Background The subacromial impingement syndrome (SIS) includes the rotator cuff syndrome, tendonitis and bursitis of the shoulder. Treatment includes surgical and non-surgical modalities. Non-surgical treatment is used to reduce pain, to decrease the subacromial inflammation, to heal the compromised rotator cuff and to restore satisfactory function of the shoulder. To select the most appropriate non-surgical intervention and to identify gaps in scientific knowledge, we explored the effectiveness of the interventions used, concentrating on the effectiveness of physiotherapy and manual therapy.
Methods The Cochrane Library, PubMed, EMBASE, PEDro and CINAHL were searched for relevant systematic reviews and randomised clinical trials (RCTs). Two reviewers independently extracted data and assessed the methodological quality. A best-evidence synthesis was used to summarise the results.
Results Two reviews and 10 RCTs were included. One RCT studied manual therapy as an add-on therapy to self-training. All other studies studied the effect of physiotherapy: effectiveness of exercise therapy, mobilisation as an add-on therapy to exercises, ultrasound, laser and pulsed electromagnetic field. Moderate evidence was found for the effectiveness of hyperthermia compared to exercise therapy or ultrasound in the short term. Hyperthermia and exercise therapy were more effective in comparison to controls or placebo in the short term (moderate evidence). For the effectiveness of hyperthermia, no midterm or long-term results were studied. In the midterm, exercise therapy gave the best results (moderate evidence) compared to placebo or controls. For other interventions, conflicting, limited or no evidence was found.
Conclusions Some physiotherapeutic treatments seem to be promising (moderate evidence) to treat SIS, but more research is needed before firm conclusions can be drawn.
- Evidence based reviews
- Exercise rehabilitation
- Soft tissue injuries
- Shoulder injuries
Statistics from Altmetric.com
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.
The subacromial impingement syndrome was included as one of the 23 disorders classified as specific disorders in the complaints of the arm, neck and/or shoulder (CANS) model. The CANS model was developed by 47 experts in the field of upper-extremity disorders. These experts were delegates from 11 medical and paramedical professional associations. Multidisciplinary agreement was achieved on the term, definition and classification of CANS. The model was developed to help professionals classify patients unambiguously in order to improve multidisciplinary cooperation and make data of scientific studies better comparable. CANS entities are common. The disorders are painful, disabling and a burden on healthcare resources.1 Women are more affected than men.2 ,3 In 19% of the patients, the discomforts were chronic.2 Of those with chronic CANS, 56% reported discomforts of the shoulder. In general practice, 85% of patients with shoulder pain were diagnosed with rotator cuff tendinopathy; 74% showed signs of impingement.4 One of the specific disorders mentioned in the CANS model is, as aforementioned, the subacromial impingement syndrome. Within this model, the term ‘subacromial impingement syndrome’ includes the rotator cuff syndrome—tendonitis of the m. infraspinatus, m. supraspinatus and m. subscapularis—and bursitis of the shoulder area. For this study, we followed this consensus. Patients with SIS suffer from pain, weakness and loss of movement of the affected shoulder.5 The occurrence of SIS is associated with highly repetitive work, forceful exertion in work, awkward postures and high psychosocial job demand.6
Surgical and non-surgical strategies are used to treat SIS. An article on the effectiveness of postsurgical interventions for SIS has already been published.7 The goal of non-surgical treatment is to decrease the subacromial inflammation, reduce the pain, allow healing of the compromised rotator cuff and restore satisfactory function of the shoulder.8 To help physicians select the most appropriate non-surgical intervention and to identify gaps in scientific knowledge, we explored the effectiveness of these interventions. Articles on the evidence for effectiveness of extracorporeal shock wave therapy and medication (oral, injected and patched) have been published elsewhere.9 ,10 This article concentrates on the effectiveness of physiotherapy and manual therapy as treatment for SIS.
The Cochrane Library, PubMed, EMBASE, PEDro and CINAHL were searched up to March 2009. Keywords related to SIS and interventions were included. Online supplementary appendix 1 shows the complete search strategy.
Systematic reviews and randomised clinical trials (RCTs) were included if they fulfilled all of the following criteria: (A) SIS, not caused by an acute trauma or any systemic disease as described in the definition of CANS, was studied (B) an intervention for treating SIS was evaluated, (C) results on pain, function or recovery were reported and (D) a follow-up period of ≥2 weeks was reported. There were no language restrictions.
After the full-text articles were included, we divided the included studies into different treatment groups for which separate reviews could be conducted. One of these groups concerned physiotherapeutic interventions. In this review, only studies were included in which physiotherapeutic interventions were compared to placebo, no treatment or another non-surgical treatment.
Two reviewers (BMAH and LG) independently applied the inclusion criteria to select potentially relevant studies from the title, abstracts and full-text articles, respectively. A consensus method was used to solve any disagreements concerning inclusion of studies, and a third reviewer (BWK) was consulted if disagreement persisted.
Categorisation of the relevant literature
Relevant articles are categorised under three headers: Systematic reviews describes all (Cochrane) reviews; Recent RCTs contains all RCTs published after the search date of the systematic review on the same intervention; and Additional RCTs describes all RCTs concerning an intervention that has not yet been described in a systematic review.
Two pairs of authors (RvdS/LG and BMAH/WDR) independently extracted data from the included articles. Information was collected on the study population, interventions and outcome measures. A consensus procedure was used to solve any disagreement between the authors. Results were reported in the short term (≤3 months), midterm (4–6 months) and long term (>6 months).
Methodological quality assessment
Two reviewers (LG, M S Randsdorp) independently assessed the methodological quality of each RCT using Furlan's 12 criteria (table 1).11 Each item was scored as ‘yes’, ‘no’ or ‘unclear’. ‘High quality’ was defined as a ‘yes’ score of ≥50%. A consensus procedure was used to solve disagreement between the reviewers.
In a (Cochrane) review, the use of a methodological quality assessment is a standard procedure. We describe the methodological quality scale/criteria that were used in the review and used the authors’ definitions of high and low quality for the included studies.
A quantitative analysis of the studies was not possible due to the heterogeneity of the outcome measures. Therefore, we summarised the results using best-evidence synthesis.12–14 The article was included in the best-evidence synthesis only if a comparison was made between the groups and the level of significance was reported. The results of the study were labelled ‘significant’ if one of the three outcome measures on pain, function or recovery reported significant results.
The levels of evidence for effectiveness are given in box 1.
Levels of evidence
Strong evidence: consistent (ie, when ≥75% of the trials report the same findings) positive (significant) findings within multiple higher quality randomised clinical trials (RCTs).
Moderate evidence: consistent positive (significant) findings within multiple lower quality RCTs and/or one high-quality RCT.
Limited evidence for effectiveness: positive (significant) findings within one low-quality RCT.
Conflicting evidence: provided by conflicting (significant) findings in the RCTs (<75% of the studies reported consistent findings)
No evidence: RCT(s) available, but no (significant) differences between the intervention and control groups were reported.
Characteristics of the included studies
The initial search resulted in five reviews from the Cochrane library. Through PubMed 5 reviews and 215 RCTs, through EMBASE 21 reviews and 193 RCTs, through CINAHL 141 reviews/RCTs and through PEDro 0 reviews and 13 RCTs were identified. Finally, 2 reviews and 10 RCTs were included.
The first systematic review of Green et al15 included nine RCTs (n=525) on acupuncture for shoulder pain versus placebo or other interventions. One of these RCTs compared ultrasound to acupuncture. The second systematic review of Green et al16 included 26 RCTs exploring physiotherapy versus placebo or other interventions for various shoulder discomforts; 10 RCTs (n=575) reported on physiotherapy to treat SIS.
The characteristics of the included studies are listed in online supplementary appendix 2A–C. A flow chart of the literature search is found in online supplementary appendix 3.
Methodological quality assessment
The results of the methodological quality assessment are presented in table 2.
Five (of the 10 included) RCTs were of high quality. The most prevalent methodological flaws were: (1) care provider not blinded and (2) unclear whether allocation was concealed.
The RCT included from the review of Green et al15 on acupuncture was of high quality. In this review, a methodological quality list was used consisting of 10 items as presented in the Cochrane handbook. The other systematic review of Green et al16 on physiotherapeutic interventions used a methodological scoring list of 11 items. All 11 RCTs in this review scored ≥50% on the quality assessment and were classified as high quality.
Effectiveness of physiotherapeutic interventions for SIS
The evidence for effectiveness of the various physiotherapeutic interventions for SIS is reported in table 3.
