Foot orthoses for patellofemoral pain in adults

Abstract

Background

Foot orthoses, which are specially moulded devices fitted into footwear, are one of the treatment options for patellofemoral or anterior knee pain.

Objectives

To assess the effects of foot orthoses for managing patellofemoral pain in adults.

Search methods

We searched the Cochrane Bone, Joint and Muscle Trauma Group’s Specialised Register (March 2010), the Cochrane Central Register of Controlled Trials (The Cochrane Library 2010, Issue 1), MEDLINE (1950 to March 2010), EMBASE (1980 to 2010 Week 11), CINAHL (1937 to March 2010), trial registers, reference lists and grey literature. No language restriction was applied.

Selection criteria

We included randomised or quasi‐randomised clinical studies that compared foot orthoses with flat insoles or another physical therapy intervention. The primary outcomes were knee pain and knee function.

Data collection and analysis

Two authors independently selected eligible trials, assessed methodological quality and performed data extraction. We calculated risk ratios and 95% confidence intervals for dichotomous variables, and mean differences with 95% confidence intervals for continuous variables. We pooled data using the fixed‐effect model.

Main results

Two trials with a total of 210 participants were included. Both trials were at some risk of performance bias. One trial had four intervention groups and the other had three. One trial found that foot orthoses when compared with flat insoles (control group) had better results at six weeks in knee pain (participants with global improvement: risk ratio 1.48, 95% confidence interval 1.11 to 1.99), but not at one year follow‐up. Participants in the orthoses group reported significantly more minor adverse effects (e.g. rubbing, blistering) compared with the flat insole group (risk ratio 1.87, 95% confidence intervaI 1.21 to 2.91). Both trials in their comparisons of orthoses plus physiotherapy versus physiotherapy alone found no statistically significant differences between the two intervention groups in knee pain or function. Results for knee pain outcomes did not show significant differences between foot orthoses versus physiotherapy. Although participants in the physiotherapy group had consistently better results for the functional index questionnaire, the clinical relevance of these results is uncertain.

Authors' conclusions

While not robust, the available evidence does not reveal any clear advantage of foot orthoses over simple insoles or physiotherapy for patellofemoral pain. While foot orthoses may help relieve knee pain over the short term, the benefit may be marginal. Patients treated with orthoses are more likely to complain of mild adverse effects and discomfort.

Plain language summary

Moulded foot insoles for adults with pain around the knee cap

Pain around the knee cap is a common problem. The pain may be brought on or made worse by day to day or sporting/exercise activities. Pain around the knee cap can have many different causes, such as the way the knee cap glides over the bones or because of knee overuse. Several different treatment options are available. Foot orthoses are specially moulded devices fitted into footwear. They are believed to be helpful because they might help improve the alignment of the leg bones.

The aim of this review is to evaluate the effects of foot orthoses on knee pain and knee usage in adults with pain over the front of the knee. We aimed to compare foot orthoses against no treatment or flat insoles, or other treatments such as physiotherapy.

We included two studies with a total of 210 participants in this review. Both trials were at some risk of bias because not enough care had been taken to ensure that groups received the same treatment other than the interventions being tested. One trial found some benefits from using foot orthoses over simple insoles at six weeks but not at one year. Participants wearing orthoses were, however, more likely to report minor adverse effects (e.g. rubbing, blistering) and discomfort compared with those wearing insoles. There were no important differences in knee pain and function in people given foot orthoses as well as physiotherapy when compared with people given physiotherapy only. Results for knee pain and function did not show important differences between foot orthoses versus physiotherapy.

On the basis of the available evidence we do not recommend foot orthoses for adults with pain around the knee cap.

Background

Description of the condition

Patellofemoral joint (PFJ) pain, or anterior knee pain, is a common musculoskeletal condition which frequently leads to medical consultation and causes significant morbidity (Collins 2008). The incidence of PFJ pain varies according to the activity level. The true incidence in the general population is not known but 26% of young adults presenting at a sports injury clinic over a two‐year period were found to suffer from PFJ pain (Brosseau 2001). People usually present with pain over the front of the knee. The pain is provoked by activities such as climbing or descending stairs, squatting, sitting for prolonged periods and running. The aetiology of PFJ pain is not well understood but patellar maltracking or overuse are believed to be contributing factors. Anterior knee pain may arise from a variety of pathologies including patellofemoral arthritis, patellofemoral instability, and tendinopathy. The term 'PFJ pain' refers specifically to a clinical presentation where there are none of the signs associated with these pathologies or of acute knee injury.

The PFJ is a complex joint with a complex movement pattern. There is some evidence to support the notion that the genesis of anterior knee pain lies in structural and biomechanical abnormalities. It has been proposed that PFJ pain may develop because of abnormally directed loads on the knee that exceed the normal physiologic tissue thresholds (Arendt 2005). Anomalous loading of the joint may occur due to abnormalities of bony alignment, joint geometry, soft tissue constraints, neuromuscular control or functional demands.

Description of the intervention

There is no consensus about what is the best management of this common clinical condition. Non‐surgical treatments are often the first mode of management for most people with PFJ pain. A wide range of treatments are used. These include foot orthoses, patellar taping, knee supports, physiotherapy, analgesics, non‐steroidal anti‐inflammatory drugs, and surgery. Foot orthoses are specially moulded devices fitted into footwear. In contrast, normal insoles are generally flat. Foot orthoses can be off‐the‐shelf or customised to individual foot shapes. A plain insole costs much less than off‐the‐shelf or customised foot orthoses.

How the intervention might work

Malalignment of the lower limb, resulting in compensatory malpositioning of the foot is thought to contribute to PFJ pain (Heintjes 2003). It has been suggested that moulded foot orthoses might help relieve PFJ pain by correcting the foot positioning during gait or altering lower limb mechanics (Bates 1979; Collins 2008; Cornwall 1995; Johnston 2004). Others have suggested that foot orthoses might act by absorbing shock, preventing excessive pronation and abnormal foot motion, as well as by enhancing balance and proprioceptive performance (Sahar 2009). However, it is unclear whether these actions translate to actual clinical improvement. Although there is some evidence that foot orthoses may reduce rearfoot movement and affect knee kinematics during certain phases of gait, the effects on knee symptoms are not clear (Eng 1994). Others have contended that orthoses act as a space filler, allowing full plantar contact and acting by reducing muscle activity and energy expenditure (Vicenzino 2008a).

Why it is important to do this review

Cochrane reviews addressing the efficacy of therapeutic ultrasound (Brosseau 2001), exercise therapy (Heintjes 2003) and pharmacotherapy (Heintjes 2004) for the treatment of PFJ pain have been completed. These reviews show that evidence for the effectiveness of these interventions is limited and conflicting. A previous Cochrane review, now withdrawn from The Cochrane Library, examined the benefits of knee orthoses for PFJ pain and concluded that the evidence was limited and weak (D'hondt 2002). D'hondt 2002 did not analyse the effectiveness of foot orthoses. Nevertheless foot orthoses seem to offer a logical and widely available treatment option for PFJ pain. Therefore a synthesis of the evidence for the use of foot orthoses and to identify gaps in the research evidence was indicated.

Objectives

To assess the effects of foot orthoses for patellofemoral joint pain in adults. More specifically, we planned to evaluate the effectiveness of foot orthoses compared with placebo, analgesia, no treatment, or other orthoses (i.e. knee) for PFJ pain.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised or quasi‐randomised (method of allocating trial participants to a treatment which is not strictly random; e.g. by date of birth, hospital record number, alternation) controlled studies. 

Types of participants

We included studies of skeletally mature adults (aged 18 years or over).  We excluded trials of adolescents because anterior knee pain in adolescents is often attributed to abnormalities of skeletal growth (e.g. Osgood Schlatter’s disease or Sinding‐Larsen syndrome). We only included mixed population trials of adults and adolescents if we could obtain separate data for the adult participants or where the proportion of adolescents was clearly low (under 10%) and balanced between the two groups. Expert clinical judgement alone was deemed to be adequate for the diagnosis of PFJ pain; thus, radiological examination of the knee was not essential.

