Article Text

Download PDFPDF

How can we implement exercise therapy for patellofemoral pain if we don’t know what was prescribed? A systematic review
  1. Sinead Holden1,2,
  2. Michael Skovdal Rathleff1,3,
  3. Martin Bach Jensen1,
  4. Christian J Barton4
  1. 1 Research Unit for General Practice in Aalborg, Department of Clinical Medicine, Aalborg University, Aalborg, Denmark
  2. 2 Center for Neuroplasticity and Pain (CNAP), SMI,Department of Health Science and Technology, Aalborg University, Aalborg, Denmark
  3. 3 SMI, Department of Health Science and Technology, Aalborg University
  4. 4 La Trobe Sport and Exercise Medicine Research Centre, School of Allied Health, La Trobe University, Melbourne, Victoria, Australia
  1. Correspondence to Dr Sinead Holden, Research Unit for General Practice in Aalborg, Department of Clinical Medicine, Aalborg University, Fyrkildevej 7,1, 9220 Aalborg, Denmark; siho{at}


Objective To evaluate the completeness of exercise prescription in randomised controlled trials (RCTs) for patellofemoral pain (PFP), identify which elements are most frequently missing and supplement recommendations based on additional data from authors.

Design Systematic review.

Data sources All studies included in the most recent Cochrane review were evaluated. Additionally, the Cochrane search was updated in June 2016 in Cochrane, MEDLINE, EMBASE, PEDro, CINAHL and AMED databases. Two raters independently assessed completeness of reporting using the Toigo and Boutellier mechanobiological exercise descriptors, and Template for Intervention Description and Replication (TIDieR) checklist. Authors were also contacted to provide additional information.

Eligibility criteria for selecting studies RCTs of exercise interventions for PFP.

Results We included 38 RCTs. The level of exercise prescription detail was low, with no study providing complete information. The most commonly reported exercise descriptors were the ’duration of the experimental period' (n=38/38) and ’number of exercise interventions' (n=35). From TIDieR, the most commonly reported items were the ’intervention name' (n=38) and ’rationale' (n=36).

The least reported items from the exercise descriptors were ’volitional muscular failure', ’temporal distribution of contraction modes', ’time under tension' and ’recovery between exercise sessions' (all n=2/38). From TIDieR, the least reported item was ‘How well (fidelity and adherence)’ (n=3/38).

36 authors were contacted, with 22 replies and 13 providing additional exercise prescription details .

Conclusion Exercise prescriptions in RCTs with proven efficacy for PFP are poorly reported, impairing their implementation in clinical practice.

PROSPERO registration number CRD42016039138.

  • exercise rehabilitation
  • evidence based
  • implementation
  • knowledge translation

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Patellofemoral pain (PFP) is a common cause of knee pain throughout the lifespan, affecting 6%–7% of all adolescents and a similar proportion of adults.1–4 It has a negative impact on physical activity and health-related quality of life.5 PFP is highly persistent and difficult to treat, with more than 50% continuing to experience pain after 12 months, despite being offered evidence-based treatment.5–8

The most recent PFP consensus statement recommends the use of exercise therapy as the cornerstone of treatment, for reducing pain in the short, medium and long term, and improving function in the medium to long term.9 This recommendation is supported by a recent Cochrane and other systematic reviews10–12 evaluating the PFP exercise literature. While these evidence syntheses provide important information on the strength of the evidence, none evaluate the completeness of reporting of the specific components in exercise interventions for PFP. If details are not fully reported, it is difficult for clinicians to implement interventions in clinical practice. Previous research indicates that reviews often do not report exercise interventions in enough detail to allow for replication,13 14 limiting their clinical use.

Exercise therapy is a generic term, which can encompass a large variability in what is actually prescribed. A number of programming variables, such as exercise type, frequency, intensity, mode, time and rest intervals, can be manipulated. These parameters can influence physiological responses and effectiveness of treatment.15 16 Therefore, accurate reporting of these parameters is essential to translate evidence-based exercise therapy into clinical practice.

The EQUATOR network outlined the omission of crucial information in the description of research methods and interventions as a major cause for concern in current health research publications.17 The Template for Intervention Description and Replication (TIDieR) checklist18 was developed to address this barrier to research translation. However, this guide is not specifically designed for exercise trials and does not cover all important exercise prescriptions specifics. Toigo and Boutellier19 outlined 13 important mechanobiological exercise descriptors, which provide a framework to standardise the design and description of resistance exercise investigations. Considering published exercise therapy trials for PFP primarily evaluate resistance exercise,10 these descriptors provide an appropriate framework to evaluate exercise prescription.

