What question did this study set out to answer?

September 28, 2005American Journal of Epidemiology

Survival Bias Associated with Time-to-Treatment Initiation in Drug Effectiveness Evaluation: A Comparison of Methods

View Full Paper

Ask AI

Bookmark

Authors

ZZZheng ZhouNorth Sichuan Medical University ERElham RahmeMcGill University Health Centre MAMichał AbrahamowiczUniversité du Québec à Chicoutimi

Discussion

Member takes

Implication

Observational cohort study uncovers survival bias across methods in post-myocardial infarction statin users, demonstrating that time-dependent modeling yields accurate risk estimates.

Key Points

To compare five statistical methods for addressing survival bias associated with time-to-treatment initiation in observational drug effectiveness evaluations.
Analyzed Quebec medical administrative databases (1996–2002) evaluating statins for secondary prevention in elderly post-acute myocardial infarction patients initiating therapy within 90 days post-discharge versus nonusers.
Evaluated two naive approaches prone to survival bias and three bias-controlling methods: 90-day landmark follow-up, prescription-time distribution matching at entry, and time-dependent treatment variables.
Classifying treatment at discharge (Method 1) exaggerated benefit to a 38% relative risk reduction at 1 year, whereas randomly assigning nonuser start dates (Method 2) attenuated it to a 10% relative risk reduction.
Following patients after a fixed 90-day window (Method 3) controlled survival bias but resulted in a substantial loss of statistical precision and efficiency.
Matching prescription-time distributions (Method 4) and time-dependent modeling (Method 5) both effectively controlled survival bias, revealing a 20% risk reduction for recurrent myocardial infarction or death.

Cite This Study

Zhou et al. (2005) studied this question.

synapsesocial.com/papers/69dff8eb93e101b251e9c3f8 https://doi.org/10.1093/aje/kwi307

View Full Paper

Ask AI

Bookmark

Implication

Observational cohort study uncovers survival bias across methods in post-myocardial infarction statin users, demonstrating that time-dependent modeling yields accurate risk estimates.

Survival Bias Associated with Time-to-Treatment Initiation in Drug Effectiveness Evaluation: A Comparison of Methods

Discussion

Member takes

Implication

Key Points

Cite This Study

Survival Bias Associated with Time-to-Treatment Initiation in Drug Effectiveness Evaluation: A Comparison of Methods

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider