This report provides guidance about the use of two-phase study designs within Mini-Sentinel in order to improve adjustment for potential confounders. Two-phase study designs can improve the efficiency of supplemental data collection, such as medical record review, and also provide additional information about outcomes or exposures. In many settings, two-phase studies can substantially reduce bias while maintaining acceptable precision, thus helping to resolve some of the uncertainty present when signals arise from routine surveillance activities based on administrative data.
The simulation tool for two-phase study designs used in this project is also posted here. Documentation for this tool is embedded in the computer program, which is written in the R programming language. Those who use this tool should be familiar with simulations, R, and two-phase methodology. The code may be altered or repurposed as needed. For further information about the tool, contact Rod Walker (email@example.com) and Sascha Dublin (firstname.lastname@example.org) at Group Health Cooperative.
The programming code provided here has not been formally audited in accordance with the Mini-Sentinel Standard Operating Procedure for Quality Control of SAS Programs. It is being provided in the spirit of supporting the exploration and development of novel scientific and statistical methods using observational healthcare data.
Sascha Dublin, MD, PhD; Group Health Research Institute and University of Washington, Seattle, WA
Dan Mines, MD, MSCE; HealthCore, Inc., Wilmington, DE
Robert Davis, MD, MPH; Kaiser Permanente Georgia, Atlanta, GA
Kevin Haynes, PharmD, MSCE; University of Pennsylvania Perelman School of Medicine, Philadelphia, PA
Erika Avila-Tang, PhD, MHS; Manuel Bayona, MD, MS, PhD; Yelizaveta Torosyan, MD, PhD; Center for Biologics Evaluation and Research, FDA, Silver Spring, MD
Aloka Chakravarty, PhD; Brad McEvoy, DrPH; Eric Frimpong, PhD, MA; Yu-te Wu, PhD; Center for Drug Evaluation and Research, FDA, Silver Spring, MD
Darren Toh, ScD; Department of Population Medicine, Harvard Pilgrim Health Care Institute and Harvard Medical School, Boston, MA
Soko Setoguchi, MD, DrPH; Duke Clinical Research Institute, Durham, NC