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.

Longitudinal Aging Study Amsterdam COVID-19 exposure index: a cross-sectional analysis of the impact of the pandemic on daily functioning of older adults.

OBJECTIVES: The aim of this study was to develop an index to measure older adults' exposure to the COVID-19 pandemic and to study its association with various domains of functioning. DESIGN: Cross-sectional study. SETTING: The Longitudinal Aging Study Amsterdam (LASA), a cohort study in the Netherlands. PARTICIPANTS: Community-dwelling older adults aged 62-102 years (n=1089) who participated in the LASA COVID-19 study (June-September 2020), just after the first wave of the pandemic. PRIMARY OUTCOME MEASURES: A 35-item COVID-19 exposure index with a score ranging between 0 and 1 was developed, including items that assess the extent to which the COVID-19 situation affected daily lives of older adults. Descriptive characteristics of the index were studied, stratified by several sociodemographic factors. Logistic regression analyses were performed to study associations between the exposure index and several indicators of functioning (functional limitations, anxiety, depression and loneliness). RESULTS: The mean COVID-19 exposure index score was 0.20 (SD 0.10). Scores were relatively high among women and in the southern region of the Netherlands. In models adjusted for sociodemographic factors and prepandemic functioning (2018-2019), those with scores in the highest tertile of the exposure index were more likely to report functional limitations (OR: 2.24; 95% CI: 1.48 to 3.38), anxiety symptoms (OR: 3.14; 95% CI: 1.82 to 5.44), depressive symptoms (OR: 2.49; 95% CI: 1.55 to 4.00) and loneliness (OR: 2.97; 95% CI: 2.08 to 4.26) than those in the lowest tertile. CONCLUSIONS: Among older adults in the Netherlands, higher exposure to the COVID-19 pandemic was associated with worse functioning in the physical, mental and social domain. The newly developed exposure index may be used to identify persons for whom targeted interventions are needed to maintain or improve functioning during the pandemic or postpandemic.

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.

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.

Older adults report cancellation or avoidance of medical care during the COVID-19 pandemic: results from the Longitudinal Aging Study Amsterdam.

PURPOSE: Delay of routine medical care during the COVID-19 pandemic may have serious consequences for the health and functioning of older adults. The aim of this study was to investigate whether older adults reported cancellation or avoidance of medical care during the first months of the COVID-19 pandemic, and to explore associations with health and socio-demographic characteristics. METHODS: Cross-sectional data of 880 older adults aged ≥ 62 years (mean age 73.4 years, 50.3% female) were used from the COVID-19 questionnaire of the Longitudinal Aging Study Amsterdam, a cohort study among community-dwelling older adults in the Netherlands. Cancellation and avoidance of care were assessed by self-report, and covered questions on cancellation of primary care (general practitioner), cancellation of hospital outpatient care, and postponed help-seeking. Respondent characteristics included age, sex, educational level, loneliness, depression, anxiety, frailty, multimorbidity and information on quarantine. RESULTS: 35% of the sample reported cancellations due to the COVID-19 situation, either initiated by the respondent (12%) or by healthcare professionals (29%). Postponed help-seeking was reported by 8% of the sample. Multimorbidity was associated with healthcare-initiated cancellations (primary care OR = 1.92, 95% CI = 1.09-3.50; hospital OR = 1.86, 95% CI = 1.28-2.74) and respondent-initiated hospital outpatient cancellations (OR = 2.02, 95% CI = 1.04-4.12). Depressive symptoms were associated with postponed help-seeking (OR = 1.15, 95% CI = 1.06-1.24). CONCLUSION: About one third of the study sample reported cancellation or avoidance of medical care during the first months of the pandemic, and this was more common among those with multiple chronic conditions. How this impacts outcomes in the long term should be investigated in future research.

Predictors of Frailty and Vitality in Older Adults Aged 75 years and Over: Results from the Longitudinal Aging Study Amsterdam.

INTRODUCTION: Frailty can be seen as a continuum, from fit to frail. While many recent studies have focused on frailty, much less attention has been paid to the other end of the continuum: the group of older adults that remain (relatively) vital. Moreover, there is a lack of studies on frailty and vitality that investigate predictors from multiple domains of functioning simultaneously. The aim of this study was to identify predictors of frailty as well as vitality among older adults aged 75 years and over. METHODS: We used longitudinal data from 569 adults aged ≥75 years who participated in the Longitudinal Aging Study Amsterdam. Predictors from the sociodemographic, social, psychological, lifestyle, and physical domains of functioning were measured at T1 (2008-2009). We used the frailty index (FI) to identify frail (FI ≥ 0.25) and vital (FI ≤ 0.15) respondents at follow-up, 3 years later (T2: 2011-2012). We conducted logistic regression analyses with backward stepwise selection to develop and internally validate our prediction models. RESULTS: The prevalence of frailty in our sample at follow-up was 49.4% and the prevalence of vitality was 18.3%. Predictors of frailty and vitality partly overlapped and included age, depressive symptoms, number of chronic diseases, and self-rated health. We also found predictors that did not overlap. Male sex, moderate alcohol use, more emotional support received, and no hearing problems, were predictors of vitality. Lower cognitive functioning, polypharmacy, and pain were predictors of frailty. The final model for vitality explained 42% of the variance and the final model for frailty explained 48%. Both models had a good discriminative value (area under ROC-curve [AUC] vitality: 0.88; AUC frailty: 0.85). CONCLUSION: Among older adults aged 75 years and over, predictors of frailty only partially overlap with predictors of vitality. The readily accessible predictors in our models may help to identify older adults who are likely to be vital, or who are at risk of frailty.

Explaining the association between frailty and mortality in older adults: The mediating role of lifestyle, social, psychological, cognitive, and physical factors.