The aim of this review was to systematically evaluate the completeness of descriptions of exercise interventions in randomised controlled trials (RCTs) of exercise therapy for PFP. Specifically, this was completed by assessing the reporting of the components of the TIDieR checklist18 and the exercise descriptors set out by Toigo and Boutellier.19 Due to the strong focus on risk of bias in peer review, a secondary aim was to examine whether the completeness of trials’ reporting correlates with traditional measures of bias. We also sought to determine if additional information could be acquired from the authors in order to provide more complete exercise guidelines. It was hoped that information obtained during this review could be used to provide clinicians with more precise information on ‘how to’ provide exercise therapy for patients with PFP.


The review methods were informed by Cochrane guidelines and the protocol was registered a priori on International Prospective Register of Systematic Reviews (ID: CRD42016039138; link: The review was reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

Data sources

In 2015, a Cochrane review was published, evaluating the effect of exercise therapy for reducing pain and improving function in RCTs for PFP.10 As this is the highest evidence available to support the use of exercise therapy for PFP, all articles originally included in this Cochrane review were included, and supplemented by an updated search, to identify recently published articles. van der Heijden et al 10 searched for and included studies from MEDLINE (1946 to 2014 week 20), EMBASE (1980 to 2014 week 20), CINAHL (1982 to 2014), Cochrane Library (May 2014) and AMED (1985 to May 2014).

To identify new articles, the search of van der Heijden et al 10 was used, and was limited to publications since June 2014 (when the Cochrane searches were last completed) to identify any new studies between June 2014 and 21 June 2016. The search was conducted in the following electronic bibliographic databases: MEDLINE (via OVID), EMBASE, CINAHL (via EBSCO), Cochrane Library and AMED (via OVID) and is detailed in the online supplementary material S1. No language restrictions were applied to searches.

Supplementary file 1

Inclusion/exclusion criteria

The same inclusion and exclusion criteria used in the original Cochrane review were applied10 to newly identified studies from the updated search.

Specifically, only RCTs evaluating an exercise intervention for treating PFP were included. Any exercise intervention applied on its own, or in combination with other non-surgical interventions were included, providing the other intervention was also applied to the comparator (or control) group. Studies evaluating surgery, other invasive treatment or pharmacological therapy were excluded. Participants were required to be diagnosed with PFP, which is defined as pain around or behind the patella, which is aggravated by at least one activity that loads the patellofemoral joint during weight bearing on a flexed knee (eg, squatting, stair ambulation, jogging/running, hopping/jumping).7


Potentially eligible articles were independently screened by title and abstract by two authors (SH and CJB). Full-text articles were then assessed for inclusion by both authors. In cases of disagreement, consensus was reached through discussion. An independent arbiter was available for consultation, but there were no disagreements that required this.

Primary outcome

The primary outcome was the completeness of descriptions of exercise interventions, which was evaluated using the 12-point TIDieR checklist18 and the exercise descriptor model proposed by Toigo and Boutellier.19

Data extraction

Phase 1

Data were independently extracted by two reviewers (SH and MSR) for each item of the TIDieR checklist,18 and Toigo and Boutellier exercise descriptors.19 If reference was made to additional details in a preregistered protocol, or published article, then the relevant information was extracted from these sources. The extracted information was used to compile a checklist of the complete and incomplete items for each study, with studies receiving one point for each item marked as complete. Items were considered complete if they were clearly and unambiguously described, to an extent which would allow them to be replicated. Items were considered incomplete if they were not described or if they were only partially described (e.g., described for some, but not all exercises). Scores were compiled for each study, for number of complete items on each of the checklists (Toigo and Boutellier and TIDieR). Data extraction was completed in a specifically developed Excel spreadsheet. Disagreements were addressed through a consensus discussion.

Phase 2

Attempts were made to contact authors for additional material relating to the exercise intervention for all trials. Initially, corresponding authors were emailed a spreadsheet containing the extracted information and a request to provide any additional information regarding the exercise prescription that was not originally included in the publication or a published study protocol. Authors were given 2 weeks to respond, then were sent a reminder, which was also sent to the coauthors, in case they could also provide any more details. If the corresponding author’s email address was not valid or if there was no listed email address (for older trials), we attempted to track down current contact details first for the corresponding author/first author, and if unsuccessful the senior and/or coauthors were contacted. The search for current details involved searching directories at the affiliated institutions listed on articles, Google searches and identifying recent publications. The completeness of the relevant items was subsequently reassessed after authors responded with additional information.