Types of interventions

For inclusion, studies had to have at least one intervention arm where foot orthoses were compared with no treatment, placebo, analgesia or another orthotic device. Examples of orthoses are off‐the‐shelf orthoses or custom‐made foot orthoses. We extended the scope of the review to include comparisons of foot orthoses plus another physical therapy intervention versus the same physical therapy intervention on its own; and comparisons of foot orthoses versus another physical therapy intervention.

Types of outcome measures

Primary outcomes

• Knee pain as quantified by a visual analogue scale or any other validated method, preferably at one year or more follow‐up.

• Knee function (any validated and commonly used knee function score, e.g. WOMAC score, IKSS score), preferably at one year or more follow‐up

Secondary outcomes

• Patient satisfaction assessed using a Likert scale, visual analogue scale (VAS) or any other validated score.

• Patient reported quality of life score: e.g. SF 36 or EQ5D or any other validated score.

• Use of analgesics and/or non‐steroidal anti‐inflammatory drugs.

• Adverse effects including foot pain, discomfort, or difficulty in walking and any other important adverse effect reported in the trials.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Bone, Joint and Muscle Trauma Group Specialised Register (March 2010), the Cochrane Central Register of Controlled Trials (The Cochrane Library 2010, Issue 1), MEDLINE (1950 to March Week 2 2010), EMBASE (1980 to 2010 Week 11), CINAHL ‐ the Cumulative Index to Nursing and Allied Health Literature (1937 to March 2010), the Turning Research into Practice (TRIP) database (to 29/03/10), and PEDro ‐ the Physiotherapy Evidence Database (to 29/03/10).

In MEDLINE, the subject‐specific search (lines 1 to 18) was combined with the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE: sensitivity‐maximizing version (Lefebvre 2009), and in EMBASE with the Scottish Intercollegiate Guidelines Network (SIGN) trial search strategy (seeAppendix 1 for all strategies).

We also searched the WHO International Clinical Trials Registry Platform and Current Controlled Trials for ongoing trials.

We did not exclude studies on the basis of language or publication status.

Searching other resources

We checked reference lists of articles and also searched for unpublished studies via contact with firms producing orthotic devices. We also contacted authors and experts to identify further studies. We searched for conference proceedings via British Library's Zetoc table of contents database of conference proceedings. We also searched the ProQuest database for theses and the Opensigle database for unpublished studies (last search on 29/03/10).

Data collection and analysis

Selection of studies

Two authors (MH and PA) independently screened search results and selected studies for possible inclusion, based on titles and abstracts. A final decision on inclusion was made from the full text. Disagreement was resolved by a third author (PJ). We contacted authors for additional information where necessary.

Data extraction and management

Two authors (MH and PA) independently extracted the data using a data extraction form. The data extraction form was piloted on the first trial and any necessary changes made. A third author (PJ) resolved disagreements. We contacted authors for additional information and data.

Assessment of risk of bias in included studies

Two authors (MH and PA) independently assessed the risk of bias of included studies using The Cochrane Collaboration’s 'Risk of bias' tool (Higgins 2009). The domains of this tool include: adequate sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other potential sources of bias such as that arising from major imbalances in baseline characteristics and important differences in care programmes other than the interventions under test. Any disagreement in assessment was resolved by a third author (PJ). Assessors were not blinded to author or institution. We contacted trial authors for clarification when there was uncertainly about trial methodology. The results of the assessment are displayed in the risk of bias table and figures, and summarised in the text.

Measures of treatment effect

For dichotomous outcomes we calculated risk ratios and 95% confidence intervals (95% CI). We used the mean difference (and 95% CI) as the effect measure for continuous outcomes. We would have used the standardised mean difference (and 95% CI) where data were pooled from studies reporting comparable, but different, continuous outcome measures.

Unit of analysis issues

We were alert to potential unit of analysis issues such as where there is disparity between the unit of randomisation and analysis for individuals with bilateral involvement. This did not arise in the current review and we also did not include any study with a cross‐over design.

Dealing with missing data

Where possible, we attempted to perform intention‐to‐treat analyses that included all people randomised to the intervention groups. Where we could not obtain missing data from the authors, we used the denominators reported in the study. We did not assume values where missing standard deviations could not be obtained from trial authors or derived from confidence interval data.

Assessment of heterogeneity

We did not have enough studies to explore heterogeneity. If this changes in a future update, we plan to assess studies for heterogeneity based on a combination of visual inspection of the forest plots, the chi² test and the I² statistic; and also to assess the clinical significance of any statistical heterogeneity.

Assessment of reporting biases

We currently do not have enough data to explore publication bias with funnel plots. However, we were successful in our attempts to locate unpublished and ongoing studies.

Data synthesis

Only limited pooling was possible. As stated in the protocol, we used a fixed‐effect model to pool data as there was no statistical heterogeneity. We stipulated that meta‐analysis would only be performed if included studies are sufficiently homogenous to allow for a meaningful summary estimate. Additionally, we stated that where there is a moderate degree of statistical heterogeneity, we would check the results using a random‐effects model. We stipulated that we would calculate pooled risk ratios and 95% CIs for dichotomous data, and pooled mean differences (or standardised mean differences) and 95% CI for continuous data.

Subgroup analysis and investigation of heterogeneity

There were insufficient data available to carry out our planned subgroup analyses based on gender, activity levels (professional athlete or military recruits versus recreational athletes or others), type of orthoses (custom‐made versus off‐the‐shelf orthoses).

Sensitivity analysis

There were insufficient data to perform our planned sensitivity analyses. We had intended to look at the effects of including trials at high risk of bias, such as from lack of allocation concealment or blinding of assessors, or trials reported only in conference abstracts, by removing these studies from the analyses. We also planned to perform sensitivity analyses to explore the effects of missing data.

Results

Description of studies

See the Characteristics of included studies,Characteristics of excluded studies and Characteristics of ongoing studies .

Results of the search

We found seven relevant trials. Two trials were included (Collins 2008a; Wiener‐Ogilvie 2004); four were excluded (reasons given below) and one unpublished trial is listed as ongoing (KP Trial). A study protocol (Vicenzino 2008a) and a duplicate publication (Collins 2008b) were available for Collins 2008a. Results for Wiener‐Ogilvie 2004 were available also in a research dissertation (Wiener‐Ogilvie 2001).

The complete search yielded 391 references. Figure 1 shows the flow chart of search strategy results. We excluded 382 references from preliminary screening of the titles and abstracts. Among these, 150 were duplicated citations, 161 not relevant, seven did not include adults, 19 only studied knee arthritis, 12 studies did not involve foot orthoses, five were case series, and 28 were review articles. This left seven potentially relevant studies, as described above.

1.

Flow chart of search strategy results

Included studies

Two randomised controlled trials were included in this review (Collins 2008a; Wiener‐Ogilvie 2004). Figure 2 and Figure 3 show the participant flow for each trial.

2.

Participant flow Collins 2008a

3.

Participant flow Wiener‐Ogilvie 2004

Design of studies

Both included trials were single centre randomised clinical trials. Collins 2008a was conducted in Australia and Wiener‐Ogilvie 2004 in the UK. Neither trial had an untreated control group. Participants of Collins 2008a were allocated to one of four groups: foot orthoses; insole; physiotherapy; and foot orthoses plus physiotherapy. Participants of Wiener‐Ogilvie 2004 were allocated to one of three groups: exercise therapy; foot orthoses; and foot orthoses plus exercise therapy.

Participants

A total of 210 participants were included in the included trials. Follow‐up data were available for 198 participants: one person died and seven were lost to follow‐up in Collins 2008a; and four did not complete the trial in Wiener‐Ogilvie 2004. Knee symptoms were mostly chronic: the median duration of knee pain in Collins 2008a was 28 months (interquartile range: 12 to 84 months).

Demographic characteristics

The age range in Collins 2008a was 18 to 40 years (mean 29.3 years). Wiener‐Ogilvie 2004 did not give an age range; however, the mean ages for the three groups were 38.7 years (orthoses group), 51.0 years (exercise group), and 61.8 years (combined group). There were 79 male and 100 female participants in Collins 2008a and nine male and 22 female participants in Wiener‐Ogilvie 2004.

