Statistician report
The primary outcome of interest is the change from baseline to 12-month follow up in the iHOT-12, comparing the PRP and placebo group. The study will be run as a superiority trial. The cut-off for statistical significance is set at 5% and desired power at 90%, with two-tailed tests applied. Data on change from baseline are also assumed to be parametric and the t test will be applied to the data to assess statistical significance.
The minimally clinically important difference (MCID) for the iHOT-12 has been reported by Sansone et al. as 10 (from 100) and the standard deviation for the change in score from baseline as no greater than 21 [
25]. Although, there are few previous data on which to base a sample size calculation, Monto et al. compared change from baseline to 12-month follow up in a group of 40 patient with hip bursitis, with steroids and PRP as the interventions being compared. In the PRP group Harris Hip scores increased from 51.7 to 87.4 whilst in the steroid group, scores increased from 50.5 to 58.8 at 12 months [
22]. Since the Harris Hip score is also scored from 100, our sample size calculation is based on these figures. We conservatively assume that change in iHOT score from baseline in the placebo group will be no more than in the steroid group reported by Monto et al., and estimate a maximal change of 10. We also estimate that the change in the iHOT score from baseline in the PRP group will be no less than 27. Using these figures, a minimal sample size at follow up of 66 (33 in each group) will be required.
Pilot data obtained by our team suggest that the rate of refusal to participate should be no more than 25% and the dropout rate no more than 35% over 12 months. Refusal rates tend to be low in this patient group given the chronic nature of the condition and the fact that patients will only be approached once conservative management has failed. Thus, we will need to approach 135.4 (rounded up to 136) patients, and recruit 102 patients to achieve our target sample size.
Data will be analysed using standard statistical software (e.g. SPSS and SAS). In the first instance, data will be analysed using simple descriptive statistics to compare the two groups in terms of demographics, clinical characteristics at baseline and outcomes. The primary outcome of interest will be the change from baseline to 12-month follow up in the iHOT-12, comparing the PRP and placebo group. The iHOT-12 data are expected to be parametric and so the unpaired t test will be used to compare the difference in change from baseline in the two groups. Since randomisation will not be stratified, we will also look to adjust for the possible confounding influence of differences in baseline characteristics (e.g. age, gender and body mass index (BMI)) using multivariable methods (e.g. linear regression). The significance level for all inferential tests will be set at 5%.
In secondary analysis, we will investigate changes in visual analogue pain score, EQ-5D 3 L and modified Harris Hip score in the two groups, as for the primary outcome. In sub-group analysis, we will investigate the data based on specific previous treatment, patients who required more than one treatment and patients with the highest levels of baseline pain.
The test group (PRP) alone: for all outcome measures the difference between the follow up scores with baseline will be assessed (e.g. baseline versus 3 months, baseline versus 12 months). The difference between one follow up period to the next will be examined (e.g. 3 months versus 6 months, 6 months versus 12 months). A significant difference will be set at a p value of less than 0.05.
The placebo group (normal saline) alone: data will be analysed as for the test group, as described.
Test versus placebo: for all outcome measures, the differential change at each follow up time point will be compared between the trial arms and the significance value calculated.
Statistical tests will be performed to ensure that gender, BMI and age differences are not significantly associated with a particular result. Care will be taken to minimise missing responses and to continue to follow up those who withdraw from treatment.
We will only be using the completed cases for final analyses; the remainder will be excluded.