Transcript Slide 1 - Annals of Internal Medicine

From: Survivor Treatment Selection Bias in Observational Studies: Examples from the AIDS Literature
Ann Intern Med. 1996;124(11):999-1005. doi:10.7326/0003-4819-124-11-199606010-00008
Figure Legend:
Hypothetical example of survivor treatment selection bias.dashed linePlots of expected Kaplan-Meier estimates of survival in a hypothetical observational cohort of 1000 patients with the acquired immunodeficiency
syndrome who are untreated ( ) or treated (solid line) with a drug that has no effect on survival. Regardless of drug use, 50 patients die each month so that all have died by 20 months. Beginning after the first month of
observation, 10% of patients who are alive and untreated at each successive monthly interval initiate therapy with the ineffective drug so that about 56% of patients will use the treatment before they die. Even though the
treatment is ineffective, it erroneously appears to prolong survival because of survivor treatment selection bias: Living longer increases the chance to use the treatment.
Date of download: 4/11/2017
Copyright © American College of Physicians. All rights reserved.