difference, -12.24; 95% CI, -18.44 to -6.04) than mothers in SNC, and specifically had lower scores for mother-newborn separation (adjusted mean difference, -1.273; 95% CI, -1.835 to -0.712). Mothers in the FICare model were present more (>8 hours per day: 105 of 125 [84.0%] mothers vs 42 of 115 [36.5%]; adjusted odds ratio, 19.35; 95% CI, 8.13 to 46.08) and participated more in neonatal care (mean [SD] score: 46.7 [6.9] vs 40.8 [6.7]; adjusted mean difference, 5.618; 95% CI, 3.705 to 7.532). Active parent participation was a significant mediator of the association between the FICare model and less maternal depression and anxiety (adjusted indirect effect, -0.133; 95% CI, -0.226 to -0.055), higher maternal self-efficacy (adjusted indirect effect, 1.855; 95% CI, 0.693 to 3.348), and better mother-newborn bonding (adjusted indirect effect, -0.169; 95% CI, -0.292 to -0.068). Conclusions and Relevance: The FICare model in our study was associated with less maternal stress at discharge; mothers were more present and participated more in the care for their newborn than in SNC, which was associated with improved maternal mental health outcomes. Future intervention strategies should aim at reducing mother-newborn separation and intensifying active parent participation in neonatal care. Trial Registration: Netherlands Trial Register identifier NL6175.

Trajectories of Frailty With Aging: Coordinated Analysis of Five Longitudinal Studies.

BACKGROUND AND OBJECTIVES: There is an urgent need to better understand frailty and its predisposing factors. Although numerous cross-sectional studies have identified various risk and protective factors of frailty, there is a limited understanding of longitudinal frailty progression. Furthermore, discrepancies in the methodologies of these studies hamper comparability of results. Here, we use a coordinated analytical approach in 5 independent cohorts to evaluate longitudinal trajectories of frailty and the effect of 3 previously identified critical risk factors: sex, age, and education. RESEARCH DESIGN AND METHODS: We derived a frailty index (FI) for 5 cohorts based on the accumulation of deficits approach. Four linear and quadratic growth curve models were fit in each cohort independently. Models were adjusted for sex/gender, age, years of education, and a sex/gender-by-age interaction term. RESULTS: Models describing linear progression of frailty best fit the data. Annual increases in FI ranged from 0.002 in the Invecchiare in Chianti cohort to 0.009 in the Longitudinal Aging Study Amsterdam (LASA). Women had consistently higher levels of frailty than men in all cohorts, ranging from an increase in the mean FI in women from 0.014 in the Health and Retirement Study cohort to 0.046 in the LASA cohort. However, the associations between sex/gender and rate of frailty progression were mixed. There was significant heterogeneity in within-person trajectories of frailty about the mean curves. DISCUSSION AND IMPLICATIONS: Our findings of linear longitudinal increases in frailty highlight important avenues for future research. Specifically, we encourage further research to identify potential effect modifiers or groups that would benefit from targeted or personalized interventions.

Association of a Family Integrated Care Model With Paternal Mental Health Outcomes During Neonatal Hospitalization.

Importance: During newborn hospitalization in the neonatal unit, fathers often feel anxious and excluded from their child's caregiving and decision-making. Few studies and interventions have focused on fathers' mental health and their participation in neonatal care. Objective: To study the association of a family integrated care (FICare) model (in single family rooms with complete couplet-care for the mother-newborn dyad) vs standard neonatal care (SNC) in open bay units with separate maternity care with mental health outcomes in fathers at hospital discharge of their preterm newborn and to study whether parent participation was a mediator of the association of the FICare model on outcomes. Design, Setting, and Participants: This prospective, multicenter cohort study was conducted from May 2017 to January 2020 as part of the fAMily Integrated Care in the Neonatal Ward Study, at level-2 neonatal units in the Netherlands (1 using the FICare model and 2 control sites using SNC). Participants included fathers of preterm newborns admitted to participating units. Data analysis was performed from January to April 2021. Exposure: FICare model in single family rooms with complete couplet-care for the mother-newborn dyad during maternity and/or neonatal care. Main Outcomes and Measures: Paternal mental health was measured using the Parental Stress Scale: NICU, Hospital Anxiety and Depression Scale, Post-partum Bonding Questionnaire, Perceived (Maternal) Parenting Self-efficacy Scale, and satisfaction with care (EMpowerment of PArents in THe Intensive Care-Neonatology). Parent participation (CO-PARTNER tool) was assessed as a potential mediator of the association of the FICare model with outcomes with mediation analyses (prespecified). Results: Of 309 families included in the fAMily Integrated Care in the Neonatal Ward Study, 263 fathers (85%) agreed to participate; 126 fathers were enrolled in FICare and 137 were enrolled in SNC. In FICare, 89 fathers (71%; mean [SD] age, 35.1 [4.8] years) responded to questionnaires and were analyzed. In SNC, 93 fathers (68%; mean [SD] age, 36.4 [5.5] years) responded to questionnaires and were analyzed. Fathers in FICare experienced less stress (adjusted ß, -10.02; 95% CI, -15.91 to -4.13; P = .001) and had higher participation scores (adjusted odds ratio, 3.424; 95% CI, 0.860 to 5.988; P = .009) compared with those in SNC. Participation mediated the beneficial association of the FICare model with fathers' depressive symptoms (indirect effect, -0.051; 95% CI, -0.133 to -0.003) and bonding with their newborns (indirect effect, -0.082; 95% CI, -0.177 to -0.015). Conclusions and Relevance: These findings suggest that the FICare model is associated with decreased paternal stress at discharge and enables fathers to be present and participate more than SNC, thus improving paternal mental health. Supporting fathers to actively participate in all aspects of newborn care should be encouraged regardless of architectural design of the neonatal unit.