Effectiveness of exercise
Exercise versus placebo
One high-quality RCT34 (n=80) compared exercise (ie, supervised low-resistance exercises) to placebo laser as treatments for SIS. After 2.5 years, all pain scores found no significant differences between the two groups. At 6 months, ‘good or excellent function’ was found to be significantly better in favour of exercise (relative risk (RR) 2.45 (95% CI 1.24 to 4.86)).
There is moderate evidence in favour of the effectiveness of exercise versus placebo laser in the midterm and no evidence in the long term.
Exercise versus control
One high-quality trial37 investigated the effectiveness of individualised physiotherapy plus a home exercise programme (tissue and joint mobilisation, exercise, postural advice, strapping and very occasionally electrotherapy) versus controls (waiting for shoulder surgery; n=84) to treat SIS. At 6 months of follow-up, the mean improvement on the Constant score was better for the treatment group, but no statistical comparison was made between the groups. Significantly more patients in the control group required surgery versus the physiotherapy group (p=0.0008).
Another high-quality study21 compared physiotherapy (progressive muscle training programme, twice a week for 8 weeks) to controls in patients with SIS (n=60), with both groups being on a surgery waiting list. At 2 months of follow-up, the study group showed significantly better improvement versus the controls for ‘pain at rest’, ‘pain during movement’, DASH (Disability of Arm, Shoulder and Hand) 2 score and DASH 3 score (p=0.001, 0.001, 0.007 and 0.013, respectively). The physiotherapy group used significantly fewer analgesics and non-steroidal anti-inflammatory drugs versus the controls (p<0.041 and p<0.01, respectively). The range of movement measurements showed significantly better improvement in the physiotherapy group versus the controls for abduction and extension (p=0.001 and p=0.032, respectively), but there were no significant differences on flexion, medial and lateral rotations.
There is moderate evidence for the effectiveness of exercises versus controls for SIS in the short term and midterm.
Exercise versus treatment with a hyperthermia machine
A high-quality study17 (n=25) compared exercise (pendular swinging and stretching, twice a day (5 min) to treatment with a hyperthermia machine (heat application, 434 MHz, one session (30 min) 3 times/week for 4 weeks). At 6 weeks, significantly better results were found in favour of the hyperthermia group on the pain scores (p=0.04) and the Constant score (p=0.03).
There is moderate evidence that hyperthermia is more effective than exercise in the short term.
Exercise versus shoulder brace
One low-quality recent RCT23 (n=40) compared conventional physiotherapy (stretching, centring (exercises to centre the humeral head), strengthening, 10 sessions) to a functional shoulder brace for SIS. There were no significant differences between the groups on the pain scores and the Constant score at 12 weeks of follow-up.
There is no evidence for the effectiveness of exercise versus shoulder brace in the short term.
Exercise versus ultrasound
A high-quality study17 (n=23) compared exercise (ie, pendular swinging plus stretching) to ultrasound (1 MHz, 2 w/cm², three times a week) for supraspinatus tendinopathy. No significant differences between the groups were found on pain and the Constant score at 6 weeks of follow-up.
We found no evidence for the effectiveness of exercise versus ultrasound in the short term.
Exercise versus physiotherapy
Two low-quality recent RCTs compared self-training (ie, exercises using an elastic band) with formally supervised physiotherapy for SIS.
In the first study24 (n=40), physiotherapy consisted of strengthening exercises for the rotator cuff. At 12 weeks of follow-up, significant differences were found on the Constant score within the groups. No significant differences were found between the groups.
In the second study23 (n=40), physiotherapy consisted of stretching, centring and strengthening exercises. At 12 weeks of follow-up, no significant results were found between the two groups for ‘rest pain’, ‘pain at night’, ‘pain at load’ or the Constant score.
There is no evidence for the effectiveness of self-training versus physiotherapy in the short term.
Workplace-based hardening versus clinical-based hardening
A low-quality study26 (n=103) compared workplace-based work hardening (WWH, ie, biomechanics and ergonomic education plus shoulder stretch, scapular control exercise and shoulder strengthening exercise) with clinic-based work hardening (CWH, ie, conventional training) for rotator cuff tendinitis. At 4 weeks of follow-up, significant changes in favour of WWH between the groups were found on shoulder flexion (p=0.015) and on the Shoulder Pain and Disability Index score (p=0.034).
There is limited evidence that WWH is more effective than CWH in the short term.
