LAUSR

Conundrums in musculoskeletal research are common. Halliday and colleagues1 demonstrated similar outcomes for patients with chronic low back pain who received two seemingly different interventions: (1) a McKenzie approach and (2) motor control exercises. Chmielewski and associates2 reported similar knee function and psychosocial status scores in those who received either low intensity or high intensity plyometric exercise programmes. Finally, van Beijsterveildt et al found no differences in the prevention of injuries among amateur footballers when usual care was compared against an exercise strategy that consisted of core stability, eccentric training of thigh muscles, proprioceptive training, dynamic stabilisation and plyometrics exercises.3 These findings beg the question: Why do we see similar outcomes in randomised trials, which compare treatment strategies that are ostensibly different? Our goal is to provide explanations beyond the obvious answers —such as treatments that have similarly effective or ineffective causal mechanisms or failing to account for patient heterogeneity in trial design and consequent insufficient statistical power to detect differences. In this editorial, we present the reader with four additional yet plausible explanations for why trials report similar outcomes. ### Reason 1: the type of outcome assessment may bias findings Our first explanation is the most obvious one. Outcome measures that capture direct measurements of strength and/or flexibility, or other indices of physical performance, may yield findings that are notably different from outcome measures that capture self-reports of pain, function, overall health status or quality of life. Therefore, null findings (no difference among groups) may be related to the fact that different sets of outcome measures are being used to compare the groups being studied. Examples of variant findings among outcomes are present in the literature. One month after total hip arthroplasty, Dayton and colleagues4 report that performance-based function (eg, timed up and go, 6 min walk test and stair climbing test) declined compared with baseline, whereas self-reported …