Comparison of Usual Podiatric Care and Early Physical Therapy for Plantar Heel Pain

This investigation will be a randomized clinical trial comparing participant-reported outcomes and cost-effectiveness of ePT and uPOD with follow up extending to 1 year. Following eligibility screening and completion of study questionnaires participants will be randomized using concealed envelopes and proceed with the intervention of their assigned group. Outcomes will be collected at 6 weeks, 6 months, and 1 year after initial presentation. All outcomes will be completed by the patient without any influence from investigators aware of the group assignment. Data entry and processing will occur by an investigator blinded to group allocation. A small financial incentive will be provided to facilitate completion of outcome measures over the study duration.

Statistical Analysis: Baseline group variables will be summarized using the mean and standard deviation for continuous measures and percentages for categorical measures. Independent t-tests (p<0.05, two-tailed) or the appropriate nonparametric test will be used to compare between group differences in baseline characteristics. Parametric test assumptions will be analyzed for all continuous variables by visual inspection of histograms, use of skewness score within double the standard error of skewness criteria, and Levene's test for homogeneity of the variance. A repeated measures analysis of variance will be used to compare group differences in the FAAM, NPRS, EQ-5D, and number of office visits associated with treatment at each time point. Results will be reported as the group mean, mean difference between groups, 95% confidence intervals, f-value, p-value, power, and effect size. The chi-square test will be used to compare GROC scores between groups at each time point and results will be reported as the chi-square value, p-value, and the frequency per category. The types of treatment based on Current Procedural Terminology (CPT) and Healthcare Common Procedure Coding System (HCPS) codes provided per group will be reported as percentages. Analysis of covariance will be used if any group differences are observed in participant characteristics. Intention to treat analysis will be performed by comparing the complete case analysis to multiple imputation analysis. Multiple imputed data sets will be generated using the Multivariate Imputation by Chained Equations algorithm with SPSS 19.0 for Windows (SPSS Inc., Chicago, IL, USA). Multinomial logistic regression will be used to obtain pooled regression estimates from the data sets which will be used for analysis. Post-randomization exclusion of included participants that did not meet eligibility criteria or that did not receive intervention will be considered for exclusion of the analysis by an independent, blinded adjudication committee that will evaluate all randomized participants. Any cross-overs will be analyzed in the original group they were assigned to and compared to an analysis that excludes cross-overs.

Participant expectations and preferences for ePT or uPOD will be categorized into 3 categories; matched, unmatched and neutral. Participants will be labelled 'matched' if they are allocated to a group for which they have expressed a higher expectation of benefit. A higher expectation will be denoted by comparison of visual analog ratings between the groups (ePT and uPOD). Unmatched participants will be those who are allocated to the treatment group for which they have a lower expectation of benefit. Neutral participants will be those who indicate the same level of expectation for both treatments. Similar categorizations will be made for participant preference based upon their response to the treatment preference question. General expectations of improvement will be dichotomized into met or unmet based upon rankings at the 6 week, 6 month, and 1 year follow-up relative to baseline rankings. Participants that demonstrate 6 week, 6 month, and 1 year ranks equal to or higher than baseline will be considered to have met their global expectations of improvement. Participants that demonstrate rankings below baseline expectations will be considered to have unmet global expectations of improvement. A chi-square test of independence will be used to compare the proportions of individuals with matched/unmatched/neutral expectations or preferences to global expectations (met or unmet). In addition, differences in expectation and preference categories will be analyzed relative to treatment success using a chi-square test of independence. Treatment success will be determined by the GROC at each time period with success defined as a GROC of +5, "a great deal better," or greater.

Power Analysis: Sample size estimates were calculated based on the primary outcome measure, the FAAM, at 6 months. In the absence of research comparing physical therapy and podiatry interventions or podiatry interventions using the FAAM as an outcome measure, sample size calculations were based on achieving a clinically meaningful difference between the groups and details from a recent clinical trial by Cleland et al that used similar methods to this investigation. Sample size estimate was made using G*Power 3.1.5 based on detecting a difference between groups greater than the minimal clinically important difference (MCID; ie, 9 point change) of the FAAM at 6 months with an alpha level of 0.05, 80% power, and pooled sample variance of 14.5 from Cleland et al This resulted in an effect size of 0.62 and 42 participants needed per group. This estimate is similar to the effect size of 0.69 achieved in the Cleland et al study that had 27 subjects per 2 treatment groups. To account for participants who drop-in or drop-out of treatment, in addition to the possibility of some participants not returning the FAAM questionnaire, the sample will be increased by 33% resulting in 56 participants per group. This estimate is conservative relative to the 3% drop-out rate and 7% rate of individuals that did not return outcome forms at the 6 month follow-up in the Cleland et al investigation. Significant effort will be made to retain subjects to include financial incentives and follow-up by the research assistant to assure completion and return of all outcome forms. The focus, and consequently the power analysis, of this investigation is on functional outcome (based on FAAM scores) and therefore may result in underpowered analysis of secondary variables.