BackgroundConfounding 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.MethodsA 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.ResultsThe 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.ConclusionIn 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.
MULTIFILE
Objectives: Promoting unstructured outside play is a promising vehicle to increase children’s physical activity (PA). This study investigates if factors of the social environment moderate the relationship between the perceived physical environment and outside play. Study design: 1875 parents from the KOALA Birth Cohort Study reported on their child’s outside play around age five years, and 1516 parents around age seven years. Linear mixed model analyses were performed to evaluate (moderating) relationships among factors of the social environment (parenting influences and social capital), the perceived physical environment, and outside play at age five and seven. Season was entered as a random factor in these analyses. Results: Accessibility of PA facilities, positive parental attitude towards PA and social capital were associated with more outside play, while parental concern and restriction of screen time were related with less outside play. We found two significant interactions; both involving parent perceived responsibility towards child PA participation. Conclusion: Although we found a limited number of interactions, this study demonstrated that the impact of the perceived physical environment may differ across levels of parent responsibility.
MULTIFILE
PURPOSE: To investigate the longitudinal association between the macronutrient composition of the diet and frailty.METHODS: Data were obtained from 5205 Dutch middle-aged and older adults participating in the Rotterdam Study. Frailty was measured using a frailty index based on the accumulation of 38 health-related deficits, score between 0 and 100, and a higher score indicating more frailty. Frailty was assessed at baseline and 11 years later (range of 23 years). Macronutrient intake was assessed using food-frequency questionnaires. The association between macronutrients and frailty over time was evaluated using multivariable linear regression, adjusted for the frailty index at baseline, energy intake, and other relevant confounders. All analyses were performed in strata of BMI.RESULTS: Median frailty index score was 13.8 points (IQR 9.6; 19.1) at baseline and increased by a median of 2.3 points (IQR - 2.0; 7.6) after 11 years. Overall, we found no significant associations between intake of carbohydrates or fat and frailty over time. We did observe a significant positive association between an iso-energetic intake of 10 g protein and frailty over time (β 0.31 (95% CI 0.06; 0.55)) which was mainly driven by animal protein (β 0.31 (95% CI 0.07; 0.56)). It did not depend on whether it was substituted fat or carbohydrates.CONCLUSIONS: Our findings suggest that a reduction in the intake of animal protein may improve the overall health status over time in a relatively healthy population. More research is needed on the optimal macronutrient composition of the diet and frailty in more vulnerable populations.
