BEGIN:VCALENDAR VERSION:2.0 PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4// BEGIN:VEVENT UID:20250505T140750EDT-4474u0sTem@132.216.98.100 DTSTAMP:20250505T180750Z DESCRIPTION:Michael Abrahamowicz\, PhD\n\nDistinguished James 51ԹProfes sor of Biostatistics\n Department of Epidemiology\, Biostatistics and Occup ational Health | 51Թ\n\nWHEN: Wednesday\, January 15\, 2025\ , from 3:30 to 4:30 p.m.\n WHERE: Hybrid | 2001 51ԹCollege Avenue\, Roo m 1140\; Zoom\n NOTE: Michal Abrahamowicz will be presenting in-person\n\nA bstract\n\nThis is a quite “applied” talk that may be of interest for bios tatisticians who collaborate on real-world clinical or epidemiological res earch projects. The goal is to promote use of targeted simulations to expl ore particular limitations of a specific real-world prognostic or epidemio logical study. To this end\, our data-driven simulations are designed so a s to accurately reflect the salient characteristics of the real-world data set being analyzed [1]. (In contrast to methods-driven simulations typical ly reported in statistical publications\, which rely on data generated und er arbitrary assumptions). After an overview of 7 generic steps proposed t o implement our approach\, I will illustrate it in two real-world time-to- event (survival) analyses\, each dealing with a different data limitation The first illustration concerns assessing the impact of omitting an import ant prognostic factor on the adjusted Hazard Ratio (HR) for the exposure o f interest\, based on multivariable Cox proportional hazards (PH) analyses . Here\, we show how data-driven simulations permit assessing the joint im pact of (i) unmeasured confounding bias and (ii) non-collapsibility\, whil e separating their effects. The second illustration focuses on the pharmac o-epidemiological study of the association between recent use of medicatio n\, modeled as a time-varying exposure\, and the hazard of a transient cog nitive impairment. The event is interval-censored as it can be detected on ly at discrete times of medical visits. Here we use the permutational algo rithm\, validated for simulating event times conditional on time-varying e xposures and/or effects [2]. Here\, our simulation results reveal how the strength of a systematic bias toward the null varies depending on the way event times are imputed\, and help decide which of the divergent results o f alternative imputation strategies may be closer to the (unknown) truth. \n\n \n\n[1] Abrahamowicz M\, Beauchamp ME\, Boulesteix AL\, Morris TP\, S auerbrei W\, Kaufman JS. Data-driven simulations to assess the impact of s tudy imperfections in time-to-event analyses. American Journal of Epidemio logy. 2024 May:kwae058. doi: 10.1093/aje/kwae058 [Online ahead of print]. \n\n[2] Sylvestre MP & Abrahamowicz M. Comparison of algorithms to generat e event times conditional on time-dependent covariates. Statistics in Medi cine 2008\; 27: 2618-2634.\n\nSpeaker Bio\n\nMichal Abrahamowicz is a Dist inguished James 51ԹProfessor of Biostatistics at 51Թ. He develops new\, flexible statistical methodology for survival analyses\, w ith focus on time-varying exposures and effects. He also explores\, and at tempts to correct for\, different biases in epidemiological studies and pr omotes creative applications of statistical simulations. He is a co-founde r and the co-chair of the international STRATOS initiative for improving t he analyses of observational studies. He is an Honorary Lifetime member of the International Society for Clinical Biostatistics.\n DTSTART:20250115T203000Z DTEND:20250115T213000Z SUMMARY:Data-driven simulations: a new Quantitative Bias Analysis tool for real-world applications of survival analyses URL:/epi-biostat-occh/channels/event/data-driven-simul ations-new-quantitative-bias-analysis-tool-real-world-applications-surviva l-362132 END:VEVENT END:VCALENDAR