Hard-soled Shoe Versus Short Leg Walking Cast for a Fifth Metatarsal Avulsion Fracture: A Randomized Multicenter Noninferiority Trial

This study is a noninferiority, randomized controlled trial of patients presenting to multiple centers. The primary outcome measure is the 100mm VAS assessed on 6 months after a fifth metatarsal avulsion fracture. 100mm VAS was compared between the experimental group (hard-soled shoe) and the control group (short leg walking cast) on 6 months follow-up examination. The 100mm VAS is widely used self assessment tool for evaluating pain after fracture. The secondary objectives were to determine whether hard-soled shoe was superior to short leg walking cast with respect to time until return to normal daily life and patient's satisfaction.

The sample size was determined using methods appropriate for noninferiority trials, assuming 90% power and a significance level of 0.05. We performed a pilot study on 18 patients and assessed 100mm VAS on 6 months. Pooled standard deviation was 16.6. To find out whether the hard-soled shoe is not inferior to short leg walking cast for a fifth metatarsal avulsion fracture, 74 patients were required to have 90% power that the lower limit of an one-sided 95% confidence interval for the difference between two treatments will be above the noninferiority margin of -10. We assumed the drop-out rate to be 20%. Adding this number, 96 patients were required for the study.

Determination of the noninferiority margin was based on clinical significance. Todd et al. reported on a study of 48 patients that the minimum clinically significant difference of 100mm VAS was 13mm and that differences of less than this amount, even if statistically significant, are unlikely to be of clinical significance. We decided that the noninferiority margin at 10 mm difference which is less than this amount will be adequate to prove noninferiority of the experiment group (hard-soled shoe) over the control group (short leg walking cast).

After 1 weeks of immobilization in a short leg posterior splint, patients are randomized to experiment group (hard-soled shoe) and control group (short leg walking cast). Randomization was stratified by the study centers. Block randomization were used and the investigators were blinded to the block sizes being used to ensure allocation concealment. Delayed union was defined as a lack of bridging callus on 3 of 5 cortices at 12 weeks. Nonunion was defined as lack of cortical bridging or a clearly visible fracture line, at 14 weeks post injury.

If a subject had discontinued prior to completion of 6 months, the last observation was carried forward for the intent-to-treat analysis. Subjects who crossed over to the other treatment arm, for an example, patients in experiment group(hard-soled shoe) who changed to short leg walking cast or in control group (short leg walking cast) who changed to hard-soled shoe, were analyzed according to their initial group allocation for the intent-to-treat analysis. Additionally, an as-treated (per-protocal) analysis was also conducted on patients who completed the 6 months follow-up with the protocol assigned.

Although previous studied have not documented an increase risk of reduction loss with protected weightbearing in a hard-soled shoe protocol, a safety rule was put in place to assure patients, clinicians, and the health research ethics board that the safety of our patients was a priority. Reduction loss was not a primary outcome, but it was monitored carefully regardless of published evidence. A safety rule established a priori eliminated any dissent among the research team and participating surgeons about stopping the study in the event of an apparent adverse outcome related to protected weightbearing in a hard-soled shoe. If a difference of greater than three spontaneous reduction loss occurred between the groups, the intervention would be halted. A reduction loss or metal failure was defined as one that occurred without patient instigation of inappropriate activity. The reduction loss was to be monitored by study personnel not involved in the outcome assessment.