Comparing Treatment Outcomes of Eligible Patients Consenting to or Declining Randomization in a Randomized Clinical Trial: A Secondary Analysis of the FISH Trial on Humeral Shaft Fractures.

Background: The Finnish Shaft of the Humerus (FISH) trial compared open reduction and internal plate fixation (ORIF) with functional bracing in adult patients with displaced, closed humeral shaft fractures. Here, we compare the results of the patients in the randomized clinical trial (RCT [the randomized cohort]) with those of the cohort of patients who were also eligible but declined randomization (the nonrandomized cohort) to investigate if patients' treatment preference was associated with the outcomes during a 2-year follow-up. Methods: A total of 321 patients were treated at 2 university hospitals in Finland between November 2012 and January 2018. Of the 140 eligible patients, 82 were randomized to ORIF or functional bracing. Of the 58 patients declining randomization, 42 consented to participate in a nonrandomized cohort in which the patients were able to choose the treatment method. The primary outcome of this study was the Disabilities of the Arm, Shoulder and Hand (DASH) score. Patients in the randomized cohort and the nonrandomized cohort were analyzed separately in 3 groups: those who had (1) initial surgery, (2) successful functional bracing, and (3) late surgery due to failed functional bracing. We used mixed-model, repeated-measures analysis of variance to compare the treatment effect among the 3 groups. Results: In the randomized cohort, 38 patients had an initial surgical procedure. Of the 44 patients randomized to functional bracing, 30 (68%) healed successfully and 14 (32%) underwent a late surgical procedure. In the nonrandomized cohort, 9 patients preferred an initial surgical procedure. Of the 33 patients preferring functional bracing, 26 (79%) healed successfully and 7 (21%) underwent late surgery. The DASH scores in the randomized cohort and the nonrandomized cohort were 6.8 (95% confidence interval [CI], 2.3 to 11.4) and 12.3 (95% CI, 0.3 to 24.3) for the initial surgery groups, 6.0 (95% CI, 1.0 to 11.0) and 3.4 (95% CI, 0 to 9.3) for the bracing groups, and 17.5 (95% CI, 10.5 to 24.5) and 20.5 (95% CI, 9.4 to 31.6) for the late surgery groups at 2 years. Conclusions: The results of the randomized cohort and the nonrandomized cohort were comparable and suggest that patients' treatment preferences are not associated with the treatment outcomes of these injuries. Level of Evidence: Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence.