1. Introduction

Anterior knee pain (AKP), or patellofemoral pain syndrome, is a chronic musculoskeletal overuse condition presenting frequently to sports medicine and general practitioners.[1–8] It affects an individual’s ability to perform routine daily activities such as stair ambulation, walking and running, and thus impacts on work-related activities and participation in physical activity. Findings from prospective studies in active populations[9–11] reflect the chronicity of AKP, with a randomized controlled trial (RCT) reporting no recovery in half of the no-treatment control group at 12 months.[11] Furthermore, it is plausible that AKP in adults may precede the development of osteoarthritis in later years.[12]

Despite its prevalence, chronicity and impact, AKP remains one of the most challenging musculoskeletal conditions managed by practitioners.[13] Since greater pain severity and longer symptom duration are predictive of poor prognosis,[14] early effective management may be the key to limiting the longer-term impact of the condition. Considering the multifactorial nature of AKP,[15,16] the management approach should consider individual presentation and the contribution of local knee factors, as well as potential proximal (hip) and distal (foot) factors.

Surgical options for AKP appear to be inadequate.[17] This is highlighted by an RCT that revealed no additional improvement in AKP symptoms and function over 9 months when arthroscopic procedures based on pragmatically-identified abnormalities (e.g. resection of inflamed/scarred/excessive medial plicae or synovium, abrasion of chondral lesions, discision of lateral capsule, repair of meniscal tears) were added to exercise therapy.[18] Consequently, there is widespread consensus that nonsurgical interventions are the primary treatment of choice for AKP. However, in order to make informed decisions regarding optimal management, practitioners require up-to-date, high-quality evidence. Systematic reviews have drawn limited conclusions regarding RCT evidence for nonsurgical interventions for AKP.[19–26] From these, it appears that there is moderate evidence to support the short-term use of patellar taping for chronic knee pain;[19,26] however, meta-analyses conducted by one review collapsed findings for nonarthritic AKP and patellofemoral joint osteoarthritis.[26] Considering the recent increase in research output regarding AKP interventions, it is timely for an updated high-quality systematic review to assist practitioners in making informed, evidence-based decisions when managing AKP.[27] Therefore, a systematic review and meta-analysis, where possible, was conducted to evaluate the evidence for the short- and long-term efficacy of nonsurgical interventions for AKP.

2. Methodology

The study protocol was developed in consultation with guidelines provided by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) statement.[28]

2.1 Literature Search Strategy

Using guidelines provided by the Cochrane Collaboration,[29] a comprehensive search strategy was devised for the following databases: MEDLINE, EMBASE, CINAHL® and Pre-CINAHL®, PEDro, PubMed, SportDiscus®, Web of Science®, BIOSOS Previews®, and the full Cochrane Library. The MEDLINE search strategy, which was adapted for other databases, is presented in appendix 1 of the Supplemental Digital Content (SDC) http://links.adisonline.com/SMZ/A7. All publications listed up until 30 November 2009 were considered for inclusion, with no restrictions placed on the year of publication. Only full text, English language articles were included. Published abstracts were followed up for full-text publications of the study, but were not included as standalone papers. One investigator (NC) reviewed all the titles returned by the database searches, and retrieved suitable abstracts. Where abstracts suggested that papers were potentially suitable, the full-text versions were obtained and included in the review if they were found to fulfil the selection criteria. Reference lists of included papers and known published systematic reviews were hand searched to ensure the inclusion of all the available published evidence.

2.2 Selection Criteria

Studies were eligible for inclusion if they investigated participants who had an insidious onset of anterior or retropatellar knee pain aggravated by activities that load the patellofemoral joint (e.g. prolonged sitting or kneeling, squatting, jogging or running, hopping, jumping, or stair ascending/descending).[30,31] Studies were required to have investigated one or more nonsurgical interventions for AKP, compared with a control intervention, over a minimum period of 1 week. The design must have been a clinical trial that (i) followed participants over at least 2 weeks (deemed to be a clinically meaningful time period); (ii) utilized an outcome measure of pain, such as pain measured on an 11-point numeric rating scale or 100 mm visual analogue scale (VAS); and (iii) randomly assigned participants to intervention groups using methods defined by PEDro rating scale criterion 2 (see appendix 2 in the SDC).[32] Studies were excluded if they presented results reported in a previous publication or if their sole focus was chondromalacia patella verified by imaging or arthroscopy.

