The effect of targeted treatment on people with patellofemoral pain: a pragmatic, randomised controlled feasibility study

The study was a pragmatic, randomised controlled feasibility study in which participants were selected from an on-going longitudinal cohort study. Twenty-six participants were identified from larger group of 70 PFP cases on the basis of having hip abductor weakness at clinical examination and were randomised into receiving either a matched treatment (MT) or UC in a 1:1 ratio. Ethical approval was obtained prior to commencement of the study (14/NE/1131). All participants completed written informed consent prior to entering the study. The study has been reported in accordance with Consolidate Standard of Reporting Trials (CONSORT) [31] and Template for Intervention Description and Replication guidelines. (TiDieR) [32].

Participants were recruited between November 2014 to April 2016 from a large musculoskeletal and rehabilitation service via clinician referrals, their SystmOne database (a local electronic healthcare database), posters displayed in the local hospital and an university alumni volunteers website. Eligibility criteria were assessed both verbally and clinically to ensure that the inclusion criteria were addressed fully (Table 1). The most symptomatic knee was designated the index limb. Participants were stratified based on hip abductor strength measured using a Biodex isokinetic system 4 (IRPS Mediquipe, UK). Hip abductor weakness was based on thresholds defined a priori from age and gender normative data [33] (Table 1). The relevant normative mean minus one standard deviation (−1 SD) was used as the threshold for allocation to the “weak” stratum based on previous recommendations [34].

The study was designed to recruit 12 participants per group based on previous guidance for feasibility studies of this design [35]. Participants were followed up to 8 weeks post-intervention as this has previously shown to be sufficient time to demonstrate an effect in PFP [20, 27, 36].

The random allocation sequence was made according to the output from a random number generator and concealed within pre-sealed, opaque envelopes [37]. All allocation and randomisation was conducted by the lead author (BD).

The outcome assessor was unblinded, however, patient reported outcome measures (PROMs) were completed in a separate room with no input from the assessor. The biomechanical outcomes were acquired in accordance to a strict study protocol to minimise variation and bias [38]. Furthermore, the biomechanical outputs are automated so the lack of blinding is less of an issue.

Participants randomised to the MT group were asked to attend six supervised sessions of approximately 30 min in duration once per week for 6 weeks at a local hospital. Each week they also performed two additional sessions on non-consecutive days independently at home, with the intervening days allowing adequate rest [39]. Consideration was made to the recommended determinants of resistance exercise [40] when developing the study intervention. The sessions were face to face, 1:1 sessions provided by the lead author (BD) a senior musculoskeletal physiotherapist with over 8 years of clinical experience. During these sessions, participants were given education and justification of the treatment followed by three exercises aimed at targeting coronal, sagittal and transverse strength of the hip using resistance bands. Each week at least one of the exercises would change with the aim of providing variation and minimising tedium [41]. The supervision sessions served as a means of ensuring both treatment fidelity and tailoring. Fidelity was ensured by checking the exercise technique and making corrections to performance prior to these being performed independently at home. Subsequent visits ensured this instruction had been correctly applied or not. Tailoring the intervention based on progressive loading was in line with current recommendations [25]. Participants were issued yellow (least resistance), red or green (most resistance) resistance tubing (66fit Ltd. ™) and were allowed to take it home. To progress the load and resistance, a Borg Rate of Perceived Exertion scale (RPE) [42] was used based on the recommendations when using resistance band [43]. A RPE of >6 was considered desirable [39] and participants were monitored after a few repetitions to ensure this was what was being achieved. As participants were stratified for strength, the intervention required participants to perform 10 repetitions within three sets as recommended for strength training [39]. Participants were advised to ensure the time under tension was 8 s (3 s concentric, 2 s isometric hold and 3 s eccentric contraction). Strengthening was performed on each leg alternatively providing a standardised rest between sets. Exercise diaries (see Additional file 1) were issued to participants to provide a reminder of the exercises and to allow a measure of adherence. Participants were asked to document each time each exercise was performed on their diary sheet and return these at each visit.

