Experimental design
The HAPPi Kneecaps! Study (sHoe inserts for Adolescents with Patellofemoral PaIn) was a randomised, controlled, participant- and assessor-blind feasibility trial with two parallel groups (1:1 allocation ratio). Detailed study methods are available in the protocol paper [17]. We consulted the SPIRIT 2013 statement [18] and the CONSORT 2010 statement extension to randomised pilot and feasibility trials [19] in designing and reporting the study. Guidelines for trials modified due to COVID-19 [20] were also consulted. The University of Queensland Human Research Ethics Committee granted ethics approval (approval number 2018000159). The trial was prospectively registered on the Australia New Zealand Clinical Trials Registry on 08/07/2019 (ACTRN12619000957190).
Participants
Adolescent volunteers were recruited from the community in Brisbane and the Gold Coast, Queensland, Australia. Several methods were used for recruitment, including advertising at community and school sporting events, and on websites and social media pages (Facebook, Instagram). Participants were also recruited from our existing database of PFP volunteers.
The inclusion criteria were: (i) aged 12–18 years; (ii) non-traumatic onset of anterior pain rated at least 3 on an 11-point numerical rating scale (0 = no pain, 10 = worst pain imaginable); (iii) pain aggravated by PFJ-loading activities (e.g., squatting, stair ambulation, running, jumping); and (iv) pain present for two months or more, and at some time during most weeks. Exclusion criteria were: (i) concomitant pain at sites other than the anterior knee (e.g. other knee structures, hip, lumbar spine); (ii) history of surgery on the ipsilateral limb or lumbar spine, or other suspected knee joint pathology (e.g. Osgood Schlatter’s Disease); (iii) planned lower limb surgery; (iv) recent PFP treatment (e.g. physiotherapy or knee joint injections in the last 3 months; foot orthoses prescription in the last 12 months); or (v) any foot condition impeding the prescription of foot orthoses.
All participants were required to provide written informed consent prior to participation. For participants under 18 years of age, their parent/guardian was also required to give written informed consent.
Sample size
Formal sample size calculations were not conducted [21, 22]. We estimated that 40 participants (20 participants per group) would allow us to observe practicalities of recruitment, acceptability of the shoe inserts, adverse events, dropouts, and sample variability.
Procedure
Eligibility was determined by one investigator (ICO, registered Physiotherapist) through two phases: (i) online screening (Research Electronic Data Capture [REDCap], Vanderbilt University, Nashville, USA); and (ii) physical screening at The University of Queensland to confirm the clinical diagnosis of PFP. Participants (and their parent/guardian if aged under 18 years) gave written consent and, after baseline measures were obtained, were randomised to receive either contoured foot orthoses or flat shoe inserts. We used simple randomisation, with the randomisation sequence generated by one investigator (KT) using a random number generator. The research assistant (KTO) held the randomisation sequence off-site to ensure concealed allocation, and was responsible for communicating group allocation to the physiotherapists. Study physiotherapists were unable to be blinded to group allocation due to the nature of the intervention. The primary investigator (ICO) was blinded to group allocation throughout the trial and collected primary outcomes. Secondary outcomes were self-reported by participants. Blinding of participants was maintained through communicating to each participant that there were two interventions that were being investigated, however, the two types of interventions and study hypotheses were not disclosed [11, 23].
Interventions
Participants attended physiotherapy appointments at one of six clinics across Brisbane and the Gold Coast for fitting of their allocated shoe insert. Study physiotherapists were trained in fitting procedures, as used in previous studies to maximise comfort [23]. Each participant received up to four pairs of inserts, fit to shoes that would accommodate the inserts and provide support. Participants were asked to attend up to three appointments to allow for any adjustments to be made and ensure inserts were comfortable. Each person was asked to wear the inserts as much as possible throughout the week. Participants in both groups received a handbook, which provided general information and advice about PFP and activity [17].
Prefabricated contoured foot orthoses
Participants assigned to the contoured foot orthoses group received prefabricated foot orthoses (Vasyli Medical, Labrador, Australia). The orthoses are manufactured from ethylene-vinyl acetate (EVA) with options for a high (hard, Shore A 70°), medium (Shore A 55°) and low (soft, Shore A 45°) density orthosis with inbuilt medial arch support and varus wedging. Density was selected based on comfort, and modifications were made to the orthoses in the form of heat moulding or wedges to achieve a comfortable fit [17], based on our published algorithm [23].