2.3 Assessment of Methodological Quality and Risk of Bias

The methodological quality of included studies was rated using a modified version of the PEDro rating scale[32] (see appendix 2 in the SDC). This scale has been used in previous systematic reviews with very high inter-rater agreement (kappa [κ] = 0.73–0.824).[33,34] Two independent reviewers (NC, LB), one of who remained blind to authors, affiliations and the publishing journal (LB), rated each study on 14 specific criteria. Final study ratings for each reviewer were collated and examined for discrepancies. Any inter-rater disagreement was discussed in a consensus meeting, and unresolved items were taken to an independent party (BV) for resolution. Once consensus was reached for all study ratings, overall quality scores were determined for each study by summing those criteria that had scored ‘yes’, providing a score out of 14.

Risk of bias was assessed using specific criteria from the modified PEDro scale that was selected based on the PRISMA statement[28] and the Cochrane Collaboration.[35] These were adequacy of randomization (criterion 2); allocation concealment (criterion 3); between-group baseline comparability (criterion 4); blinding of outcome assessors (criterion 7); adequate follow-up (>85%; criterion 8); and intention-to-treat analysis (criterion 9). A lack of blinding of participants and therapists was not considered to be a source of bias for these studies, as it is often not plausible to blind those providing and receiving interventions, such as physiotherapy and specific exercises, particularly when the efficacy of two different interventions is compared. The RCTs were classified as having a low risk of bias (score at least 5 of the 6 criteria), moderate risk (3 or 4) or high risk (≤2); the latter were excluded from further analysis.

2.4 Data Management and Statistical Analysis

Inter-rater reliability of the modified PEDro scores was evaluated using the κ statistic, where a κ of 1 represented perfect agreement, 0.9–1 almost perfect agreement, 0.7–0.9 very high agreement, 0.5–0.7 high agreement, 0.3–0.5 moderate agreement, 0.1–0.3 small agreement and 0–0.1 very small agreement.[36]

Standardized mean differences (SMD) with 95% confidence intervals (CIs) were used to represent effect sizes for pain, and were calculated using a random effects model in Review Manager (version 4.2).[37] For studies that reported outcome on multiple pain scales, participant-reported worst pain intensity over the previous week or a similar scale (e.g. pain with activity) was used. Data were grouped by follow-up time (0–6 weeks, 7–12 weeks, 13–26 weeks, >26 weeks), using the latest timepoint in each period. Where possible, mean change scores and standard deviations were extracted from papers to calculate SMDs. Alternatively, mean change scores were calculated by subtracting the follow-up score from the baseline score, and the standard deviation estimated by taking the average of the pre- and post-score standard deviations[38] or 95% CIs. When insufficient data prevented calculation of change scores, raw follow-up scores were used when groups were not significantly different on baseline measures from a clinical perspective (>15 mm on a 100 mm pain VAS[39,40] ). Authors were contacted by email for clarification or provision of additional data. Effect size magnitudes were interpreted as being nearly perfect (SMD ≥4), very large (2–4), large (1.2–2), moderate (0.6–1.2), small (0.2–0.6) and trivial (<0.2),[36] with positive values favouring the intervention of interest. Significance was set at p > 0.05. Data were pooled where studies investigated similar interventions, and had similar comparator interventions and timing of follow-up outcome measures.

Sensitivity analyses were conducted to confirm the exclusion of papers with a high risk of bias. Since studies with higher quality scores return findings of reduced efficacy of treatment,[41] Spearman’s correlation coefficients were calculated to determine the strength of the relationship between risk of bias (low, moderate, high) and effect size (SMD), and between modified PEDro scale score and effect size. The strength of correlations was determined using the same classification as for the κ statistic.[36]

3. Results

3.1 Search Strategy

The comprehensive search strategy identified 605 publications for evaluation beyond title (figure 1). The full text of 188 articles was retrieved, with 48 of these meeting the inclusion criteria.