Clinical characteristics

There were important differences within and between the recruited cohorts. Collins 2008a included adults recruited from the local community by advertisements in various media. Wiener‐Ogilvie 2004 recruited people of any age with non‐traumatic antero‐medial knee pain and excessive foot pronation referred by their General Practitioner to the local physiotherapy department. Both trials had detailed inclusion and exclusion criteria (seeCharacteristics of included studies).

Notably, Collins 2008a did not assess foot pronation. Neither study required prior radiological examination before inclusion.

Interventions

Study grouping according to the interventions under test were:

• Comparison 1: orthoses versus insole (Collins 2008a).

• Comparison 2: orthoses plus physiotherapy versus physiotherapy alone (Collins 2008a; Wiener‐Ogilvie 2004).

• Comparison 3: orthoses versus physiotherapy (Collins 2008a; Wiener‐Ogilvie 2004)

Description of insoles and orthoses

Collins 2008a used flat insoles without any wedge arch support or heel cupping. The authors intended that flat insoles should act as a control group, that would, at least partially, act as a placebo. Unmolded flat insoles will not correct lower limb malalignment and therefore any effects could be due to a placebo effect. However, it was made of the same material as the orthoses and Collins 2008a also treated insoles with a degree of heat moulding. Both Collins 2008a and Wiener‐Ogilvie 2004 used orthoses of the same material (ethylenevinyl acetate). These are prefabricated off‐the‐shelf orthoses, customisable with heat moulding, wedges or heel raises. Those receiving foot orthoses had their orthoses moulded and, if necessary, had sequential trial fitting.

Description of physiotherapy

The 'physiotherapy' intervention differed between the two trials. The regimen in Collins 2008a included patellar mobilisation and taping, progressive vasti (muscles over the anterior aspect of the thigh) retraining exercises, hamstring and anterior hip stretches, hip external rotator retraining and a home exercise programme. Participants were encouraged to "continue exercise and activities that did not provoke their pain". Wiener‐Ogilvie 2004 prescribed quadriceps and hamstring stretching and strengthening exercise as well as hip adductor contraction and dynamic side stepping. A key difference between the two trials is that Collins 2008a employed patellar mobilisation and taping.

Outcomes

Collins 2008a reported a number of primary outcome measures. Sample size calculation was based on a clinically meaningful change of 15 mm on a 100 mm VAS for pain. The published protocol of this study refers to a few secondary outcome measures, none of which were reported in the paper. The senior author confirmed that they had collected secondary outcome measures. It is not clear why these were not reported. Collins 2008a followed up and assessed participants at six, 12 and 52 weeks. Wiener‐Ogilvie 2004 followed up participants at four and eight weeks only. 
 
 The outcomes reported by the two included trials are as follows:

Primary outcomes

• Knee pain: knee pain was evaluated in both the studies using a number of outcome measures:

  • Visual Analogue Scale (VAS): Collins 2008a reported usual and worst knee pain over the preceding week on VAS at 6, 12 and 52 weeks (100 mm scale where 100 mm = worst pain). Wiener‐Ogilvie 2004 reported usual knee pain for the preceding week on a VAS at four and eight weeks (100 mm scale,100 mm = worst pain).

  • Global improvement scale: Collins 2008a reported global improvement on a five point Likert Scale (marked improvement to marked worsening). The authors categorised and reported the global improvement scale to those reporting moderate or marked improvement or not, and reported the results as relative risk reductions and numbers needed to treat. They also reported global improvement scale on a VAS (20 cm scale: ‐100 mm = much worse, 0 mm = same, 100 mm = completely better). Wiener‐Ogilvie 2004 also reported knee pain using a global outcome score on a five point Likert scale.

  • Knee pain scale (KPS): Wiener‐Ogilvie 2004 reported the transfer and ambulation intensity scales of this outcome measure. The KPS consists of four sub‐scales: transfer frequency scale, ambulation frequency scale, transfer intensity scale, ambulation intensity scale. This scale was developed for use in patients with knee osteoarthritis (Rejeski 1995). Each scale consists of three items; each question is answered on a six point Likert scale. The scale is scored from 1 to 6 (1 = no pain; 6 = maximum pain).

  • Short‐Form 36 (SF‐36) pain scale: this was reported by Wiener‐Ogilvie 2004. The SF‐36 pain scale contains two questions, one with five and the other with six scoring options. The sum of scores is converted to a range of 0 to 100 (100 = least pain).

• Knee function:

  • Functional index questionnaire (FIQ): this was reported by Collins 2008a. The FIQ is a patient reported eight item questionnaire. Each question is answered on a four point Likert scale. The scale is scored from 0 to 16 points (16 = no disability) (Chesworth 1989).

  • Kujala anterior knee pain scale: this was reported by Collins 2008a. The Kujala anterior knee pain scale measures disability due to anterior knee pain. It consists of 13 patient reported questions (total: 0 to 100 points, 100 points = no disability) (Kujala 1993).

Secondary outcomes

• Patient satisfaction: Wiener‐Ogilvie 2004 collected limited information related to patient satisfaction with the orthoses; these results were reported in a dissertation (Wiener‐Ogilvie 2001).

• Quality of life: Wiener‐Ogilvie 2004 reported data from the SF‐36 physical function scale. The SF36 physical function scale contains 10 questions with three scoring options for each question. Scores from raw data are converted to a scale of 0 to 100 (100 = best health).

• Concurrent analgesic use: data are available from both Collins 2008a and Wiener‐Ogilvie 2004.

• Adverse effects: only Collins 2008a reported adverse effects of the interventions.

While Collins 2008a did not report compliance with the various interventions, they reported the numbers of participants in each group who did not receive the intervention according to protocol because they attended fewer than six sessions. Collins 2008a also detailed participant reported use of co‐interventions for PFJ during the trial. Participants of Wiener‐Ogilvie 2004 kept a diary of their compliance with exercises, and also completed a questionnaire at the end of the trial regarding their compliance. However, some data were missing and the authors found that there was discrepancy between the diary entries and patients' recollection of their compliance.

Excluded studies

The Characteristics of excluded studies details the reasons for excluding four trials. In summary, one study included adolescent females only (Eng 1993); one study did not trial foot orthoses (Miller 1997); one trial did not provide sufficient information to determine if the small subgroup of participants had patellofemoral pain (Trotter 2008); and one unpublished trial did not report knee pain despite purporting to do so at trial registration (Wies 2009).

Risk of bias in included studies

Figure 4 and Figure 5 show the review authors' judgement of risk of bias in the included studies.

4.

Methodological quality summary: review authors' judgements about each methodological quality item for each included study.

5.

Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.

Allocation

Both of the included studies satisfied criteria for adequate randomisation as well as allocation concealment. In Collins 2008a the randomisation sequence was drawn up and kept off site. A research assistant contacted the randomisation centre at recruitment. Wiener‐Ogilvie 2004 used sealed dark brown envelopes containing the group numbers. The envelopes were shuffled and the patient chose an envelope (Wiener‐Ogilvie 2001).

Blinding

Patients, unsurprisingly, were not blinded in either trial. Collins 2008a blinded both the outcome assessor and the data analyst. Wiener‐Ogilvie 2004 did not blind outcome assessors. The possibility remains that the side‐effects reported in Collins 2008a were influenced by lack of participant blinding.

Incomplete outcome data

Collins 2008a was considered at low risk of bias. Although eight of the 179 randomised participants dropped out (one died, seven could not be contacted), the results of the remaining 171 participants were analysed on an intention‐to‐treat basis. This was irrespective of the fact that 22 participants received fewer sessions of their allocated intervention physiotherapy than intended, and two from the insole group crossed over to using foot orthoses at 12 weeks. Patient data were available for 27 out of 31 participants in Wiener‐Ogilvie 2004. Two male and two female participants dropped out of the trial early: one participant was hospitalised, another had an accidental knee injury, one discontinued the trial for personal reasons and the last participant requested a different mode of treatment unrelated to the study interventions after three weeks of treatment (Wiener‐Ogilvie 2001). The risk of bias from incomplete data was considered unclear for this trial.