Participants randomised to the UC group continued with the same management of their condition as they were planning to receive prior to the commencement of the study. This included planned physiotherapy, podiatry or no intervention, depending upon participant preference. The type of management and number of sessions was recorded for the UC group at follow-up.

In response to lack of agreed guidelines for outcomes in feasibility studies [44], the primary feasibility outcomes were adapted from recommendations made by Bugge et al. (2013) [45] and Shanyinde et al. (2011) [46].

Statistical analysis was undertaken using SPSS (version 21.0 (Armonk, NY: IBM Corp). As hypothesis testing is not advised for this size and type of study design [44], descriptive statistics along with point estimates, confidence intervals and effect sizes were presented for all PROMs and biomechanical outcomes. Within-group changes for all kinematic variables were expressed as a percentage change of the total ROM. Feasibility outcomes were described using descriptive statistics. To determine, where possible, a quantifiable measure of the feasibility outcomes, predetermined thresholds were used to indicate either success or strategies required (Table 2). Where it was not possible to use quantitative data to demonstrate success, outcomes were reported narratively.

Figure 2 shows that 14 participants were randomised to MT and 12 participants to UC. Of the participants in the UC group, 55% received formal physiotherapy treatment, which may or may not have included a strengthening component. The remaining UC participants reported continuing with their normal self-management.

The results from the mechanistic outcomes are shown in Table 6. Evaluation of the peak torque measures showed that both MT and UC groups showed an increase in peak hip abductor torque from baseline to follow up but no evidence of a systematic effect between groups was observed (−0.63 Nm; 95% CI: -13.35, 12.09). In terms of peak knee extensor torque, the UC group showed a much larger increase yielding a MD of 7.96 Nm (95% CI -2.88, 18.79; d = 0.624).

The between-group comparisons of the kinematics showed that the MT group had a reduction in peak IR whereas the UC had a slight increase (1.70°; 95% CI: −2.56, 5.97) yielding a small effect size (d = −0.34). Both MT and UC groups showed an increase in peak ADD (−0.17° vs. -0.04° respectively). Coronal ROM showed that the MT group had a reduction whereas the UC group showed a slight increase (1.12°; 95% CI: −0.72, 3.06) yielding a medium effect size (d = −0.53). Transverse ROM showed an increase in both the MT and UC groups (−0.32° vs. -0.78° respectively.

The within-group comparisons of the kinematic outcomes are presented in Fig. 3. The MT intervention led to a reduction in peak IR of 13.1% of the total transverse ROM. There was a small reduction in coronal ROM (4.8%) whilst peak ADD and transverse ROM demonstrated a small increase. The UC group demonstrated an increase for all kinematic variables.

Based on our eligibility thresholds for selecting a ‘weak’ hip group, we predetermined that for feasibility; eligibility should reach or exceed 32%. Our observed eligibility rate of 37% provides reassurance. It is also expected that other people with PFP, ineligible for the current study based on hip strength, may well classify into other subgroups as shown in recent studies [29]. This eligibility rate for hip weakness is less than the 88% (at 1SD) reported by Selfe et al. (2016) [29] but may be explained by the current study measuring an isokinetic contraction rather than an isometric contraction and as a consequence the different strength thresholds applied. Furthermore in order to minimise potential bias, future multicentre RCTs would need to ensure cross-site calibration of the isokinetic systems and site visits to monitor fidelity of the testing procedures and the intervention.

Recent studies have reported that a greater adherence to treatment is associated with increased probability of better outcomes [2]. An adherence to treatment and adherence to appointments over 90% is promising. Approximately 30% (4/13) achieved complete adherence to all treatment sessions and 9% of appointments required rearranging. Our adherence rate is comparable to a larger RCT [18] who found a 80.3% adherence rate for a 6 week hip strengthening in PFP. It’s anticipated that rearranging appointments for participants would be more challenging for a larger sample over multiple sites. Consequently, strategies to enhance adherence with the use of activity monitoring technology and reminder services need to be considered [56].