Fig. 1
figure 1

Flow chart of the process and rationale used in selecting papers for inclusion in the review using a highly sensitive search strategy. AKP= anterior knee pain; RCT(s) = randomized controlled trial(s).

Full size image

3.2 Methodological Quality and Risk of Bias

The 48 studies scored widely on the modified PEDro scale, ranging from 2 to 13 of 14 (mean score 6.58) [see appendix 3 in the SDC]. Criteria satisfied by less than half of the included papers were allocation concealment (criterion 3, 29%), blinding (criterion 5, 8%; criterion 6, 2%; criterion 7, 41%), intention-to-treat analysis (criterion 9, 22%), justification of sample size (criterion 12, 37%), and reporting of adverse or side effects (criterion 14, 18%).

Initial inter-rater agreement on the modified PEDro criteria was very high (619 of 672 ratings; κ = 0.842). Consensus was reached on all items on initial discussion. Inter-rater reliability for individual criteria ranged from high (κ = 0.529) for criterion 11 to perfect (κ = 1.000) for criteria 1 and 2.

Twenty-one studies were considered to have a high risk of bias and were subsequently excluded from further analysis (see appendix 4 in the SDC for study characteristics). Sensitivity analyses revealed a significant moderate correlation between the risk of bias and SMD (r = 0.328; p = 0.004), and between modified PEDro score and SMD (r = 0.456; p < 0.000), supporting the exclusion of high-risk studies.

3.3 Findings

The 27 remaining studies were grouped by their primary intervention of interest (multimodal physiotherapy, manual therapy, exercise, tape, foot orthoses, electrotherapy, acupuncture, pharmacotherapy). Follow-up was predominantly within 3 months; only six studies followed participants beyond this. Table I presents characteristics of included studies with effect sizes for pain, and study conclusions when effect sizes could not be calculated.

Table I
figure Tab1

Summary of included studies (n = 27)

Full size image

3.3.1 Multimodal Physiotherapy

Evidence from a meta-analysis supports the use of multimodal physiotherapy in the short term. Pooled data from two studies that investigated identical multimodal physiotherapy programmes[30,31] showed a significant moderate effect for multimodal physiotherapy over a placebo intervention (flat inserts;[31] sham physiotherapy[30] ) at 6 weeks (SMD 1.08; 95% CI 0.73, 1.43) [figure 2].

Fig. 2
figure 2

Standardized mean difference (SMD) for improvement in pain with multimodal physiotherapy (MMP). SMD >0 favours MMP, <0 favours comparator. Ed= education; Ex = exercise; FO = foot orthoses.

Full size image

Evidence from individual RCTs largely supports the use of multimodal physiotherapy for AKP. When compared with placebo, there were significant moderate effects for multimodal physiotherapy at 12 weeks (SMD 0.69; 95% CI 0.23, 1.14), and significant small effects at 1 year (SMD 0.44; 95% CI 0.01, 0.88).[31] When multimodal physiotherapy was compared with a no-treatment or education control, the inclusion of more multimodal components appeared to increase its efficacy. Findings of Syme et al.[44] showed significant moderate effects favouring 8 weeks of manual therapy, stretches, vasti retraining and patellofemoral joint taping over no-treatment control (SMD 0.63; 95% CI 0.00, 1.26), but no significant effects were seen when only manual therapy, stretches and general lower limb exercises were used. However, there were contrasting effects found by Clark et al.,[42] who reported no significant effects when a combination of exercise, patellar taping and education was compared with education alone at 12 weeks and 1 year.

