Jan Beyersmann1, Petra Gastmeier2, Martin Wolkewitz3, Martin Schumacher4. 1. Freiburg Centre for Data Analysis and Modelling, Freiburg University, Freiburg, Germany; Institute of Medical Biometry and Medical Informatics, University Medical Center Freiburg, Freiburg, Germany. Electronic address: jan.beyersmann@fdm.uni-freiburg.de. 2. Institute of Medical Microbiology and Hospital Epidemiology, Hannover Medical School, Germany. 3. Freiburg Centre for Data Analysis and Modelling, Freiburg University, Freiburg, Germany; Institute of Medical Biometry and Medical Informatics, University Medical Center Freiburg, Freiburg, Germany. 4. Institute of Medical Biometry and Medical Informatics, University Medical Center Freiburg, Freiburg, Germany.
Abstract
OBJECTIVE: Time-dependent bias occurs when future exposure status is analyzed as being known with start of observation. As this bias is common, we sought to determine whether it always leads to biased effect estimation. We also sought to determine the direction of the effect bias. STUDY DESIGN AND SETTING: We derived an easy mathematical proof investigating the nature of time-dependent bias. We applied the general mathematical result to data from a prospective cohort study on the incidence of hospital infection in intensive care: Here, we investigated the effect of time-dependent hospital infection status on intensive care unit stay. The nature of time-dependent bias was also illustrated graphically. RESULTS: Biased effect estimation is a mathematically inevitable consequence of time-dependent bias, because the number of individuals at risk of exposure is distorted over the course of time. In case of a time-dependent exposure that prolongs time until the study endpoint, the prolonging effect will be overestimated. CONCLUSION: Because time-dependent bias inevitably leads to erroneous findings, it is a major concern that it is common in the clinical literature. Time-dependent bias can be avoided by proper hazard-based analyses.
OBJECTIVE: Time-dependent bias occurs when future exposure status is analyzed as being known with start of observation. As this bias is common, we sought to determine whether it always leads to biased effect estimation. We also sought to determine the direction of the effect bias. STUDY DESIGN AND SETTING: We derived an easy mathematical proof investigating the nature of time-dependent bias. We applied the general mathematical result to data from a prospective cohort study on the incidence of hospital infection in intensive care: Here, we investigated the effect of time-dependent hospital infection status on intensive care unit stay. The nature of time-dependent bias was also illustrated graphically. RESULTS: Biased effect estimation is a mathematically inevitable consequence of time-dependent bias, because the number of individuals at risk of exposure is distorted over the course of time. In case of a time-dependent exposure that prolongs time until the study endpoint, the prolonging effect will be overestimated. CONCLUSION: Because time-dependent bias inevitably leads to erroneous findings, it is a major concern that it is common in the clinical literature. Time-dependent bias can be avoided by proper hazard-based analyses.
Authors: Debra A Goldman; Koos Hovinga; Anne S Reiner; Yoshua Esquenazi; Viviane Tabar; Katherine S Panageas Journal: J Neurosurg Date: 2018-11-01 Impact factor: 5.115
Authors: Clare L Atzema; Peter C Austin; Bing Yu; Michael J Schull; Cynthia A Jackevicius; Noah M Ivers; Paula A Rochon; Douglas S Lee Journal: CMAJ Date: 2018-12-17 Impact factor: 8.262
Authors: Richard E Nelson; Vanessa W Stevens; Makoto Jones; Karim Khader; Marin L Schweizer; Eli N Perencevich; Michael A Rubin; Matthew H Samore Journal: Antimicrob Agents Chemother Date: 2018-10-24 Impact factor: 5.191
Authors: Clare L Atzema; Cynthia A Jackevicius; Alice Chong; Paul Dorian; Noah M Ivers; Ratika Parkash; Peter C Austin Journal: CMAJ Date: 2019-12-09 Impact factor: 8.262