Comparison of Five Analytic Techniques for Two - Group , Pre - Post Repeated Measures Designs Using SAS ®

There has been debate regarding which method to use to analyze repeated measures continuous data when the design includes only two measurement times. Five different techniques can be applied and give similar results when there is little to no correlation between preand post-test measurements and when data at each time point are complete: 1) analysis of variance on the difference between preand post-test, 2) analysis of covariance on the differences between preand post-test controlling for pre-test, 3) analysis of covariance on post-test controlling for pre-test, 4) multiple analysis of variance on post-test and pre-test, and 5) repeated measures analysis of variance. However, when there is missing data or if a moderate to high correlation between preand post-test measures exists under an intent-to-treat analysis framework, bias is introduced in the tests for the ANOVA, ANCOVA, and MANOVA techniques. A comparison of Type III sum of squares (where estimated), F-tests and p-values for a complete case and an intent-to-treat case are presented. The analysis using a complete case data set shows that all five methods produce similar results except for the repeated measures ANOVA due to a moderate correlation between preand post-test measures. However, significant bias is introduced for the tests when using the intent-to-treat data set. INTRODUCTION Many clinical trial designs use a pre-post repeated-measures, two-group design to assess the effect of a treatment over time compared to a control condition on a continuous variable. To analyze the data and determine the effect of the intervention, several analytic techniques exist, but the question becomes which method to use in the presence of a moderate to high correlation of the outcome measure between the two time points or in the presence of missing data at post-test. For non-statisticians, performing an analysis of variance (ANOVA), an analysis of covariance (ANCOVA) on the difference between preand post-test measures, or a multiple ANOVA (MANOVA) on both preand post-test is easier than performing a repeated measures mixed model. However, if a moderate to high correlation exists between the continuous measures at the two measurement times, the results of the ANOVA, ANCOVA and MANOVA are biased as the correlation between time points is not taken into account in the estimation. Similarly, if missing data exists at post-test the ANOVA, ANCOVA and MANOVA methods do not utilize all subjects with pre-test data, rather the analysis is done on subjects with complete data rather than an analysis on all available data. To overcome these biases, a repeated measures ANOVA using PROC MIXED can be performed to model the correlation structure between the measurement times and utilize all available data. METHODS STUDY DESIGN This study examined school attendance in children between the ages of 8 to 11, who were overweight or obese (BMI ≥ 85th percentile), healthy enough to participate in an exercise intervention, any race or ethnicity, and recruited from a public school system in the southeastern US. The study design was a randomized control clinical trial with two groups, exercise and control, from 4 cohorts measured at three time points, baseline, post-initial, and one-year followup. The study enrolled 175 study participants of which, 124 contained data for baseline school attendance. Children randomized to the exercise intervention spent 40 minutes per day in two 20-miniute sessions in activities eliciting vigorous exercise. The activities focused on intensity, enjoyment, and safety of the children and consisted of different games and activities that were easy to comprehend, not competitive, and brought about sporadic intervals of vigorous movement. Children allocated to the control group spent 60 minutes participating in instructor-led sedentary activities, including board games, watching videos, or similar sedentary activities.