OBJECTIVES: In an analyses of patients with type 2 diabetes (T2DM) in a large electronic medical record (EMR) database, small differences were found to be statistically significant between N=1162 patients with liraglutide versus a comparison cohort due to the comparison group sample size (n=274,922). The purpose of this study is to evaluate a bootstrapping technique to provide robust mean and variance estimates for comparison patients, thereby helping to address the issue of being over-powered.  METHODS: Study patients were age ≥18 years with T2DM, prescribed liraglutide or other antidiabetic drug January 1, 2010 to July 16, 2010 and with ≥13 months of EMR activity. Bootstrapping was used to provide cohort mean, standard deviation, and 95% CI estimates for the comparison cohort and were calculated as the mean of the mean values identified in 1000 random draws with replacement of 1162 comparison patients. Means (95% CI) were compared for continuous variables (age, HbA1c and blood pressure [BP]) for liraglutide versus the overall comparison group and to bootstrap estimates.  RESULTS: Of 1162 liraglutide patients, mean (95% CI) age was 55.5 (54.9, 56.2) years versus 60.9 (60.8, 60.9) years for all comparison patients and 60.9 (60.1, 61.6) years for comparison patients per bootstrap estimates (both p<0.05). HbA1c was 8.12% (8.00, 8.24) for liraglutide versus 7.62% (7.61, 7.63) and 7.63% (7.49, 7.76) for all comparison patients and per bootstrap, respectively (both p<0.05).  BP was 127.0 (126.1, 127.8)/75.8 (75.3, 76.4) mmHg for liraglutide versus 130.1 (130.0, 130.1)/75.4 (75.4, 75.5) mmHg and 130.1 (129.0, 131.2)/75.5 (74.8, 76.1) mmHg for all comparison patients and per bootstrap, respectively (p<0.05 except bootstrap diastolic BP p>0.05). CONCLUSIONS: A bootstrap analysis provided more robust variance and 95% CI estimates for a large comparison group.  This technique can help researchers avoid identifying statistical significance when differences are not clinically meaningful when evaluating a large patient cohort.