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
Full text met een HU Account Objective: To quantify diversity in components of self-management interventions and explore which components are associated with improvement in health-related quality of life (HRQoL) in patients with chronic heart failure (CHF), chronic obstructive pulmonary disease (COPD), or type 2 diabetes mellitus (T2DM). Methods: Systematic literature search was conducted from January 1985 through June 2013. Included studies were randomised trials in patients with CHF, COPD, or T2DM, comparing self-management interventions with usual care, and reporting data on disease-specific HRQoL. Data were analysed with weighted random effects linear regression models. Results: 47 trials were included, representing 10,596 patients. Self-management interventions showed great diversity in mode, content, intensity, and duration. Although self-management interventions overall improved HRQoL at 6 and 12 months, meta-regression showed counterintuitive negative effects of standardised training of interventionists (SMD = 0.16, 95% CI: 0.31 to 0.01) and peer interaction (SMD = 0.23, 95% CI 0.39 to 0.06) on HRQoL at 6 months. Conclusion: Self-management interventions improve HRQoL at 6 and 12 months, but interventions evaluated are highly heterogeneous. No components were identified that favourably affected HRQoL. Standardised training and peer interaction negatively influenced HRQoL, but the underlying mechanism remains unclear. Practice implications: Future research should address process evaluations and study response to selfmanagement on the level of individual patients
MULTIFILE
Background: A hospital group is an organizational integration strategy that has recently been widely implemented in Chinese urban health systems to promote integrated care. This study aims to evaluate the effect of hospital group on integrated care from the perspectives of both patients and care professionals. Methods: Two cross-sectional surveys were conducted in Shenzhen city of China, in June 2018 and July 2021. All 30 community health stations (CHSs) in the hospital group were included in the intervention group, with 30 CHSs in the same district selected as the control group by simple random sampling. All care professionals within both the intervention and the control groups were invited to participate in the surveys. Twelve CHSs were selected from 30 CHSs in the intervention and the control groups by simple random sampling, and 20 patients with type 2 diabetes mellitus (T2DM) were selected from each of these selected CHSs to participate in the survey by systematic sampling. The Chinese version Rainbow Model of Integrated Care Measurement Tool (C-RMIC-MT) was used to assess integrated care. Propensity score matching and difference-in-differences regression (PSM-DID) were used to evaluate the effect of the hospital group on integrated care. Results: After matching, 528 patients and 1896 care professionals were included in the DID analysis. Results from care professionals indicated that the hospital group significantly increased technical competence of the health system by 0.771 points, and cultural competence by 1.423 points. Results from patients indicated that the hospital group significantly decreased organizational integration of the health system by 0.649 points. Conclusion: The results suggests that the effect of the hospital group on integrated care over and above routine strategies for integrated care is limited. Therefore, it is necessary to pay attention to implementing professional, clinical and other integration strategies beyond establishing hospital groups, in urban Chinese health systems.
DOCUMENT
Zijn data-analyse en bio-informatica de sleutel naar voorspellingen over de invloed van giftige stoffen op de gezondheid van mensen? Het project DART Pathfinder is een vervolgonderzoek naar een dierproefvrije testmethode. Met moderne ICT-technieken proberen we die voorspellingen te doen.Doel Het doel van dit project is om gegevens over giftige stoffen uit verschillende data bronnen samen te brengen. In het onderzoek gebruiken we technieken uit de bio-informatica. Zo willen we de eigenschappen van giftige stoffen beter in kaart brengen en (nadelige) effecten van soortgelijke stoffen kunnen voorspellen. Veel bedrijven maken producten of stoffen, die getest moeten worden of ze veilig zijn. Met dit project helpen we bedrijven om o.b.v. bestaande gegevens een betere keuze te maken welke testen ze hiervoor het beste kunnen gebruiken. Resultaten Kennis over computer modellen die voorspellingen doen, zoals machine learning, regression tree-based models; Nieuwe algoritmen (instructies om berekeningen uit te voeren) Inzicht in nieuwe biologische mechanismen obv data science Nieuwe statische methoden om data te analysen en voorspellingen te doen. Looptijd 01 februari 2018 - 01 februari 2022 Aanpak Met de gegevens uit het onderzoek maken we een computermodel dat voorspelt of giftige stoffen invloed hebben op de voortplanting en ontwikkeling van mensen. Die voorspelling gebeurt via machine learning, algoritmen en statistische methoden. Voor dit model wordt informatie uit publieke databases over fysische en chemische eigenschappen van mogelijk gevaarlijke stoffen samengevoegd met de gegevens over de invloed van deze stoffen op levende organismen. Net als in het eerste onderzoek (PreDART) werken we met rondwormen (C.elegans) en embryo's van zebravissen, met als doel geen proeven meer met ratten en konijnen te hoeven doen.
