Patients' health outcomes after an implementation intervention targeting the physiotherapists' clinical behaviour.

The study was underpowered. It planned 230 patients, but only 155 consented and the patient characteristics table analysed 70 active-group patients and 63 passive-group patients. Seven of the 24 included physiotherapists did not recruit any patients because of sick leave, maternity leave, changed workplace or lack of interest. The physiotherapists did not report how many patients declined participation, so the true recruitment picture is incomplete. Patients who were lost to follow-up were younger and included more men than those who remained. Missing data at any measurement point led to exclusion of all data from that patient, reducing the analysed sample. There was a 35% dropout rate between the first and second measurements. The design was explorative and comparative, and the supplied text does not state randomisation of physiotherapists or patients. Sick leave was not balanced between groups at baseline, with more patients on sick leave in the active group before treatment. Treatment content was assessed by patients' self-report, which may not capture all actions taken by the physiotherapists. The study used a less conservative post hoc test and multiple comparisons, which increased the risk of false-positive findings.

Do not read the sick leave reduction in the active support group as evidence that the active implementation support worked. More patients in that group were already on sick leave before and after treatment, and the authors could not separate the effect of the support from that baseline imbalance.