Effect sizes predominantly showed favourable effects for multimodal physiotherapy compared with other nonsurgical interventions, although this appeared to be associated with the timing of outcome measures. From the study of Collins et al.[31] there were significant small effects favouring multimodal physiotherapy over foot orthoses at 6 weeks (SMD 0.51; 95% CI 0.07, 0.95) and 12 weeks (SMD 0.45; 95% CI 0.01, 0.88), but no significant differences at 1 year. Furthermore, the addition of multimodal physiotherapy to foot orthoses when compared with foot orthoses alone produced significant moderate effects at all timepoints over 1 year (6 weeks: SMD 0.87; 95% CI 0.42, 1.32; 12 weeks: SMD 0.63; 95% CI 0.16, 1.07; 52 weeks: SMD 0.70; 95% CI 0.27, 1.14). Harrison et al.[43] compared multimodal physiotherapy (patella taping, vasti retraining) with a home exercise programme (general lower limb strengthening and stretching) with and without therapist supervision over 1 year. There was a significant small effect favouring the multimodal programme over supervised exercise at 4 weeks (SMD 0.56; 95% CI 0.00, 1.12). Interestingly, multimodal physiotherapy was not significantly different to supervised or unsupervised exercise at 12, 26 or 52 weeks.

When compared with placebo, multimodal physiotherapy used in conjunction with foot orthoses produced significant large effects at 6 weeks (SMD 1.45; 95% CI 0.96, 1.94), and significant moderate effects at 12 weeks (SMD 0.86; 95% CI 0.40, 1.33) and 52 weeks (SMD 0.77; 95% CI 0.33, 1.21).[31] However, the combined effects of multimodal physiotherapy and foot orthoses were not significant when compared with physiotherapy alone at 6, 12 or 52 weeks (figure 2).

3.3.2 Exercise

Evidence from individual RCTs supports the use of various forms of exercise for AKP (figure 3). Three studies showed significant effects favouring exercise over a no-treatment control.[11,45,47] Herrington and Al-Sherhi[45] compared two different 6-week programmes of knee extension exercises (closed kinetic chain; open kinetic chain) to no-treatment control. Effect sizes calculated from data provided by the authors showed significant large to very large effects favouring both types of exercise over control (closed chain: SMD 3.02; 95% CI 1.93, 4.11; open chain: SMD 1.82; 95% CI 0.95, 2.69). Effect sizes for Song et al.[47] showed significant moderate effects favouring leg press with hip adduction (SMD 0.83; 95% CI 0.26, 1.40) and standard leg press (SMD 1.01; 95% CI 0.43, 1.59) over control after 8 weeks. Similarly, findings of van Linschoten et al.[11] showed significant small effects for supervised exercise therapy over the control at 12 weeks (SMD 0.44; 95% CI 0.09, 0.78) and at 1 year (SMD 0.49; 95% CI 0.14, 0.83).

Fig. 3
figure 3

Standardized mean difference (SMD) for improvement in pain with exercise (Ex). SMD >0 favours Ex, <0 favours comparator. add= adduction; CKC = closed kinetic chain; Ed = education; FO = foot orthoses; MT = manual therapy; OKC = open kinetic chain.

Full size image

Interestingly, time effects were found among studies that compared closed and open kinetic chain strengthening exercises. Contrasting effects were found for two short-term studies, with Herrington and Al-Sherhi[45] showing a significant moderate effect favouring closed chain over open chain exercises at 6 weeks (SMD 1.01; 95% CI 0.25, 1.78), while a similar study by Bakhtiary and Fatemi[50] showed no significant effects at 5 weeks. In comparison, a 5-year follow-up study showed a significant small effect in favour of open chain exercises (SMD −0.57; 95% CI −1.14, 0).[53]

Two studies evaluated hip exercises as an addition to standard exercise programmes (figure 3). Findings of Nakagawa et al.[46] showed no significant effect on worst pain severity for the addition of hip abduction and external rotation exercises to standard quadriceps exercises. Similar outcomes were noted in the study of Song et al.,[47] with effect sizes showing no significant effects when leg press was performed in hip adduction compared with the standard leg press exercise.

Comparison of two groups from the study of Harrison et al.[43] investigated the effect of physical therapist supervision of a home exercise programme (figure 3). Effect sizes revealed no significant difference between supervised and unsupervised exercise at 4, 12, 26 or 52 weeks.

