Abstrak: Outcome misclassification is widespread in epidemiology, but methods to account for it are rarely used. We describe the use of multiple imputation to reduce bias when validation data are available for a subgroup of study participants. This approach is illustrated using data from 308 participants in the multicenter Herpetic Eye Disease Study between 1992 and 1998 (48% female; 85% white; median age, 49 years). The odds ratio comparing the acyclovir group with the placebo group on the gold-standard outcome (physician-diagnosed herpes simplex virus recurrence) was 0.62 (95% confidence interval (CI): 0.35, 1.09). We masked ourselves to physician diagnosis except for a 30% validation subgroup used to compare methods. Multiple imputation (odds ratio (OR) = 0.60; 95% CI: 0.24, 1.51) was compared with naive analysis using self-reported outcomes (OR = 0.90; 95% CI: 0.47, 1.73), analysis restricted to the validation subgroup (OR = 0.57; 95% CI: 0.20, 1.59), and direct maximum likelihood (OR = 0.62; 95% CI: 0.26, 1.53). In simulations, multiple imputation and direct maximum likelihood had greater statistical power than did analysis restricted to the validation subgroup, yet all 3 provided unbiased estimates of the odds ratio. The multiple-imputation approach was extended to estimate risk ratios using log-binomial regression. Multiple imputation has advantages regarding flexibility and ease of implementation for epidemiologists familiar with missing data methods.

Abstrak: Internal validity of a risk model can be studied efficiently with bootstrapping to assess possible optimism in model performance. Assumptions of the regular bootstrap are violated when the development data are clustered. We compared alternative resampling schemes in clustered data for the estimation of optimism in model performance. A simulation study was conducted to compare regular resampling on only the patient level with resampling on only the cluster level and with resampling sequentially on both the cluster and patient levels (2-step approach). Optimism for the concordance index and calibration slope was estimated. Resampling of only patients or only clusters showed accurate estimates of optimism in model performance. The 2-step approach overestimated the optimism in model performance. If the number of centers or intraclass correlation coefficient was high, resampling of clusters showed more accurate estimates than resampling of patients. The 3 bootstrap schemes also were applied to empirical data that were clustered. The results presented in this paper support the use of resampling on only the clusters for estimation of optimism in model performance when data are clustered.