OBJECTIVE: To explore the differences among missing data imputation methods when comparing treatment options. METHODS: Data from a randomized clinical trial evaluating the effects of treatment on asthma quality of life (QOL) were obtained. The instrument employed to measure QOL was the Asthma Quality of Life Questionnaire. Six common statistical techniques for handling missing data were evaluated: Complete case analysis (CCA), Markov modeling, hot deck, general imputation (GI), last value carried forward (LVCF), and regression. Scores in the tenth week were modeled using ANCOVA with the baseline score as the covariate. RESULTS: A total of 303 subjects were available for analysis, however only 134 subjects had data available for CCA. Two of the three treatment groups had 100 subjects while the remaining treatment group had103 subjects. Overall LVCF had the lowest average score (5.37, ?=1.03), indicating lower quality of life, whereas CCA produced the highest average score (5.83, ?=0.84). The effect of treatment group was found to be statistically insignificant across all six methods; however, the LVCF method (p=0.1667) and the GI method (p=0.1227) were substantial closer to achieving statistical significance than other methods. CONCLUSIONS: Differing imputation methods did not affect the decision of the analysis. CCA produced the highest average imputation scores while LVCF produced the lowest average imputation scores. Table 1. AQLQ Scores at the Tenth Week Utilizing Different Imputation Methods Imputation Method Treatment 1 Treatment 2 Treatment 3 Overall Significance of Treatment Group CCA (n=134) 5.84 (0.91) 5.68 (0.93) 5.95 (0.63) 5.83 (0.84) 0.228 Markov Model 5.71 (0.92) 5.82 (0.87) 5.85 (0.82) 5.80 (0.87) 0.367 Hot Deck 5.74 (0.91) 5.82 (0.91) 5.88 (0.71) 5.82 (0.84) 0.449 GI 5.62 (1.00) 5.53 (0.98) 5.76 (0.74) 5.64 (0.92) 0.123 LVCF 5.42 (1.06) 5.23 (1.10) 5.47 (0.93) 5.37 (1.03) 0.168 Regression 5.69 (0.83) 5.76 (0.77) 5.82 (0.63) 5.76 (0.74) 0.362