دورية
Abstract 11521: Prevalence of Immortal Time Bias in High-Impact Cardiovascular Publications
| العنوان: | Abstract 11521: Prevalence of Immortal Time Bias in High-Impact Cardiovascular Publications |
|---|---|
| المؤلفون: | Samuel, Michelle, Kutcher, Stephen, Flatman, Leah K, Emmalin, Buajitti, Mahmoud, Farida, Bandargal, Saruchi, Brophy, Jay |
| المصدر: | Circulation (Ovid); November 2022, Vol. 146 Issue: Supplement 1 pA11521-A11521, 1p |
| مستخلص: | Background:Immortal time bias (ITB) is a consequence of non-uniform time zeros (ie. start of follow-up) between treatment groups, resulting in misclassification or selection bias in observational cohort studies. ITB poses a substantial problem as its presence favors the treatment, leading to an overestimation of the protective effect or an underestimation of the harmful effects of the treatment group compared to the control group. Depending on the amount of person-time misclassified or excluded, the magnitude of bias due to ITB may be substantially greater than other biases, such as confounding.Objective:To determine the prevalence of ITB in observational cohort studies published between 2020 and 2021 in high-impact cardiovascular journals.Methods:Observational cohort studies published in Circulation, European Heart Journal, and the Journal of the American College of Cardiology between January 2020 and December 2021 were screened for inclusion. Cross-sectional studies, case series, case-control, time-series analyses, and survey or surveillance studies were excluded. Two independent reviewers (ie. epidemiologists) evaluated each article for the presence and type of ITB.Results:Of 1,558 screen articles, 154 published articles were eligible for inclusion. Twenty of 154 (13%) publications had ITB present. ITB was most frequently due to misclassification bias (17 of 20 articles, 85%). ITB due to selection bias was present in 5 of 20 (25%) articles. Two articles had ITB due to both misclassification and selection biases. Most studies (75%) with ITB did not have an active comparator group. Among studies with ITB, event-based cohorts were the most frequent (65%), followed by event-exposure based cohorts (15%), and exposure-based (10%) and time-based (10%) cohorts. ITB was present in studies with various exposures, including medications (7), surgeries or procedures (5), devices (4), and diseases (4).Conclusion:A substantial proportion (13%) of published observational cohort studies in high-impact cardiovascular journals had ITB present and may result in an overestimation of treatment effect. As ITB is preventable with study design techniques, researchers need to be cognizant of this bias. |
| قاعدة البيانات: | Supplemental Index |
Full Text Finder | تدمد: | 00097322 15244539 |
|---|---|
| DOI: | 10.1161/circ.146.suppl_1.11521 |