Risk of bias assessment

For the studies included in the original Cochrane review,10 the reported Cochrane domain-based risk of bias20 scores were used. Using the same criteria, two authors (MBJ and CJB) independently evaluated newly identified studies. The following domains were assessed: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias. Other sources of bias included bias from major imbalance in baseline characteristics and performance bias such as from lack of comparability in clinicians’ experience with the interventions under test, differences in care other than the interventions under test or compliance with the intervention.

Data analysis

Completeness of reporting from the original research reports and after contact to authors was compiled for each study in Excel.

To compare risk of bias with completeness of reporting, we created a risk of bias score, by scoring all items at low risk of bias as 1 and all items at unclear or high risk of bias as 0. They were summed to give each study a final score out of 10. This summary score was then plotted against the number of complete items for the TIDieR and Toigo and Boutellier descriptors, and Spearman’s correlation was used to test the association between risk of bias and completeness of reporting.

Due to the large range in publication dates, we explored whether the completeness of exercise description has increased by year of publication. This was exploratory and was not predefined according to the protocol. To do this, the total number of complete items on Toigo and Boutellier, and TIDieR, were plotted against the year of publication for each study.


Characteristics of studies

Thirty-one studies were included from the 2015 Cochrane review21–50 with seven additional articles from the updated search deemed eligible for inclusion3 51–56 (figure 1, PRISMA flow diagram). From these 38 studies, there were 36 individual trials. A list of the included trials, including a brief description of the intervention(s) and comparator(s), is provided in supplementary material S2. The studies were published between 1988 and 2016.

Supplementary file 2

Figure 1
Figure 1

PRISMA flow diagram for updating the search. PFP, patellofemoral pain; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; RCT, randomised controlled trial.

Completeness of exercise descriptions

Figures 2 and 3 show the number of interventions that were clearly described in publications for each of the items rated for the TIDieR checklist and the Toigo and Boutellier exercise descriptors, respectively . No study provided complete information for all of the items. However, all studies described the ‘duration of the experiment’ (from Toigo and Boutellier) and most provided the ’intervention name' and ’rationale' (from TIDieR). The least reported items from the Toigo and Boutellier exercise descriptors were ’volitional muscular failure', the ’fractional distribution of the contractions', ’time under tension' and ’recovery between exercise sessions', which were each reported by only two studies. From the TIDieR checklist, the least reported item was ‘How well (fidelity and adherence)’.

Completeness of exercise descriptions after contact to authors

Authors from 36 of the 38 included studies were contacted for additional information. Contact details could not be obtained for authors of two studies from 1992.32 33 Of those 36 contacted, 22 replied, with 13 supplying additional information. Once supplemented by authors, the completeness of descriptions were increased, although no study had complete data for all items. Details on load magnitude and range of motion were the most commonly supplemented items from Toigo and Boutellier (n=6 authors provided information both items; figure 2), while ‘What (materials and procedure)’, ‘tailoring’ and ‘modifications’ were the most commonly supplemented items from the TIDieR checklist (n=3 authors provided information for these items; figure 3).

Figure 2
Figure 2

Number of interventions (out of 38) with complete reporting for each of the TIDieR checklist items. TIDieR, Template for Intervention Description and Replication.

Figure 3
Figure 3

Number of interventions (out of 38) with complete reporting for each of the Toigo and Boutellier exercise descriptors.

Completeness according to year of publication

An exploratory analysis was undertaken to explore whether the completeness of descriptions has improved over time, and as figure 4 demonstrates, there has been no increase in the details reported.

Figure 4
Figure 4

Number of complete items from both Toigo and Boutellier exercise descriptors, and TIDieR checklist (maximum=26) according to year of publication. TIDieR, Template for Intervention Description and Replication.

Risk of bias

Full details of the risk of bias for the individual trials are provided in online supplementary material S3. Spearman’s correlation calculation demonstrated a statistically significant (p=0.019), weak positive correlation (Spearman’s r=0.38) between exercise description completeness and number of low bias items on the Cochrane risk of bias tool. This relationship is plotted in online supplementary material S4.

