Premium
Sampling Patients within Physician Practices and Health Plans: Multistage Cluster Samples in Health Services Research
Author(s) -
Adams John L.,
Wickstrom Steven L.,
Burgess Margaret J.,
Lee Paul P.,
Escarce José J.
Publication year - 2003
Publication title -
health services research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.706
H-Index - 121
eISSN - 1475-6773
pISSN - 0017-9124
DOI - 10.1111/j.1475-6773.2003.00196.x
Subject(s) - sampling frame , computer science , sample (material) , sampling (signal processing) , cluster sampling , sample size determination , sampling design , research design , data collection , health care , principal (computer security) , multistage sampling , frame (networking) , data science , medicine , environmental health , statistics , mathematics , population , filter (signal processing) , telecommunications , chemistry , chromatography , pathology , computer vision , economic growth , operating system , economics
Objective. To better inform study design decisions when sampling patients within health plans and physician practices with multiple analysis goals. Study Setting. Chronic eye care patients within six health plans across the United States. Study Design. We developed a simulation‐based approach for designing multistage samples. We created a range of candidate designs, evaluated them with respect to multiple sampling goals, investigated their tradeoffs, and identified the design that is the best compromise among all goals. This approach recognizes that most data collection efforts have multiple competing goals. Data Collection. We constructed a sample frame from all diabetic patients in six health plans with evidence of chronic eye disease (glaucoma and retinopathy). Principal Findings. Simulations of different study designs can uncover efficiency gains as well as inform potential tradeoffs among study goals. Simulations enable us to quantify these efficiency gains and to draw tradeoff curves. Conclusions. When designing a complex multistage sample it is desirable to explore the tradeoffs between competing sampling goals via simulation. Simulations enable us to investigate a larger number of candidate designs and are therefore likely to identify more efficient designs.