Forty participants, women and men with knee OA, were recruited and then randomly assigned either to the intervention group (n = 21), which was designed to evaluate effects of a self-massage protocol, or the control group (n = 19), the members of which continued their usual care returning only for assessments. We verified that with our results (small standard deviations), this did not affect the statistical validity of a two-tailed test of intervention compared to no intervention, at 80% power.