Supplementary file 3

Supplementary file 4


Reporting of exercise interventions in RCTs evaluating the efficacy of exercise therapy for PFP was poor. No study reported their exercise programme in enough detail to allow full replication. This limited reporting was highlighted by both the commonly accepted TIDieR checklist and our addition of the Toigo and Boutellier criteria, which relate specifically to resistance training interventions. This limitation in reporting creates a clear barrier to implementation of recent recommendations to provide exercise therapy of proven efficacy to patients with PFP.9

Most studies clearly described their intervention names and rationale. However, few reported important components of exercise prescription, such as rest intervals, time under tension and distribution of contraction modes. These parameters constitute the exercise stimulus. The physiological mechanisms underpinning skeletal muscle adaptation to exercise are highly dependent on, and distinct to this stimulus.57–59. Only 3 of 38 studies reported fidelity and adherence. This is important, as interventions are rarely delivered exactly as intended, and adherence may impact on the clinical benefit.3 Not reporting these items impedes accurate interpretation of study findings and could prevent future interventions from targeting strategies to improve fidelity and adherence, as the problem may be understated at present.

There was no association between completeness of reporting and year of publication, despite the strong focus on transparency in research reporting in recent years (e.g., by initiatives from the EQUATOR network).18 There was a weak but significant association between completeness of reporting and number of items scoring low risk of bias. This highlights that although there may be some relationship between the two, papers with high internal validity do not necessarily have strong external applicability. Caution must be taken in relying solely on risk of bias during appraisal of clinical trials. There needs also to be emphasis on the importance of clear and unambiguous reporting of interventions to facilitate greater translation of research, thereby increasing the impact of research articles.

Exercise interventions, in other populations, such as cardiac rehabilitation60 and peripheral arterial disease,61 also demonstrate poor reporting (when assessed according to the TIDieR criteria). This highlights that incomplete reporting of exercise interventions is not isolated to PFP, but also a problem in relation to exercise for other non-musculoskeletal conditions. Similarly, other reviews highlight the lack of reproducibility associated with physiotherapy interventions62 and other non-pharmacological interventions.14 63 Considering this, we believe these results are likely representative of a widespread problem in scientific trials using exercise therapy.

We chose to combine the TIDieR checklist with the Toigo and Boutellier exercise descriptors criteria to gain greater insight into important reporting details related to resistance exercise interventions. Specifically, items from the Toigo and Boutellier criteria can influence physiological adaptations64–68 and analgesic response69 to resistance exercise, which is the primary form of exercise therapy intervention in the PFP literature.10 Even if we look at the ‘classical’ or traditional resistance exercise descriptors from the Toigo and Boutellier criteria (including load, repetitions, sets, rest between sets, frequency and duration),19 70 these are not commonly reported. Of these traditional descriptors, load was the most commonly reported, yet it was only available for 55% of trials, even after authors provided additional details.

Moving forward in clinical practice

It was our goal to provide a ‘how to’ guide for clinicians to prescribe exercise for PFP, and we were hopeful that this may be possible after receiving supplemental details from authors. As this was not possible solely from the descriptions of the interventions, we used the strength and conditioning literature, and exercise prescription principles in order to provide some clinical recommendations (online supplementary material S5), which can be used to guide clinical practice until more detailed evidence is available. Our recommendations are available as a printable leaflet for clinicians and cover each aspect of the Toigo and Boutellier criteria based on key exercise prescription principles and PFP literature.3 10 11 15 16 46 49 69–75 The leaflet will be freely available from the iPFRN (International Patellofemoral Research Network:

Supplementary file 5

Moving forward in research

This review highlights a need to standardise the reporting of specific components of exercise interventions beyond current accepted guidelines. The Consensus on Exercise Reporting Template (CERT) was published with this goal in mind.76 However, CERT was not used in this review as it was published after completion of the protocol registration, and commencement of data extraction and analysis. CERT specifically advocates reporting components such as ‘Detailed description of each exercise to enable replication’, ‘Describe the setting in which the exercises are performed’, ‘Detailed description of the exercise intervention’ as well as how exercises are progressed and tailored. Despite the importance of these factors in describing exercises, the checklist does not explicitly require the mechanobiological exercise descriptors outlined by Toigo and Boutellier, such as the relative load, time under tension, rest intervals between sets or repetitions, and so on. Therefore, in the future, we also encourage authors and journal editors to consider using the Toigo and Boutellier exercise descriptors in conjunction with the CERT to guide reporting standards, as they focus on different aspects of exercise prescription and would be complementary. Another option is for researchers to use tools such as the Toigo and Boutellier exercise descriptors in combination with the CERT and TIDieR checklists during trial registrations (e.g., and protocol publications (e.g., as in Riel et al 77). Box outlines a full list of recommendations that researchers can consider to ensure that their interventions are accessible.