Flat shoe insoles
Participants assigned to this group received flat shoe insoles made of the same high-density EVA as the contoured foot orthoses, which were a uniform thickness of 3 mm along their length. To facilitate blinding regarding the true intervention, the contoured foot orthoses and flat insoles were covered in the same fabric and markings. Participants randomised to this group were advised that the intervention enhanced sensory feedback, and the insoles were heat moulded as necessary to enhance comfort [17].
Outcome assessment
At baseline, participant details were recorded, including demographics, knee/s affected, symptom duration and aggravating activities. Patient-reported outcome measures were collected at baseline, 6 weeks, and 3 months post-randomisation via an online platform (REDCap). For the duration of the study, participants were asked to complete a paper logbook to record details of daily activities, pain, and footwear [17]. Study physiotherapists recorded attendance, prescription notes and adverse events during fitting and follow-up appointments. 3 months was nominated as the primary endpoint.
The primary outcome was the feasibility of conducting a full-scale RCT. Feasibility was assessed by evaluating the following outcomes:
-
(i)
Number of eligible volunteers (from recruitment database).
-
(ii)
Willingness of participants to enrol in the study (from recruitment database).
-
(iii)
Recruitment rate (from recruitment database).
-
(iv)
Adherence with allocated intervention and logbook completion (from participant logbook).
-
(v)
Adverse events (from Study Practitioner notes, adverse events database, and participant logbook).
-
(vi)
Success of blinding (risk of performance and detection bias, from the Credibility and Expectancy Questionnaire) [24].
-
(vii)
Drop-out rate (from trial database; defined as participants who did not complete 3-month outcome measures).
Participants completed the Credibility and Expectancy Questionnaire [24] at baseline and at their second visit with the study physiotherapist (~ 2 weeks post-randomisation). The Credibility and Expectancy Questionnaire consists of six items in two sections; four items related to thoughts, and two items related to feelings. Credibility was derived from the first three thought items, and expectancy was derived from the fourth thought question and the two feeling questions [25]. The Credibility and Expectancy Questionnaire has been used in previous adolescent studies [26, 27].
Prior to the study, we set three criteria to determine the feasibility of a full-scale RCT: (i) a recruitment rate of 1 participant per week; (ii) adherence of at least 2 h of shoe insert wear per day for 5 days a week; and (iii) ≤20% drop out rate at 3 months [17].
The following secondary outcomes were collected by online questionnaires and participant logbooks (details provided in the protocol paper) [17]:
-
(i)
Usual and worst knee pain severity during a self-nominated aggravating activity, measured using a 100 mm visual analogue scale (0 mm = no pain and 100 mm = worst pain imaginable) [28].
-
(ii)
Knee injury and Osteoarthritis Outcome Score Child Version (KOOS-Child) [29].
-
(iii)
Knee injury and Osteoarthritis Outcome Score – Patellofemoral Subscale (KOOS-PF) [30].
-
(iv)
Global Rating of Change (GROC), using a 7-point Likert scale [13].
-
(v)
Patient Acceptable Symptom State (PASS) [31].
-
(vi)
Anterior Knee Pain Scale [32].
-
(vii)
EuroQOL-5D-5L. The UK time-trade-off scoring algorithm was used to weight each participant’s profile data to produce a single EQ-5D index score (as there are no data published in Australian adult or adolescent populations) [33].
-
(viii)
Use of co-interventions (e.g., pain medication, physiotherapy, knee brace, other footwear interventions).
Statistical analyses
Analyses were conducted by a blinded investigator (ICO) using SPSS software (IBM SPSS Statistics for Macintosh, Version 27.0, IBM Corp., Armonk, NY, USA). Baseline data were checked for normality using Shapiro-Wilk tests and presented as mean and standard deviation (normal distribution) or median and interquartile range (not normal distribution). Primary feasibility outcomes for this study were presented using descriptive statistics. Descriptive statistics were calculated for secondary outcomes at all timepoints, separately for each group.
Protocol deviations
There are two instances where our methods deviated from those reported in our protocol paper [17]. We were not able to report data from the Youth Quality of Life Short Form at any of the measured timepoints. This was due to an error with the online data collection platform. We also planned to provide a range of plausible estimates of treatment effects for prefabricated foot orthoses, compared to flat insoles, by reporting between-group differences (with 95% confidence intervals) for secondary outcome measures. Due to experiencing greater loss to follow-up than anticipated (including the impact of COVID-19), and the likely imprecision of between-group estimates from a small pilot trial [34], we elected not to report this. Instead, we only present descriptive statistics for secondary outcomes at each timepoint.