Selective reporting

The study protocol of Collins 2008a was published prior to the trial (Vicenzino 2008a). The authors reported all primary outcomes described in the protocol. This was also the only trial to have reported any adverse effects of foot orthoses. However, the protocol refers to a number of secondary outcomes, none of which were reported in the published study. A protocol was not available for Wiener‐Ogilvie 2004.

Other potential sources of bias

Similarity of baseline characteristics

Baseline characteristics of the participants allocated to different treatment groups differed in the two trials. Patients allocated to physiotherapy had more females compared with the other intervention groups in Collins 2008a; the duration of knee pain also differed between the groups. However, Collins 2008a noted that “including baseline data as covariates did not significantly influence outcomes”. In Wiener‐Ogilvie 2004, there were more male and relatively young participants with higher mean height in the orthoses group. Conversely, more participants in the exercise group took anti‐inflammatory drugs or pain killers as well as having the lowest mean duration of symptoms; and the combined group had the highest mean age and weight and the longest mean duration of symptoms.

Equal treatment of different groups

Care programmes, apart from the trial interventions, were not identical in both trials. This resulted in a high risk of performance bias in the included trials. Patients who received foot orthoses had a greater degree of contact and care from health professionals during the fitting of their foot orthoses, whereas the flat insole group only received minimal care in Collins 2008a. Collins 2008a did not report either the contact time or the number of sessions required for fitting of insoles or orthoses. Whereas the participants in the foot orthoses group were prescribed the home exercise regimen for twice daily performance for both legs, home exercise for the flat insole group was not reinforced. The number of sessions allocated for contact with care givers also differed between the different intervention groups in Wiener‐Ogilvie 2004. The trial authors did not report any attempts to control for such differential levels of care in these two studies.

Summary of overall risk of bias

Risk of bias was judged as low in terms of sequence generation and allocation concealment. Due to the nature of the interventions it was not possible to blind the participants to the intervention. However, Collins 2008a achieved some objectivity by blinding the assessors. There was some imbalance in baseline characteristics in both Collins 2008a and Wiener‐Ogilvie 2004. Participant contact with health professionals and attention to detail for ensuring comparable care other than the trial interventions differed in the included studies (Collins 2008a; Wiener‐Ogilvie 2004). Performance bias was such that outcomes could have been influenced significantly. Incomplete reporting was identified for Collins 2008a, which did not report some secondary outcomes and for Wiener‐Ogilvie 2004, which did not describe adverse effects.

Effects of interventions

While long term follow‐up data are more appropriate for participants whose chronic knee pain lasted from between nine and 96 months in the included cohort, we present also short term data given these were available from both trials. For Collins 2008a, we present data for six weeks (at the end of treatment) and for 52 weeks. Wiener‐Ogilvie 2004 limited their follow‐up to eight weeks only. Where data are pooled from both the studies, these usually apply to six weeks follow‐up in Collins 2008a and eight weeks follow‐up in Wiener‐Ogilvie 2004.

Collins 2008a reported knee pain using a number of outcome measures. We present global improvement scale and worst knee pain using VAS for this trial in the following.

Only Wiener‐Ogilvie 2004 provided data on patient satisfaction. Fifteen out of the 18 participants prescribed orthoses found them comfortable or very comfortable. Only seven could fit the orthoses into "almost any shoes", while the rest found it difficult to fit orthoses or could only fit them into one or two pairs of shoes. A majority (14) got used to the orthoses within a few days but five had to buy new shoes to accommodate the prescribed orthoses. Most participants found most of the nine types of prescribed exercises easy or moderately easy to perform; although seven rated one prescribed exercise as difficult.

Comparison 1: orthoses versus insole

Only Collins 2008a performed this comparison.

Primary outcomes

Knee pain

Although early results for global improvement scale (categorical) favoured the orthoses group (Risk Ratio (RR) 1.48, 95% CI 1.11 to 1.99), the difference between the two groups was not statistically significant at 52 weeks (RR 1.15, 95% CI 0.92 to 1.44) (seeAnalysis 1.1). A similar finding applied to the results for worst pain VAS (see Analysis 1.2). However, the differences between the two groups were neither statistically nor clinically significant at either follow‐up time (6 weeks: MD ‐8.20, 95% CI ‐17.67 to 1.27; 52 weeks: MD 1.50, 95% CI ‐8.57 to 11.57).

1.1. Analysis.

Comparison 1 Orthoses versus insole, Outcome 1 Knee pain: numbers with 'global improvement'.

1.2. Analysis.

Comparison 1 Orthoses versus insole, Outcome 2 Worst knee pain during preceding week (VAS: 0 to 100 mm: worst pain).

Knee function

There were no statistically significant differences between the two groups in knee function, assessed using the Functional Index Questionnaire, at either six or 52 weeks follow‐up (6 weeks: MD 0.70, 95% CI ‐0.30 to 1.70; 52 weeks: MD ‐0.40, 95% CI ‐1.50 to 0.70; seeAnalysis 1.3). The results of the anterior knee pain scale were marginally better in the orthoses group at six weeks (MD 4.30, 95% CI 0.34 to 8.26) but not at 52 weeks (MD ‐1.40, 95% CI ‐5.50 to 2.70) (seeAnalysis 1.4).

1.3. Analysis.

Comparison 1 Orthoses versus insole, Outcome 3 Functional index questionnaire (0 to 16: no disability).

1.4. Analysis.

Comparison 1 Orthoses versus insole, Outcome 4 Knee function: anterior knee pain (0 to 100: no disability).

Secondary outcomes

Neither patient satisfaction nor quality of life were reported. There was no statistically significant difference in concurrent analgesic use between the two groups (RR 0.89, 95% CI 0.45 to 1.78; seeAnalysis 1.5). Patients treated with orthoses reported significantly more adverse effects (rubbing, blistering, discomfort, pain in toes) compared with the insole group (RR 1.87, 95% CI 1.21 to 2.91, seeAnalysis 1.6). All adverse effects were rated as mild.

1.5. Analysis.

Comparison 1 Orthoses versus insole, Outcome 5 Concurrent analgesic use.

1.6. Analysis.

Comparison 1 Orthoses versus insole, Outcome 6 Adverse effects.

Comparison 2: orthoses plus physiotherapy versus physiotherapy alone

Primary outcomes

Knee pain

Neither the pooled short‐term results for the global improvement scale (RR 1.00, 95% CI 0.86 to 1.17) nor the long‐term results for Collins 2008a alone favoured one group over the other (RR 1.01 95% CI 0.82 to1.23) (seeAnalysis 2.1). Collins 2008a found no significant differences between the two groups in worst pain VAS at either six weeks (MD ‐3.70, 95% CI ‐12.99 to 5.59) or at one year follow‐up (MD ‐3.40, 95% CI ‐13.52 to 6.72) (seeAnalysis 2.2). Wiener‐Ogilvie 2004 found no statistically significant difference between the two groups in the change scores over eight weeks for the SF‐36 pain scale (MD 9.60, 95% CI ‐8.99 to 28.19, seeAnalysis 2.3). Notably, the mean score for the combined intervention group did not change over the eight week period.

2.1. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 1 Knee pain: numbers with 'global improvement'.

2.2. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 2 Worst knee pain during preceding week (VAS: 0 to 100 mm: worst pain).

2.3. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 3 SF36 pain scale (change scores at 8 weeks: positive scores = pain reduction).

Knee function

There was no statistically significant differences between the two groups in knee function, assessed using the Functional Index Questionnaire, at either six or 52 weeks follow‐up (6 weeks: MD 0.40 95% CI ‐0.59 to 1.39; 52 weeks: MD ‐0.40, 95% CI ‐1.51 to 0.71; seeAnalysis 2.4). There were no significant differences between the two groups in the anterior knee pain scale results at six weeks (MD 0.20, 95% CI ‐3.72 to 4.12) or at 52 weeks (MD 3.60, 95% CI ‐0.52 to 7.72); the clinical significance of the small difference potentially favouring the combined intervention group is doubtful (seeAnalysis 2.5).

