Use of insurance claims databases to evaluate the outcomes of ophthalmic surgery.

This article reviews methodological issues in research using Medicare claims data and reviews how those issues affect the interpretation of study results. Although studies using Medicare claims data can improve our knowledge of surgical outcomes, the ability to infer that one type of surgery or one surgeon is better than another is limited by such factors as bias due to inaccurate record keeping, confounding of treatment effects by other covariates, effect modification due to factors associated with treatment, and selective loss of follow-up of patients. Five studies on the outcomes of cataract surgery and capsulotomy are reviewed, providing illustrations of methodological strengths and weaknesses of this type of research.