Three studies investigated the use of exercise as an adjunct to other interventions (figure 3). Effect sizes revealed no significant effect for the addition of exercise to education over 12 and 52 weeks,[42] to patellar mobilization/manipulation over 5 weeks,[48] or to foot orthoses over 4 or 8 weeks.[49]

3.3.3 Foot Orthoses

Evidence from one RCT supports the short-term use of foot orthoses over placebo (figure 4). Collins et al.[31] showed a significant small effect for foot orthoses over flat inserts at 6 weeks (SMD 0.59; 95% CI 0.15, 1.04), but no differences in effect at 12 weeks or 1 year. One study compared foot orthoses to exercise.[49] While effect sizes showed no significant difference at 4 or 8 weeks, sample size calculations provided by the authors indicated that the study was underpowered to detect significant effects.

Fig. 4
figure 4

Standardized mean difference (SMD) for improvement in pain with manual therapy (MT), tape, foot orthoses (FO), electrotherapy or acupuncture intervention. SMD >0 favours intervention, <0 favours comparator. Ed= education; EMG = electromyography; EMS = electric muscle stimulation; Ex = exercise; FKC = full kinetic chain; manip = manipulation; MMP = multimodal physiotherapy; MT = manual therapy; ST = soft tissue treatment.

Full size image

Effect sizes calculated from 4- and 8-week data provided by Wiener-Ogilvie and Jones[49] were not significant when foot orthoses were used in conjunction with exercise, compared with exercise alone (figure 4), although post hoc power calculations conducted by the authors suggested a high likelihood of a type II error.

3.3.4 Patella Taping

Evidence from one RCT supports the short-term use of patella taping (figure 4). Effect sizes from Whittingham et al.[54] showed significant large to very large effects favouring 4 weeks of patella taping and exercise over exercise alone (SMD 2.47; 95% CI 1.25, 3.70), and over placebo tape with exercise (SMD 1.35; 95% CI 0.36, 2.35). In contrast, longer-term data from Clark et al.[42] showed no significant between-group effects when patella taping and education was compared with education alone, and when patella taping was added to exercise and education (12 and 52 weeks).

3.3.5 Acupuncture

Evidence provided by one RCT supports the use of acupuncture in AKP (figure 4). Effect sizes calculated from data provided by Jensen et al.[63] showed a significant moderate effect favouring acupuncture treatment over control at 5-month follow-up (SMD 0.65; 95% CI 0.13, 1.16). Although outcome measures were also taken at two other timepoints over 1 year, only the acupuncture group was followed up at 6 weeks, and pain severity was not recorded at 1 year.

3.3.6 Manual Therapy

All three studies that investigated manual therapy techniques compared with control or in conjunction with other interventions revealed no significant findings (figure 4). van den Dolder and Roberts[55] showed no significant effect for 2 weeks of treatment with medial glide and tilt mobilizations and local lateral retinacular massage when compared with a no-intervention control. Similar outcomes were found for joint manipulation. Findings of Brantingham et al.[56] showed no significant effect at 6 or 14 weeks for either knee manipulation or full lower limb kinetic chain manipulation when each was added to exercise and soft tissue treatment. In the study by Stakes et al.,[57] there was no significant effect for the addition of spinal manipulation to patellar mobilization over 4 weeks.

3.3.7 Electrotherapy

Evidence from meta-analysis does not support the use of electromyography (EMG) biofeedback in addition to an exercise programme (figure 4). Pooled data from two studies[59,60] showed no significant benefit of using EMG biofeedback with exercise at 4 weeks (SMD −0.21; 95% CI −0.64, 0.21), or at 8 to 12 weeks (SMD −0.22; 95% CI −0.65, 0.20).

Electric muscle stimulation (EMS) was investigated by two studies (figure 4).[61,62] Effect sizes from Akarcali et al.,[61] who evaluated the addition of high voltage pulsed galvanic stimulation to an exercise programme, showed no significant effect for either group at 6 weeks. Callaghan and Oldham[62] compared different EMS devices, finding no significant effects for either simultaneous mixed frequency or sequential mixed frequency EMS. One study compared low-level laser treatment to sham laser over 5 weeks;[58] however, insufficient data was provided for effect size calculation.