Recommendations for research

  • Authors, journal editors and reviewers are advised to comply with appropriate reporting guidelines for reporting exercise interventions in original trials (eg, Template for Intervention Description and Replication, Consensus on Exercise Reporting Template and Toigo and Boutellier exercise descriptors).

  • Journal article authors may wish to include full details in protocol registrations or supplementary materials, due to restrictions with word counts, allowing them refer to these in trial publications.

  • Other additional avenues to communicate exercise therapy interventions are available for researchers to use and should be considered, including:

    • Uploading videos, and manuals, including visual material such as pictures or infographics to journal websites, via video abstracts, dedicated trial websites or via social media platforms.

    • Where possible, smartphone apps that host interventions can be made freely available on iOS and Android app stores.

    • Journals such as BJSM offer the opportunity to publish infographics, which display information in a graphic format.78 If authors pursue this, they can ensure that sufficient information regarding the exercise prescription is provided in the infographic.

  • Inclusion of full descriptions of the explicit components of exercise prescriptions should be considered in future high-quality reviews (e.g., Cochrane reviews) to increase the clinical use of this high-quality evidence.


We were only able to get additional information from authors involved in 13 of the 38 included studies, limiting our ability to improve on the detail of reporting as originally intended. A large proportion of the authors were unresponsive to requests for additional information despite reminders and contact to the non-corresponding coauthors. Although we chose to combine the TIDieR with the Toigo and Boutellier criteria, it remains unknown which of these additional descriptors may be the most important for patients or what the ‘optimal’ dose is for decreasing pain or increasing self-reported function in patients with PFP. Due to incomplete reporting from previous studies evaluating resistance training in PFP, we were unable to provide any further insight into this.

The Toigo and Boutellier guidelines are specific to resistance training and may not be directly applicable to other types of exercise therapy. In fact, two of the ‘exercise therapy’ studies, which were included from the Cochrane review,39 43 used stretching interventions rather than active exercises, preventing any data for the Toigo and Boutellier exercise descriptors from being extracted for these trials. There are also other barriers to translating research into clinical practice, which were not considered in this review. Unfortunately, factors such as how long it took to locate and extract the relevant information, and whether or not reports were open access were not documented over the course of this review. We recommend future research should also document this process.


Exercise prescriptions in RCTs with proven efficacy for PFP are poorly reported. This is a barrier to implementing current exercise therapy programmes with reported efficacy in clinical practice. In addition to standardising reporting of exercise interventions in original trial reports, and systematic reviews, we advocate that researchers should use a combination of the Toigo and Boutellier and TIDieR guidelines to design their future exercise interventions. This will ensure all components that influence response to training, and perhaps clinical outcome, are considered from the outset.

What is already known?

  • Exercise therapy including resistance training can improve pain and function in patients with patellofemoral pain (PFP).

  • The most recent consensus statement recommends exercise therapy for PFP, which is supported by the Best Practice Guide for conservative management of PFP.

  • Many resistance exercise variables can be manipulated in the prescription of an exercise intervention, which this will impact on physiological responses.

What are the new findings?

  • The overall detail related to specific exercise prescriptions in randomised controlled trials of exercise therapy for patellofemoral pain (PFP) is low.

  • Current exercise therapy programmes with reported efficacy for PFP cannot be directly replicated.

  • We provide guidance to clinicians and researchers on how to prescribe exercise to people with PFP by considering previously published clinical trials alongside key exercise prescription principles and literature (see online supplementary material S5).


We would like to thank all authors who responded to our requests for additional information.



  • Handling editor Karim M Khan

  • Contributors SH, MSR, MBJ and CJB all participated in the study design, protocol and registration. SH and CJB were responsible for selecting articles for inclusion. MBJ and CJB conducted the risk of bias assessment. SH and MSR were responsible for data extraction. SH drafted the manuscript and all authors provided critical input and final approval. SH is the guarantor.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.