2.4. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 4 Functional index questionnaire (0 to 16: no disability).

2.5. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 5 Knee function: anterior knee pain (0 to 100: no disability).

Secondary outcomes

Wiener‐Ogilvie 2004 found no statistically significant difference between the two groups in the change scores over eight weeks for the SF‐36 physical function scale (MD ‐4.50, 95% CI ‐17.36 to 8.36, seeAnalysis 2.6). There was no statistically significant difference in concurrent analgesic use between the two groups (RR 0.56, 95% CI 0.28 to 1.13; seeAnalysis 2.7). Data were not available for total adverse effects for the physiotherapy only group in Collins 2008a.

2.6. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 6 SF36 physical function (change scores at 8 weeks: positive scores = decrease in function).

2.7. Analysis.

Comparison 2 Orthoses plus physiotherapy versus physiotherapy alone, Outcome 7 Concurrent use of analgesics.

Comparison 3: orthoses versus physiotherapy

Primary outcomes

Knee pain

There were no significant differences between the two groups in the global improvement scale results at both follow‐up times (6 weeks: RR 0.99, 95% CI 0.80 to 1.21; 52 weeks: RR 1.04, 95% CI 0.86 to 1.27; seeAnalysis 3.1). Similar findings applied for worst pain VAS results measured in Collins 2008a (6 weeks: MD 7.60, 95% CI ‐1.77 to 16.97; 52 weeks: MD 5.40, 95% CI ‐4.57 to 15.37; seeAnalysis 3.2). Wiener‐Ogilvie 2004 found no statistically significant difference between the two groups in the change scores over eight weeks for the SF‐36 pain scale (MD 13.30, 95% CI ‐9.19 to 35.79; seeAnalysis 3.3).

3.1. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 1 Knee pain: numbers with 'global improvement'.

3.2. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 2 Worst knee pain during preceding week (VAS: 0 to 100 mm: worst pain).

3.3. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 3 SF36 pain scale (change scores at 8 weeks: positive scores = pain reduction).

Knee function

The physiotherapy group of Collins 2008a had statistically significant better knee function, assessed using the Functional Index Questionnaire, compared with the orthoses group at both follow‐up points (6 weeks: MD ‐1.10, 95% CI ‐2.10 to ‐0.10; 52 weeks: MD ‐1.20, 95% CI ‐2.29 to ‐0.11; seeAnalysis 3.4). There were no significant differences between the two groups in the anterior knee pain scale results at six weeks (MD ‐3.70, 95% CI ‐7.64 to 0.24) or at 52 weeks (MD ‐2.40, 95% CI ‐6.48 to 1.68); the clinical significance of the small differences potentially favouring the physiotherapy group is doubtful (seeAnalysis 3.5).

3.4. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 4 Functional index questionnaire (0 to 16: no disability).

3.5. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 5 Knee function: anterior knee pain (0 to 100: no disability).

Secondary outcomes

Wiener‐Ogilvie 2004 found no significant difference between the two groups in the change scores over eight weeks for the SF‐36 physical function scale (MD 0.24, 95% CI ‐0.68 to 1.17, seeAnalysis 3.6). There was no significant difference between the two groups in concurrent analgesic use (RR 0.81, 95% CI 0.44 to 1.49; seeAnalysis 3.7). Data were not available for total adverse effects for the physiotherapy only group of Collins 2008a.

3.6. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 6 SF36 physical function (change scores at 8 weeks: positive scores = decrease in function).

3.7. Analysis.

Comparison 3 Orthoses versus physiotherapy, Outcome 7 Concurrent analgesic use.

Discussion

Summary of main results

Two randomised trials with a total of 210 participants were included. There are three comparisons. Only limited pooling for short‐term outcomes was possible.

Orthoses versus insoles

Based on the available evidence from one trial, foot orthoses compared with insoles (control intervention) improved knee pain at six weeks follow‐up coinciding with the end of the initial treatment period, but this benefit was not maintained at one year follow‐up, nor was it evident for any of the other outcome measures. There is evidence that patients using orthoses had significantly more adverse effects, of a minor nature, than those using insoles.

Orthoses plus physiotherapy versus physiotherapy alone

Providing orthoses in addition to physiotherapy did not lead to improved knee pain or function outcome compared with physiotherapy alone. Analgesic requirements were also similar between the two groups. There were insufficient data to conclude on patient satisfaction or adverse effects.

Orthoses versus physiotherapy

There were no significant differences between foot orthoses versus physiotherapy in knee pain outcomes or in measures of knee function except for the results for the functional index questionnaire which favoured the physiotherapy group. The clinical relevance of these latter findings is unclear, especially in the context of the reports of poor reliability and validity for the functional index questionnaire. Given these, Crossley 2004 recommended that this outcome measure should not be used for patellofemoral pain.

Overall completeness and applicability of evidence

Neither of the included trials included an untreated control group, although Collins 2008a used flat insoles as a 'placebo'. Thus it is unclear whether the effects seen represent a spontaneous improvement or benefits arising from paying care and attention to the patient's needs rather than specific aspects of the intervention based on beliefs about pathophysiology. It has been shown that there can be spontaneous improvement in patellofemoral pain. Up to 22% patients with anterior knee pain diagnosed in adolescence were completely pain free at a mean follow‐up of 16 years (Nimon 1998). Unequal treatment of different intervention groups by providing different levels of attention and care from care‐givers might have given rise to performance bias in the trials but also impedes the attribution of trial findings to the stated interventions. There were insufficient data to perform subgroup analyses and to determine if the effect of the intervention differed according to different patient characteristics such as age, gender, activity levels and duration of knee pain. Vicenzino 2008b, in a post‐hoc analysis of the orthoses group of Collins 2008a, identified age, height, pain severity and mid foot morphometry as variables possibly affecting treatment success. These observations may warrant investigation in future studies. Currently though there is no evidence to determine whether foot orthoses might be useful for a particular subgroup of people with PFJ pain. Notably, the association between rearfoot eversion and PFJ pain remains controversial and the usefulness of orthoses, based on its suggested mechanism of action of restoration of knee biomechanics, in a subgroup of patients with rearfoot eversion or patellar malalignment might merit further investigation.

Quality of the evidence

Wiener‐Ogilvie 2004 was a small trial with a short and inadequate follow‐up. Although they did not find any significant difference between the intervention groups the possibility of type II error (false negative) cannot be ruled out. Collins 2008a presented data from a longer follow‐up and also performed a priori sample size calculation. In both studies allocation was concealed and random. This, however, did not prevent imbalances in baseline characteristics in both trials; those in Wiener‐Ogilvie 2004 put this trial at a high risk of bias. Neither of the trials were exempt from the risk of performance bias and, we believe, this could also have influenced the findings of both trials.

Potential biases in the review process

The review was conducted in accordance with a previously published protocol. We searched extensively for both published and unpublished studies to avoid publication bias. We also contacted study authors and acknowledged experts for more information on studies and were able to include unpublished information in our review. We did not locate any studies in the non‐English literature. It is possible that such studies might have been missed despite our efforts. The stages of study search, selection, data collection and analyses were all performed by more than one person to minimise bias.

Agreements and disagreements with other studies or reviews

Although we have found no other review specifically investigating the efficacy of foot orthoses in managing PFJ pain, a previous review investigating the efficacy of various physical interventions including foot orthoses found limited evidence to support the use of physical interventions in general in the management of PFJ pain (Crossley 2001). Our findings are similar to those of Bizzini 2003 where authors found significant methodological limitations in the available RCTs on management of PFJ pain that cast doubts about the validity of the study results.

Authors' conclusions

Implications for practice.

Despite the high incidence of patellofemoral joint (PFJ) pain and widespread use of foot orthoses, very few randomised trials have investigated the efficacy of this intervention. Though limited, the available evidence does not reveal any clear advantage of foot orthoses over simple insoles or physiotherapy for PFJ pain. While foot orthoses may help relieve knee pain over the short term, the benefit is likely to be marginal and patients treated with orthoses are more likely to have minor adverse effects such as rubbing, blistering and discomfort.