3.3.8 Pharmacotherapy

One study investigated pharmacological interventions for AKP,[65] comparing 1 week of NSAIDs with placebo; however, the authors did not report data for effect size calculation.

4. Discussion

The comprehensive search strategy identified a wide variety of conservative interventions for AKP that have been investigated by RCTs. Meta-analyses of pain data, the primary symptom of this condition, provides evidence for multimodal physiotherapy when compared with a placebo over 6 weeks, as well as evidence of no additional benefit in adding EMG biofeedback to exercise over 12 weeks. As a result of a lack of further opportunities for data pooling, the best evidence for other interventions, such as exercise, patella taping, foot orthoses and acupuncture, comes from individual RCTs.

Findings from this review indicate that multimodal physiotherapy, compared with a placebo intervention that controls for therapist-patient interaction effects, has the best evidence for reducing AKP using a nonsurgical approach. Importantly, the magnitude represents a clinically meaningful improvement in pain, with the weighted mean difference of 20.3 mm of 100 exceeding the minimal clinically important difference.[39,40] The two studies incorporated in the meta-analysis received the highest ratings on the modified PEDro scale, scoring 13[30] and 12[31] of 14. Importantly, identical multimodal programmes were used, and incorporated interventions targeting local factors (patellar taping, patellar mobilization, vasti retraining with EMG biofeedback), proximal factors (gluteal strengthening) and global factors (lower limb stretches). Syme et al.[44] also included these components in their multimodal programme, which was more efficacious than no treatment. In contrast, the multimodal programmes of Clark et al.[42] and Harrison et al.,[43] who largely reported nonsignificant differences to comparators, only utilized local interventions and did not target proximal or distal factors. This suggests that the inclusion of interventions that target specific proximal factors in conjunction with local interventions,[66] such as utilized by Crossley et al.[30] and Collins et al.,[31] may be the key to ensuring success in reducing AKP symptoms. Importantly, these findings highlight the multifactorial nature of AKP,[67] and the need for practitioners to use clinical judgement to address all necessary lower limb factors to effectively manage this condition. However, more RCTs are required to facilitate further meta-analyses, particularly to compare multimodal physiotherapy with a no-treatment control, and to evaluate the effects of proximal interventions used in isolation. Furthermore, considering that only one study investigated the long-term outcomes of multimodal physiotherapy,[31] this should be considered in future studies.

Findings of the three studies that compared exercise with wait-and-see indicate that a predominantly quadriceps-based programme is more effective than no treatment.[11,45,47] However, it appears that the addition of hip components, supervision or other adjunct interventions to quadriceps-based programmes does not change AKP outcomes. It is important to highlight that these studies tended to lack specific vasti retraining, instead aiming for general quadriceps strengthening. This may explain differences in effects of combined treatments when multimodal physiotherapy outcomes are compared with those of exercise studies. Furthermore, these findings highlight the importance of targeted exercise programmes based on sound clinical assessment, with consideration of additional components and interventions as necessary. While it is difficult to draw conclusions regarding direct comparison of open and closed kinetic chain exercises, the greater emphasis on closed chain exercises in the other exercise and multimodal programmes suggests that this is the preferred clinical approach, and fits with evidence of greater vasti activity during closed than open kinetic chain exercises.[68]

It should be noted that the apparent lack of evidence for other interventions, such as pharmacotherapy and manual therapy, does not imply that these interventions are not effective; rather, it highlights deficits in the current literature with respect to study methodology, as well as the need for more high-quality RCTs. Small sample sizes utilized by some of these studies may have increased the risk of a type II error. It is also important to consider those interventions that were not sufficiently represented in the 27 studies, such as knee braces and trunk muscle retraining. Indeed, six studies evaluated the efficacy of knee and patella braces,[69–74] but were excluded because of a high risk of bias, while only one study incorporated transversus abdominus retraining into their exercise programme.[46] Importantly, no studies were found that investigated interventions targeted to individual participant presentations, such as core stability deficits. Thus, these simple and commonly used interventions require further investigation in high-quality RCTs.