No differences between groups were found for either the average or worst NRS values. This might be explained by the difference in almost a score of one in average baseline NRS, a feature that would likely be minimised in a larger full-scale trial. Previous RCTs [18, 57] have also used eligibility criteria requiring a minimum NRS score of three out of 10 pain score. Setting a minimum pain score as part of the inclusion criteria is suggested for a future RCT.

The difference between groups for the AKP score did not reach the predetermined minimal clinically important difference (MCID) of eight points [58], although there was a trend towards a meaningful benefit. These findings are similar to the only other RCT to have stratified a PFP cohort, used in a study of a foot orthotic intervention over 6 weeks. Mills et al. (2011) [12] selected their participants based on predictors shown to predict success with orthotics, which included age, height, baseline pain severity and a static foot measure. They also found a significant difference between groups in terms of GROC with no differences in AKP score or VAS pain. They suggest that GROC is able to capture the multidimensional nature of PFP (characterised by pain, disability and functional limitation) compared to AKP score and VAS pain which are more one dimensional [12].

Stratifying for hip abductor weakness led to a reduction of approximately 13% for peak IR in the MT group following strengthening treatment. This is important considering that an increase peak IR has been associated with PFP during stair descent [59, 60]. This reduction in peak IR occurred with a slight worsening of their transverse ROM suggesting that following treatment, people in the MT group were initiating stance phase in a more desirable externally rotated hip position. A reduction in peak IR and a slight increase in peak ADD are perhaps surprising considering that participants were stratified for hip abductor weakness. However, recent strength measures conducted on 501 healthy athletes [61] have shown that hip abductor and hip external rotation strength are highly correlated (r = 0.66) indicating this subgroup were likely to have also demonstrated weakness into both hip abduction and external rotation.

Hip strength increased in both groups by a similar difference, which is likely the result of over a third of participants in the UC group being engaged in physiotherapy. Post-hoc analysis of those participants in the UC group who received no treatment show an increase of only 3.9 Nm in hip abductor strength (results not shown). In the MT group the change in hip strength was 9%, which is a comparable improvement to previous hip strengthening programmes over a similar training duration [21, 27]. The increase in strength for both groups but with only kinematic improvements seen in the MT group might suggest that strength, on its own, cannot explain the improvement seen in peak IR. Direct comparison with previous studies [17, 26‐28] remains difficult due to the differences in assessment tasks (e.g. running, stairs etc.) and the specific kinematic outcomes (e.g. peak, average angles etc.) investigated. Previous studies [27, 28] that have observed the effect of hip strengthening on running kinematics in people with PFP found no change in kinematics despite increases in hip abductor strength. Only Baldon et al. (2014) [17] reported changes in kinematics, during a single leg squat, following a hip strengthening programme. Yet, this training programme, did include constant feedback on lower limb alignment which suggests a more movement retraining approach [62] rather than pure strength training. It remains possible that the improvements observed in peak IR in the current study was the result of using progressive loading within a tailored treatment regime and selecting participants who were most likely to benefit from strengthening.

This feasibility study presents with several limitations. Firstly, the study was performed in a single centre. Future RCTs would be required to be multicentre to improve generalisability, which is anticipated to introduce new feasibility issues. The findings of the current study would, however, inform the documentation of standard operating procedures in terms of recruitment; data collection and intervention provision to ensure any future study could be operationalised across different geographical locations. Secondly, the current study did not blind assessors to group allocation, which could lead to potential bias [63]. Every effort was made for participants to complete PROMs in isolation and objective biomechanical outcomes were acquired in accordance with strict protocols with little chance of introducing bias. Future RCTs should make every effort to introduce outcome assessor blinding and consider measuring the level of this outcome assessor blinding [64] Thirdly, the use of a UC group was intended to represent the heterogeneity of management available in real, daily practice and thus improve the external validity [65]. For the purposes of exploring the mechanism of hip strengthening, however, the fact that over half of the UC group received physiotherapy input potentially dilutes the between-group findings. Comparison with a control group receiving no active intervention would remedy this issue.