Implications for research.

Further research, via randomised trials with adequate sample size, is needed to establish whether there is a role for foot orthoses in the treatment of PFJ pain. We suggest that future research should directed at comparisons of foot orthoses versus no treatment, supportive care alone and a placebo group in a randomised trial over a one year period. Such trials should include also participants with PFJ pain who have excessive foot pronation or patellar malalignment, and subgroup analysis should be performed to determine if the results differ for these participants. Prior specification of subgroups should help to assess the clinical relevance of patellar maltracking and rear foot motion, and any other pathophysiological aspects currently believed to be important for PFJ pain. Apart from using validated measures of knee pain and knee function, the trial should also report on physical and mental well being of the participants, compliance, adverse effects, and costs. Any such trials should follow established standards for good practice and careful consideration must be given to 'placebo' effects and performance bias.

Appendices

Appendix 1. Search strategies

Cochrane Central Register of Controlled Trials (Wiley InterScience interface)

#1        MeSH descriptor Patellofemoral Pain Syndrome, this term only 
 #2        MeSH descriptor Patella, this term only 
 #3        MeSH descriptor Knee Joint, this term only 
 #4        MeSH descriptor Knee, this term only 
 #5        (#2 OR #3 OR #4) 
 #6        MeSH descriptor Arthralgia, this term only 
 #7        MeSH descriptor Pain, this term only 
 #8        (#6 OR #7) 
 #9        (#5 AND #8) 
 #10      (anterior knee pain):ti,ab,kw 
 #11      ((patell* or femoropatel* or femoro‐patell* or retropatell*) NEAR/2 (pain or syndrome or dysfunction)):ti,ab,kw 
 #12      ((lateral compression or lateral facet or lateral pressure or odd facet) NEAR syndrome) :ti,ab,kw 
 #13      ((chondromalac* or chondropath*) NEAR/2 (knee* or patell* or femoropatell* or femoro‐patell* or retropatell*)):ti,ab,kw 
 #14      MeSH descriptor Chondromalacia Patellae, this term only 
 #15      (#1 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14) 
 #16      MeSH descriptor Orthotic Devices, this term only 
 #17      MeSH descriptor Braces, this term only 
 #18      (orthos?s or orthotic* or insert):ti,ab,kw 
 #19      (#16 OR #17 OR #18) 
 #20      MeSH descriptor Foot, this term only 
 #21      MeSH descriptor Shoes, this term only 
 #22      (foot or feet or shoe*):ti,ab,kw 
 #23      (#20 OR #21 OR #22) 
 #24      (#19 AND #23) 
 #25      (#15 AND #24)

MEDLINE (Ovid interface)

1. Patellofemoral Pain Syndrome/      

2. Patella/ or Knee Joint/ or Knee/       

3. Arthralgia/ or Pain/              

4. and/2‐3 

5. anterior knee pain.tw

6. ((patell$ or femoropatell$ or femoro‐patell$ or retropatell$) adj2 (pain or syndrome or dysfunction)).tw.

7. ((lateral compression or lateral facet or lateral pressure or odd facet) adj syndrome).tw.      

8. ((chondromalac$ or chondropath$) adj2 (knee$1 or patell$ or femoropatell$ or femoro‐patell$ or retropatell$)).tw.              

9. Chondromalacia Patellae/               

10. or/1,4‐9 

11. Orthotic Devices/ or Braces/             

12. (orthos#s or orthotic$ or insert).tw. 

13. or/11‐12

14. Foot/ or Shoes/   

15. (foot or feet or shoe$).tw.  

16. or/14‐15

17. and/3,16             

18. and/10,17

19. randomized controlled trial.pt.         

20. controlled clinical trial.pt.  

21. randomized.ab.  

22. placebo.ab.         

23. randomly.ab.       

24. trial.ab. 

25. groups.ab.           

26. or/19‐25            

27. exp Animals/ not Humans/           

28. 26 not 27              

29. and/18,28

EMBASE (Ovid interface)

1. Patellofemoral Pain Syndrome/ 

2. Patellofemoral Joint/ or Patella/ or Knee/          

3. Arthralgia/ or Pain/       

4. and/2,3

5. Knee Pain/      

6. anterior knee pain.tw.  

7. ((patell$ or femoropatell$ or femoro‐patell$ or retropatell$) adj2 (pain or syndrome or dysfunction)).tw.           

8. ((lateral compression or lateral facet or lateral pressure or odd facet) adj syndrome).tw.

9. ((chondromalac$ or chondropath$) adj2 (knee$1 or patell$ or femoropatell$ or femoro‐patell$ or retropatell$)).tw.           

10. Patella Chondromalacia/          

11. or/1,4‐10        

12. "Prostheses and Orthoses"/ or Foot orthosis/ or Orthosis/         

13. (orthos#s or orthotic$ or insert).tw.      

14. or/12‐13          

15. Foot/ or Shoe/ or Orthopedic Shoe/     

16. (foot or feet or shoe$).tw.        

17. or/15‐16         

18. and/14,17

19. and/11,18        

20. Clinical trial/     

21. Randomized controlled trial/     

22. Randomization/

23. Single blind procedure/

24. Double blind procedure/          

25. Crossover procedure/  

26. Placebo/          

27. Randomi?ed controlled trial$.tw.          

28. Rct.tw.

29. Random allocation.tw. 

30. Randomly allocated.tw.

31. Allocated randomly.tw.

32. (allocated adj2 random).tw.     

33. Single blind$.tw.          

34. Double blind$.tw.        

35. ((treble or triple) adj blind$).tw.           

36. Placebo$.tw.   

37. Prospective study/       

38. or/20‐37         

39. Case study/     

40. Case report.tw.           

41. Abstract report/ or letter/         

42. or/39‐41         

43. 38 not 42        

44. limit 43 to human         

45. and/19,44        

Lines 20‐43 = Scottish Intercollegiate Guidelines Network search filter for randomised controlled trials

CINAHL Plus (EBSCOhost)

S1        (MH "Patellofemoral Pain Syndrome") 
 S2        (MH "Patella") or (MH "Knee Joint") or (MH "Knee") 
 S3        (MH "Arthralgia") or (MH "Pain") 
 S4        S2 and S3        
 S5        TX anterior knee pain   
 S6        TX ( patell* or femoropatell* or retropatell* ) and TX ( pain or syndrome or dysfunction )         
 S7        TX ( lateral compression or lateral facet or lateral pressure or odd facet ) and TX syndrome      
 S8        TX ( chondromalac* or chondropath* ) and TX ( knee* or patell* or femoropatell* or femoro‐patell* or retropatell* )     
 S9        (MH "Chondromalacia Patella")            
 S10      S1 or S4 or S5 or S6 or S7 or S8 or S9 
 S11      (MH "Foot Orthoses") or (MH "Orthoses") or (MH "Orthoses Design") or (MH "Orthoses Fitting") 
 S12      TX orthos?s or orthotic$*or insert        
 S13      S11 or S12      
 S14      (MH "Foot") or (MH "Shoes") or (MH "Orthopedic Footwear")           
 S15      TX foot or feet or shoe*           
 S16      S14 or S15      
 S17      S13 and S16    
 S18      S10 and S17    
 S19      (MH "Clinical Trials+")             
 S20      (MH "Evaluation Research+")   
 S21      (MH "Comparative Studies")    
 S22      (MH "Crossover Design")         
 S23      PT Clinical trial             
 S24      S19 or S20 or S21 or S22 or S23        
 S25      TX ( clinical or controlled or comparative or placebo or prospective or randomised or randomized ) and TX ( trial or study )         
 S26      TX random* and TX ( allocat* or allot* or assign* or basis* or divid* or order* )        
 S27      TX ( singl* or doubl* or trebl* or tripl* ) and TX ( blind* or mask* )    
 S28      TX crossover* or cross‐over or "cross over"    
 S29      TX ( allocat* or allot* or assign* or divid* ) and TX ( condition* or experiment* or intervention* or treatment* or therap* or control* or group*            
 S30      S25 or S26 or S27 or S28 or S29        
 S31      S24 or S30      
 S32      S18 and S31    