Frailty is associated with a higher risk of mortality, but not much is known about underlying pathways of the frailty-mortality association. In this study, we explore a wide range of possible mediators of the relation between frailty and mortality. Data were used from the Longitudinal Aging Study Amsterdam (LASA). We included 1477 older adults aged 65 years and over who participated in the study in 2008-2009 and linked their data to register data on mortality up to 2015. We examined a range of lifestyle, social, psychological, cognitive, and physical factors as potential mediators. All analyses were stratified by sex. We used causal mediation analyses to estimate the indirect effects in single-mediator analyses. Statistically significant mediators were then included in multiple-mediator analyses to examine their combined effect. The results showed that older men (OR = 2.79, 95% CI = 1.23;6.34) and women (OR = 2.31, 95% CI = 1.24;4.30) with frailty had higher odds of being deceased 6 years later compared to those without frailty. In men, polypharmacy (indirect effect OR = 1.21, 95% CI = 1.03;1.50) was a statistically significant mediator in this association. In women, polypharmacy, self-rated health, and multimorbidity were statistically significant mediators in the single-mediator models, but only the indirect effect of polypharmacy remained in the multiple-mediator model (OR = 1.16, 95% CI = 1.03;1.38). In conclusion, of many factors that were considered, we identified polypharmacy as explanatory factor of the association between frailty and mortality in older men and women. This finding has important clinical implications, as it suggests that targeting polypharmacy in frail older adults could reduce their risk of mortality.

Intention-to-treat analysis when only a baseline value is available.

OBJECTIVES: How to perform an intention to treat (ITT) analysis when a patient has a baseline value but no follow-up measurements is problematic. The purpose of this study was to compare different methods that deal with this problem, i.e. no imputation (standard and alternative mixed model analysis), single imputation (i.e. baseline value carried forward), and multiple imputation (selective and non-selective). STUDY DESIGN AND SETTING: We used a simulation study with different scenarios regarding 1) the association between missingness and the baseline value, 2) whether the patients did or did not receive the treatment, and 3) the percentage of missing data, and two real life data sets. RESULTS: Bias and coverage were comparable between the two mixed model analyses and multiple imputation in most situations including the real life data examples. Only in the situation when the patients in the treatment group were simulated not to have received the treatment, selective imputation using this information outperformed all other methods. CONCLUSIONS: In most situations a standard mixed model analysis without imputation is appropriate as ITT analysis. However, when patients with missing follow-up data allocated to the treatment group did not received treatment, it is advised to use selective imputation, using this information, although the results should be interpreted with caution.

Frailty Combined with Loneliness or Social Isolation: An Elevated Risk for Mortality in Later Life.

BACKGROUND/OBJECTIVES: Frailty, loneliness, and social isolation are all associated with adverse outcomes in older adults, but little is known about their combined impact on mortality. DESIGN: Prospective cohort study. SETTING: The Longitudinal Aging Study Amsterdam. PARTICIPANTS: Community-dwelling older adults aged 65 and older (n = 1,427). MEASUREMENTS: Frailty was measured with the frailty phenotype (Fried criteria). Loneliness was assessed with the De Jong Gierveld Loneliness Scale. Social isolation was operationalized using information on partner status, social support, and network size. Two categorical variables were created, for each possible combination regarding frailty and loneliness (FL) and frailty and social isolation (FS), respectively. Mortality was monitored over a period of 22 years (1995-2017). Survival curves and Cox proportional hazard models were used to study the effects of the FL and FS combinations on mortality. Analyses were adjusted for sociodemographic factors, depression, chronic diseases, and smoking. RESULTS: Frailty prevalence was 13%, and 5.9% of the sample were frail and lonely, and 6.2% frail and socially isolated. In fully adjusted models, older adults who were only frail had a higher risk of mortality compared with people without any of the conditions (hazard ratio [HR] range = 1.40-1.48; P < .01). However, the highest risk of mortality was observed in people with a combined presence of frailty and loneliness or social isolation (HRFL = 1.83; 95% confidence interval [CI] = 1.42-2.37; HRFS = 1.77; 95% CI = 1.36-2.30). Sensitivity analyses using a frailty index based on the deficit accumulation approach instead of the frailty phenotype showed similar results, confirming the robustness of our findings. CONCLUSION: Frail older adults are at increased risk of mortality, but this risk is even higher for those who are also lonely or socially isolated. To optimize well-being and health outcomes in physically frail older adults, targeted interventions focusing on both subjective and objective social vulnerability are needed.

Ignoring competing events in the analysis of survival data may lead to biased results: a nonmathematical illustration of competing risk analysis.

OBJECTIVE: Competing events are often ignored in epidemiological studies. Conventional methods for the analysis of survival data assume independent or noninformative censoring, which is violated when subjects that experience a competing event are censored. Because many survival studies do not apply competing risk analysis, we explain and illustrate in a nonmathematical way how to analyze and interpret survival data in the presence of competing events. STUDY DESIGN AND SETTING: Using data from the Longitudinal Aging Study Amsterdam, both marginal analyses (Kaplan-Meier method and Cox proportional-hazards regression) and competing risk analyses (cumulative incidence function [CIF], cause-specific and subdistribution hazard regression) were performed. We analyzed the association between sex and depressive symptoms, in which death before the onset of depression was a competing event. RESULTS: The Kaplan-Meier method overestimated the cumulative incidence of depressive symptoms. Instead, the CIF should be used. As the subdistribution hazard model has a one-to-one relation with the CIF, it is recommended for prediction research, whereas the cause-specific hazard model is recommended for etiologic research. CONCLUSION: When competing risks are present, the type of research question guides the choice of the analytical model to be used. In any case, results should be presented for all event types.