Assessing the Robustness of Mediation Analysis Results Using Multiverse Analysis.

There is an increasing awareness that replication should become common practice in empirical studies. However, study results might fail to replicate for various reasons. The robustness of published study results can be assessed using the relatively new multiverse-analysis methodology, in which the robustness of the effect estimates against data analytical decisions is assessed. However, the uptake of multiverse analysis in empirical studies remains low, which might be due to the scarcity of guidance available on performing multiverse analysis. Researchers might experience difficulties in identifying data analytical decisions and in summarizing the large number of effect estimates yielded by a multiverse analysis. These difficulties are amplified when applying multiverse analysis to assess the robustness of the effect estimates from a mediation analysis, as a mediation analysis involves more data analytical decisions than a bivariate analysis. The aim of this paper is to provide an overview and worked example of the use of multiverse analysis to assess the robustness of the effect estimates from a mediation analysis. We showed that the number of data analytical decisions in a mediation analysis is larger than in a bivariate analysis. By using a real-life data example from the Longitudinal Aging Study Amsterdam, we demonstrated the application of multiverse analysis to a mediation analysis. This included the use of specification curves to determine the impact of data analytical decisions on the magnitude and statistical significance of the direct, indirect, and total effect estimates. Although the multiverse analysis methodology is still relatively new and future research is needed to further advance this methodology, this paper shows that multiverse analysis is a useful method for the assessment of the robustness of the direct, indirect, and total effect estimates in a mediation analysis and thereby to inform replication studies.

Modeling non-linear relationships in epidemiological data: The application and interpretation of spline models.

Objective: Traditional methods to deal with non-linearity in regression analysis often result in loss of information or compromised interpretability of the results. A recommended but underutilized method for modeling non-linear associations in regression models is spline functions. We explain spline functions in a non-mathematical way and illustrate the application and interpretation to an empirical data example. Methods: Using data from the Amsterdam Growth and Health Longitudinal Study, we examined the non-linear relationship between the sum of four skinfolds and VO2max, which are measures of body fat and cardiorespiratory fitness, respectively. We compared traditional methods (i.e., quadratic regression and categorization) to spline methods [1- and 3-knot linear spline (LSP) models and a 3-knot restricted cubic spline (RCS) model] in terms of the interpretability of the results and their explained variance (radj2). Results: The spline models fitted the data better than the traditional methods. Increasing the number of knots in the LSP model increased the explained variance (from radj2=0.578 for the 1-knot model to radj2=0.582 for the 3-knot model). The RCS model fitted the data best (radj2=0.591), but results in regression coefficients that are harder to interpret. Conclusion: Spline functions should be considered more often as they are flexible and can be applied in commonly used regression analysis. RCS regression is generally recommended for prediction research (i.e., to obtain the predicted outcome for a specific exposure value), whereas LSP regression is recommended if one is interested in the effects in a population.

Mediation analysis methods used in observational research: a scoping review and recommendations.

BACKGROUND: Mediation analysis methodology underwent many advancements throughout the years, with the most recent and important advancement being the development of causal mediation analysis based on the counterfactual framework. However, a previous review showed that for experimental studies the uptake of causal mediation analysis remains low. The aim of this paper is to review the methodological characteristics of mediation analyses performed in observational epidemiologic studies published between 2015 and 2019 and to provide recommendations for the application of mediation analysis in future studies. METHODS: We searched the MEDLINE and EMBASE databases for observational epidemiologic studies published between 2015 and 2019 in which mediation analysis was applied as one of the primary analysis methods. Information was extracted on the characteristics of the mediation model and the applied mediation analysis method. RESULTS: We included 174 studies, most of which applied traditional mediation analysis methods (n = 123, 70.7%). Causal mediation analysis was not often used to analyze more complicated mediation models, such as multiple mediator models. Most studies adjusted their analyses for measured confounders, but did not perform sensitivity analyses for unmeasured confounders and did not assess the presence of an exposure-mediator interaction. CONCLUSIONS: To ensure a causal interpretation of the effect estimates in the mediation model, we recommend that researchers use causal mediation analysis and assess the plausibility of the causal assumptions. The uptake of causal mediation analysis can be enhanced through tutorial papers that demonstrate the application of causal mediation analysis, and through the development of software packages that facilitate the causal mediation analysis of relatively complicated mediation models.