Data and analyses

Comparison 1. Orthoses versus insole.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Knee pain: numbers with 'global improvement'	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 6 weeks (end of treatment)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 52 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Worst knee pain during preceding week (VAS: 0 to 100 mm: worst pain)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.1 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Functional index questionnaire (0 to 16: no disability)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.1 at 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Knee function: anterior knee pain (0 to 100: no disability)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.1 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Concurrent analgesic use	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
6 Adverse effects	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected

Comparison 2. Orthoses plus physiotherapy versus physiotherapy alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Knee pain: numbers with 'global improvement'	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 6 or 8 weeks (end of treatment)	2	101	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.86, 1.17]
1.2 52 weeks	1	85	Risk Ratio (M‐H, Fixed, 95% CI)	1.01 [0.82, 1.23]
2 Worst knee pain during preceding week (VAS: 0 to 100 mm: worst pain)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.1 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 SF36 pain scale (change scores at 8 weeks: positive scores = pain reduction)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Functional index questionnaire (0 to 16: no disability)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.1 at 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Knee function: anterior knee pain (0 to 100: no disability)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 SF36 physical function (change scores at 8 weeks: positive scores = decrease in function)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7 Concurrent use of analgesics	2	101	Risk Ratio (M‐H, Fixed, 95% CI)	0.56 [0.28, 1.13]

Comparison 3. Orthoses versus physiotherapy.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Knee pain: numbers with 'global improvement'	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 6 or 8 weeks (end of treatment)	2	105	Risk Ratio (M‐H, Fixed, 95% CI)	0.99 [0.80, 1.21]
1.2 52 weeks	1	87	Risk Ratio (M‐H, Fixed, 95% CI)	1.04 [0.86, 1.27]
2 Worst knee pain during preceding week (VAS: 0 to 100 mm: worst pain)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.1 at 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 at 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 SF36 pain scale (change scores at 8 weeks: positive scores = pain reduction)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4 Functional index questionnaire (0 to 16: no disability)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.1 at 6 weeks (end of treatment)	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.2 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Knee function: anterior knee pain (0 to 100: no disability)	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1 at 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 at 52 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 SF36 physical function (change scores at 8 weeks: positive scores = decrease in function)	1		Std. Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7 Concurrent analgesic use	2	101	Risk Ratio (M‐H, Fixed, 95% CI)	0.81 [0.44, 1.49]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Collins 2008a.

Methods	Method of randomisation: by generation of randomisation sequence by an independent off‐site body Assessor blinding: yes Intention‐to‐treat analysis: yes
Participants	Location: single centre trial within a community setting in Brisbane, Australia. Number of participants (N): 179 Inclusion criteria: age 18 to 40 years insidious onset of anterior knee or retropatellar pain of greater than six weeks’ duration pain provoked by at least two of: prolonged sitting or kneeling, squatting, running, hopping, or stair walking tenderness on palpation of the patella, or pain with step down or double leg squat worst pain over the previous week of at least 30 mm on a 100 mm VAS. Exclusion criteria: concomitant injury or pain from the hip, lumbar spine, or other knee structures previous knee surgery; patellofemoral instability; knee joint effusion any foot condition that precluded use of foot orthoses allergy to strapping tape; use of physiotherapy or foot orthoses within the previous year or use of anti‐inflammatory drugs. Age: range 18 to 40 years (mean 29.3 years) Gender (male:female): 79:100 Assigned: 46 (group 1), 44 (group 2), 45 (group 3), 44 (group 4) Analysed: 45 (group 1), 41 (group 2), 42 (group 3), 43 (group 4)
Interventions	Four intervention groups: Group 1: off the shelf foot orthoses, Group 2: flat insoles, Group 3: physiotherapy, Group 4: foot orthoses plus physiotherapy Timing of intervention: 6 appointments, each lasting 20 to 60 minutes, over 6 weeks with the physiotherapists. Self‐management for the rest of the study period thereafter (52 weeks). Description of interventions: Group 1: received prefabricated, commercially available orthoses fitted to their shoes and customised, if necessary, to optimise comfort through heat moulding and by adding wedge or heel raises. Group 2: flat insoles, manufactured from the same material with identical covering fabric, of uniform thickness, with no inbuilt arch or wedging, underwent heat moulding. Group 3: physiotherapy consisted of patellar mobilisation, patellar taping, a progressive programme of vasti muscle retraining exercises with electromyographic biofeedback, hamstring and anterior hip stretches, hip external rotator retraining, and a home exercise programme. Group 4: participants received both orthoses and physiotherapy as described above and had an extra appointment with the physiotherapist if necessary. The participants were advised to continue exercise and activities that did not provoke their pain. Both the foot orthoses group and the flat insole group were prescribed additional home exercise programme. Whereas this was prescribed to be performed bilaterally twice daily for the orthoses group, the exercise programme was not reinforced for the flat insole group.
Outcomes	Length of follow‐up: all the groups were followed up and assessed at 6, 12 and 52 weeks. Primary outcomes: Knee pain: global improvement on a 5 point Likert Scale and Visual Analogue Scale (VAS) (20 cm scale: ‐10 cm = much worse, 10 cm = completely better), usual and worst pain over the preceding week on VAS (10 cm scale: 0 to 10; 10 cm = worst pain), Knee function: the anterior knee pain scale (0 to 100 points, 100 point = no disability), the functional index questionnaire (0 to 16 points, 16 = no disability) Secondary outcomes: Adverse effects and co‐interventions used were recorded in diaries, reported to the research assistant, or detailed in an "exit questionnaire". The published protocol of the paper refers to a number of other secondary outcome measures, none of which were reported in the published paper.
Notes	Interventions given by 17 qualified and trained physiotherapists. Off the shelf orthoses used, but customised if patients complained of discomfort. Study performed and reported a priori sample size calculation.
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	The authors used “a random number generator in blocks of eight with no stratification”
Allocation concealment?	Low risk	Randomisation sequence drawn up and kept off site, contact with randomisation centre via a research assistant .
Blinding? All outcomes	Unclear risk	"the assessor responsible for outcome measurement and data analysis remained blind to group allocation". Participants were not blinded.
Incomplete outcome data addressed? All outcomes	Low risk	Initial randomisation = 179. Drop out = 8 (1 died, 7 lost to follow‐up). 20 participants received fewer sessions of physiotherapy than intended, two participants from insole group crossed over to foot orthoses group at 12 weeks. All participants were analysed on an intention‐to‐treat basis
Free of selective reporting?	High risk	All primary outcomes were reported. However, various secondary outcomes mentioned in the protocol were not reported in the published study.
Similarity of baseline characteristics	Low risk	The median duration of knee pain was considerably greater in patients treated with foot orthoses (42 months) compared with the foot orthoses plus physiotherapy group (24 months). Groups were otherwise comparable.The authors, however, state that including the baseline data as covariates did not significantly influence outcomes.
Equal treatment of different groups	High risk	The insole group had less contact with the physiotherapist compared with the orthoses group. Home exercise programme was not reinforced in the flat insole group compared with the orthoses group.

Wiener‐Ogilvie 2004.