This is the first systematic review to incorporate meta-analyses of data from RCTs investigating conservative interventions solely for nonarthritic AKP. It also considers 23 RCTs of low-to-moderate risk of bias that were not included in the most recently published systematic review of all nonsurgical interventions,[21] or those that have been published since. New evidence from pooled data was found regarding multimodal physiotherapy and EMG biofeedback. However, other nonsurgical interventions including pharmacology require ongoing investigation. This is particularly important considering the role that early effective intervention, aimed at reducing pain severity and duration, may play in limiting the longer term impact of AKP.[14]

Despite the strengths of this systematic review, there are limitations that need to be considered when interpreting findings. Studies were not eligible for inclusion if they were published in non-English languages. While it is arguable that authors of high-quality RCTs would aim to widely disseminate their findings via high-impact journals published in the English language, it is plausible that this may influence the outcomes of analyses. Furthermore, while the use of a rigorous and systematic methodology limits the influence of potential biases, assessment of publication bias was not conducted. Considering that publication bias has been reported for studies investigating patellar taping and bracing for chronic knee pain,[26] it is possible that publication bias also exists among studies of other interventions for AKP, and only highlights the need for further high-quality RCTs to ensure that the literature is characterized by more balanced findings.

A number of methodological issues were identified among the included studies that should be addressed in future AKP RCTs. First, the 48 studies initially rated for their methodological quality had a mean modified PEDro rating of less than half of the total possible score, and almost half of these studies were excluded from further analysis because of a high risk of bias associated with the study design or inadequate reporting. In order to enhance the quality of published studies on AKP, and maximize the potential for consolidation of findings in systematic reviews and meta-analyses, future RCTs should utilize CONSORT (Consolidated Standards of Reporting Trials) guidelines[75] during the methodological design phase and when reporting study findings. This would also address the inconsistencies and inadequacies with reporting outcome data that were observed, and the subsequent effect that this has on the calculation of effect sizes and meta-analyses. Second, participant numbers were generally low, with final group sizes below 30 for the majority of studies. Only 37% of the 48 studies reported sample size calculations, which suggests that they may not have been adequately powered to show significant between-group differences. Third, only six studies that were included in final analyses investigated treatment effects beyond 3 months. In the context of a chronic condition such as AKP,[9] studies of longer duration are required to determine the long-term efficacy of interventions.

While the aim of this systematic review was to investigate the effect of interventions on pain, future RCTs should consider using a range of outcome measures that also address other signs and symptoms of AKP. The Kujala Patellofemoral Score[76] that encompasses pain, function and other symptoms, has been shown to be reliable, valid and responsive in AKP,[40] and predictive of short- and long-term outcome.[14] Indeed, it was used as an additional outcome measure in 11 of the 48 studies initially included.[11,14,40,45,51,53,56,62,69,77,78] Ratings of global improvement[31] provide an overall opinion of treatment effects on the condition as a whole, and can be represented by clinically meaningful statistics, such as relative risk reduction and numbers needed to treat. More widespread use of such measures would facilitate further between-study comparisons and meta-analyses involving dimensions other than pain.

5. Conclusions

Pooled data from a limited subset of studies supports the use of multimodal physiotherapy incorporating proximal as well as local interventions to reduce AKP in the short term, but does not support the addition of EMG biofeedback to exercise. Because of a lack of further opportunities for data pooling, individual RCTs provide the best evidence for other interventions, such as exercise, patella taping, foot orthoses and acupuncture. Until further high-quality RCTs are conducted addressing issues of sample size, long-term follow up and adherence to the CONSORT statement, sports medicine practitioners should prescribe local, proximal and distal components of multimodal physiotherapy for appropriate AKP patients, but also consider foot orthoses or acupuncture as adjunct or alternative interventions.