The use of traditional and causal estimators for mediation models with a binary outcome and exposure-mediator interaction.

An important recent development in mediation analysis is the use of causal mediation analysis. Causal mediation analysis decomposes the total exposure effect into causal direct and indirect effects in the presence of exposure-mediator interaction. However, in practice, traditional mediation analysis is still most widely used. The aim of this paper is to demonstrate the similarities and differences between the causal and traditional estimators for mediation models with a continuous mediator, a binary outcome, and exposure-mediator interaction. A real-life data example, analytical comparisons, and a simulation study were used to demonstrate the similarities and differences between the traditional and causal estimators. The causal and traditional estimators provide similar indirect effect estimates, but different direct and total effect estimates. Traditional mediation analysis may only be used when conditional direct effect estimates are of interest. Causal mediation analysis is the generally preferred method as its casual effect estimates help unravel causal mechanisms.

Noncollapsibility and its role in quantifying confounding bias in logistic regression.

BACKGROUND: Confounding bias is a common concern in epidemiological research. Its presence is often determined by comparing exposure effects between univariable- and multivariable regression models, using an arbitrary threshold of a 10% difference to indicate confounding bias. However, many clinical researchers are not aware that the use of this change-in-estimate criterion may lead to wrong conclusions when applied to logistic regression coefficients. This is due to a statistical phenomenon called noncollapsibility, which manifests itself in logistic regression models. This paper aims to clarify the role of noncollapsibility in logistic regression and to provide guidance in determining the presence of confounding bias. METHODS: A Monte Carlo simulation study was designed to uncover patterns of confounding bias and noncollapsibility effects in logistic regression. An empirical data example was used to illustrate the inability of the change-in-estimate criterion to distinguish confounding bias from noncollapsibility effects. RESULTS: The simulation study showed that, depending on the sign and magnitude of the confounding bias and the noncollapsibility effect, the difference between the effect estimates from univariable- and multivariable regression models may underestimate or overestimate the magnitude of the confounding bias. Because of the noncollapsibility effect, multivariable regression analysis and inverse probability weighting provided different but valid estimates of the confounder-adjusted exposure effect. In our data example, confounding bias was underestimated by the change in estimate due to the presence of a noncollapsibility effect. CONCLUSION: In logistic regression, the difference between the univariable- and multivariable effect estimate might not only reflect confounding bias but also a noncollapsibility effect. Ideally, the set of confounders is determined at the study design phase and based on subject matter knowledge. To quantify confounding bias, one could compare the unadjusted exposure effect estimate and the estimate from an inverse probability weighted model.

An examination of reciprocal effects between cardiovascular morbidity, depressive symptoms and loneliness over time in a longitudinal cohort of Dutch older adults.

BACKGROUND: Unidirectional studies suggest that the effects between cardiovascular disease, depressive symptoms and loneliness are reciprocal, but this has not been tested empirically. The aim was to study how cardiovascular morbidity, depressive symptoms and loneliness influence each other longitudinally. METHODS: Data from 2979 older adults from the Longitudinal Aging Study Amsterdam were analysed. Depressive symptoms (≥16 points on the Center for Epidemiologic Studies Depression Scale), loneliness (≥3 points on the De Jong Gierveld Loneliness Scale) and cardiovascular morbidity were measured five times during 13-year follow-up. With structural equation modelling, a full cross-lagged panel model was compared to nine nested models reflecting different sets of temporal effects. RESULTS: The best-fitting cross-lagged panel model showed reciprocal risk increasing effects between depressive symptoms and loneliness and a risk increasing effect of cardiovascular morbidity on depressive symptoms. LIMITATIONS: A cross-lagged panel model has technical limitations, such as that the chosen time lag may not be appropriate for each effect. In addition, differential loss to follow-up and collider bias may have led to an underestimation of the effects. CONCLUSIONS: Reciprocal effects tend to occur only between depressive symptoms and loneliness. Their interplay with cardiovascular morbidity seems more complex and mostly indirect, highlighting the potential of interventions to reduce depressive symptoms, loneliness and cardiovascular morbidity in concert to improve health at old age.