Methods	Method of randomisation: sealed dark brown envelopes containing numbers (1, 2 or 3: 1 = exercise therapy, 2 = foot orthoses, 3 = combination therapy) were randomly shuffled and the patient chose to open the envelope (Wiener‐Ogilvie 2001). Assessor blinding: no Intention‐to‐treat analysis: no
Participants	Location: single centre trial in Wishaw, Lanarkshire, UK Number of participants (N): 31 Inclusion criteria: patients of all ages with antero‐medial knee pain referred by their GP for physiotherapy, associated with excessive foot pronation Exclusion criteria: systemic rheumatoid conditions knee pain associated with previous direct trauma to the knee back injuries including neurological damage, sciatica and disc problems residual knee pain in patients where hip pain was the main complaint previous knee surgery intra‐articular knee injections in the last six months pregnant patients. Age: mean ages (and standard deviations) of different groups were as follows: orthoses 38.7 (17.2) years, exercise 51.0 (22.5) years, combined group 61.8 (10.3) years Gender (male:female): 9:22 (2 males and 2 females dropped out early) Assigned: 11 (group 1), 10 (group 2), 10 (group 3) Analysed: 9 (group 1), 9 (group 2), 9 (group 3)
Interventions	Three intervention groups: Group 1: functional foot orthoses (orthoses group), Group 2: exercise therapy for the knee (exercise group), Group 3: a combination of exercise therapy and foot orthoses (combined group) Timing of intervention: Patients received supervised treatment over a 4 week period. Patients receiving the knee exercise regimen (groups 2 and 3) were seen twice a week for two weeks and then once a week for a further two weeks by a physiotherapist (6 sessions in total). Patients receiving foot orthoses (groups 1 and 3) were assessed by the podiatrist once a week for 3 weeks to check for the fitness of the orthoses. Patients were advised to continue to wear the orthoses and perform exercises on their own accord after 4 weeks. Description of interventions: Group 1: Functional foot orthoses. Foot orthoses group were provided with foot orthoses with a 40° rearfoot post. Rearfoot posts and forefoot posts were adjusted using 20° or 40° additional wedges, if necessary. Group 2: Exercise therapy for the knee (exercise group). The exercise therapy included the following exercises: Isometric quadriceps contractions in full knee and hip extension. Isometric quadriceps contractions with hip slightly flexed and knee in full extension. Isotonic quadriceps contractions without resistance. Isotonic quadriceps contractions with resistance. Isotonic hamstring contractions. Dynamic stepping exercise. Hamstrings stretching exercises. Isometric hip adductors contractions. Dynamic side stepping. Group 3: A combination of exercise therapy and foot orthoses (combined group)
Outcomes	Length of follow‐up: 8 weeks Primary outcomes: knee pain: global improvement scale (5 point Likert scale), knee pain: VAS (10 cm scale: 0 least pain, 10 worst pain) knee pain: transfer and ambulation intensity scales of the knee pain scale (each scale consists of a patient reported three items questionnaire, each question answered on a six point Likert scale. The scale is scored from 1 to 6 (1 = no pain, 6 = maximum pain) knee pain: SF‐36 pain scale. The scale contains 2 questions, one with five and the other with six scoring options. The sum of scoring is converted to a range of 0 to 100 (100 = least pain). physical function: SF‐36 physical function scale. The scale contains 10 questions with three scoring options for each question. The scoring from the raw data is converted to a scale of 0 to 100 (100 = best health). Secondary outcomes: The authors did not report any adverse effects but did report concurrent analgesic use.
Notes	One podiatrist and two physiotherapists assessed all patients and were given a two‐hour training session to familiarise themselves with the standardised assessment and treatment protocol.
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	"randomisation carried out...containing numbers 1, 2, 3". Each sealed envelope contained a number. Each number represented an allocated group. Envelopes were randomly mixed up.
Allocation concealment?	Low risk	"sealed dark brown envelope, patients chose the envelope randomly"
Blinding? All outcomes	High risk	Care givers and participants not blinded.
Incomplete outcome data addressed? All outcomes	Unclear risk	27/31 completed the trial. Two male and two female participants dropped out of the trial early: one participant was hospitalised, another had an accidental knee injury, one discontinued the trial for personal reasons and the last participant requested a different mode of treatment unrelated to the study interventions after three weeks of treatment (Wiener‐Ogilvie 2001).
Free of selective reporting?	Unclear risk	Reported all the listed outcomes. However, the authors did not report adverse effects of orthoses. The authors did collect limited information on participant satisfaction of using orthoses, which is available in a dissertation by the first author (Wiener‐Ogilvie 2001).
Similarity of baseline characteristics	High risk	There were a number of demographic differences between the treatment groups. Patients treated with orthoses were more likely to be male, younger, and taller. Those treated with exercise alone were more likely to be taking anti‐inflammatory drugs or pain killers and had symptoms for the least time. The combined group had the highest mean age and weight and the longest mean duration of symptoms. These differences may have biased the findings of the trial.
Equal treatment of different groups	High risk	Different groups received different levels of attention from the physiotherapist with the orthoses group receiving the least number of sessions.

VAS = visual analogue scale

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Eng 1993	The authors compared the efficacy of orthoses versus flat insoles. Both groups also received additional physiotherapy. Knee pain as assessed with a VAS over eight weeks. But all participants were adolescent females with non‐traumatic bilateral retropatellar knee pain beginning insidiously and with excessive foot pronation and as such the trial did not meet inclusion criteria.
Miller 1997	This study was a randomised controlled trial but only examined the efficacy of knee orthoses on patellofemoral pain.
Trotter 2008	This was a randomised controlled trial with a cross‐over design. The authors compared the efficacy of foot orthoses and prefabricated insoles on lower extremity musculoskeletal pain. The trial included five patients with knee pain but the authors were not able to give any further details of these five patients. It is not clear how many, if any, of these five participants with knee pain actually had patellofemoral pain.
Wies 2009	Although this randomised controlled trial was stated at trial registration to be looking at the effects of custom semi‐rigid full‐foot orthoses "on planta‐surface foot pain and anterior knee pain", the lead investigator confirmed by personal communication that the study is now completed but that knee pain was not in fact recorded.

VAS = visual analogue scale

Characteristics of ongoing studies [ordered by study ID]

KP Trial.

Trial name or title	A randomised controlled trial to evaluate the effectiveness of insoles to treat anterior knee (patella femoral) pain
Methods	Randomised, subject‐blinded, single‐centre trial
Participants	100 male or female participants meeting the following criteria 1. Aged between 18 and 65 years 2. With self‐reported anterior knee pain (pain at the front of the knee) with a minimum of 4 weeks duration 3. With self‐reported anterior knee pain who score at least 40 mm on a 100 mm visual analogue scale (VAS) for pain 4. With self‐reported anterior knee pain in the 2 days prior to recruitment 5. Who agree to wear the insoles provided for at least 4 hours each day and complete diary cards and postal questionnaire as required 6. Who give written informed consent
Interventions	The participants will be randomly allocated to the following two arms: 1. Orthaheel® regular. This is a 3/4 length insole with arch shape. It is inserted into shoes and worn daily for 12 months 2. Sham control insole. This is a full length 3 mm deep soft flat insole. It is inserted into shoes and worn daily for 12 months
Outcomes	1. Knee pain 2. Knee function
Starting date	28/03/2008
Contact information	Dr Christopher Nester Director Centre for Rehabilitation & Human Performance Research Brian Blatchford Building, University of Salford, Salford United Kingdom, M6 6PU
Notes	This study was funded by a commercial foot care firm. The lead investigator has confirmed that the study is complete but the findings have not yet been reported. There was significant loss to follow‐up in this series.

Differences between protocol and review

We clarified that foot orthoses did not include flat insoles.

We extended the scope of the review to include comparisons of foot orthoses plus any another physical therapy intervention versus the same physical therapy intervention on its own; and comparisons of foot orthoses versus any other physical therapy interventions.

Contributions of authors

MH conceived the idea for the review and developed and finalised it with advice from PJ and AB. MH wrote the preliminary draft which was revised by PA, AB and PJ. MH and PA developed the search strategy with the help of Joanne Elliott (Trials Search Co‐ordinator, Cochrane Bone, Joint and Muscle Trauma Group). MH and PA independently extracted data and judged risk of bias of the included studies. This was reviewed and adjudicated by PJ. MH analysed and interpreted data which was overseen by PJ and AB. MH wrote the final draft of the review which was revised by PJ and AB. PJ provided clinical expertise in interpretation and analysis. AB provided general advice for the review at all stages.

Sources of support

Internal sources

• Betsy Cadwaladr University Health Board, Bangor, UK.

• University of Oxford, UK.

• University Hospitals of Birmingham, UK.

External sources

• No sources of support supplied

Declarations of interest

None known

New