Activity Number:
|
333
- SPEED: Biopharmaceutical Statistics, Medical Devices, and Mental Health
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 1, 2017 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biopharmaceutical Section
|
Abstract #324030
|
View Presentation
|
Title:
|
Subgroup Analyzes with Survival Data in Retrospective Cohort Studies
|
Author(s):
|
Rima Izem* and Jiemin Lao and Mao Hu and Yuqin Wei and Michael Wernecke and Jeffrey A Kelman and David J Graham
|
Companies:
|
Food and Drug Administration and Sphere Institute and Acumen LLC and Sphere Institute and Acumen LLC and Centers for Medicaid and Medicare Services (CMS) and Food and Drug Administration
|
Keywords:
|
subgroup analyses ;
propensity score matching ;
inverse probability of treatment weighting ;
Cox regression
|
Abstract:
|
Subgroup analyses examine whether benefits and/or risk of health interventions are similar in different subsets of the population. In this era of precision medicine, these analyses are more important than ever for regulatory decisions and public health. Subgroup analyses have higher power to detect risks in large cohort studies compared to smaller clinical trials. However, analyses in cohort studies need to control for confounding for proper causal inference. Analysis methods used to estimate risk in subgroups of a cohort vary in terms of amount of information borrowed across subgroups to control for confounding or to estimate risk. This presentation will give practical advice for subgroup analyses to estimate hazard ratio with survival data when using propensity score methods (matching or weighting) to control for confounding. This presentation will emphasize the trade-offs between bias and precision with different methods to control for confounding. The advice is informed by literature review, extensive simulation work and real world application in drug safety assessment using Centers for Medicare and Medicaid Services data.
|
Authors who are presenting talks have a * after their name.