Effectiveness of mobilisation
Mobilisation as add-on therapy to exercise
One high-quality RCT33 (n=49) compared mobilisation as an add-on therapy to exercise (flexibility and strength training) for SIS. After 3 weeks, significant differences were found on pain in favour of the exercise plus mobilisation group (weighted mean difference (WMD) −186.23 (95% CI −319.33 to −53.13)) and on the composite strength score (WMD 173.67 (95% CI 64.79 to 282.55)). A significant difference was found on function in favour of the exercise group (WMD 4.96 (95% CI 1.30 to 8.62)).
Another high-quality RCT31 (n=14) compared mobilisation as an add-on therapy to exercise (ie, hot packs, active ROM, stretching, strengthening, soft tissue mobilisation and education). At 3 weeks of follow-up, the group that received mobilisation as an add-on therapy showed significantly better results on pain (WMD −32.07 (95% CI −58.04 to −6.10)). No significant differences were found on the range of abduction, elevation, internal or external rotation.
There is conflicting evidence for the effectiveness of mobilisation as an add-on therapy to exercise versus exercise alone in the short term.
Effectiveness of manual therapy
Manual therapy as an add-on therapy to self-training
One recent high-quality RCT22 (n=30) compared self-training (ie, active ROM, stretching, strengthening exercise programme) with and without manual therapy (ie, joint and soft tissue mobilisation, techniques, ice applications) for SIS. At 12 weeks of follow-up, significant results on pain (no p value given) in favour of the manual group were found.
There is limited evidence for the effectiveness of manual therapy plus self-training versus self-training alone in the short term.
Effectiveness of ultrasound
Ultrasound versus placebo
Two RCTs comparing ultrasound to placebo were included in the review of Green et al.16 One high-quality RCT30 (n=24) found no significant differences between both groups on pain at 4 weeks of follow-up, range of abduction at 2 weeks or recovery (or substantial improvement) in the short term.
Another high-quality study27 (n=72) found no significant differences between the groups on pain, range of abduction (unknown follow-up) or function at 4 or 12 months of follow-up.
We found no evidence for the effectiveness of ultrasound versus placebo in the short term, midterm and long term.
Ultrasound versus hyperthermia
A high-quality RCT17 (n=26) compared ultrasound (US (1 MHz 2 W/cm2, 1 session (15 min) 3 times/week for 4 weeks)) to treatment with a hyperthermia machine (heat application, 434 MHz, 1 session (30 min) 3 times/week for 4 weeks).
Significant differences between the groups were found in favour of the hyperthermia group on the pain scores (p=0.045) and the Constant score (p=0.04).
There is moderate evidence that ultrasound is less effective than hyperthermia in the short term.
Ultrasound plus ionthophoresis with acetic acid versus placebo
One high-quality RCT35 (n=22) compared ultrasound plus ionthophoresis to placebo for calcific tendinitis. No significant differences between the groups were found on improvement in abduction after treatment (follow-up time not given) or on improvement in change in calcium deposit after treatment.
There is no evidence for the effectiveness of ultrasound plus ionthophoresis versus placebo.
Ultrasound versus corticosteroid injection
One high-quality study30 (n=24) compared ultrasound to a steroid injection (40 mg methylprednisolone plus 2 mL 2% lignocaine using the anterior approach into the shoulder joint) for rotator cuff problems. No significant differences were found between both groups on pain or range of abduction at 2 weeks of follow-up, and for treatment success (success or failure of the treatment at the end of 4 weeks, defined as a need for a steroid injection).
There is no evidence for the effectiveness of ultrasound versus steroid injections in the short term (4 weeks).
Ultrasound versus acupuncture
One high-quality RCT30 (n=24) compared ultrasound (10 min, 8 sessions) to acupuncture (once a week with moxibustion) for rotator cuff lesions. There were no significant differences between the two treatments for postintervention pain, range of abduction and success rate in the short term.
A high-quality study18 (n=85) compared ultrasound (twice a week for 5 weeks, 10 treatment sessions of 10 min, mode: 1 MHz, 1 W/cm2) to acupuncture (needle placement at 4 local points, 10 sessions, in addition to home exercises for both groups) for SIS. There were no significant differences between the groups measured by a combined score, which included the University of California—Los Angeles Shoulder Rating Scale, the Adolfsson-Lysholm Shoulder Score and the Constant score at 3, 6 and 12 months of follow-up.
There is no evidence for the effectiveness of ultrasound versus acupuncture (both in addition to home exercises) in the short term, midterm and long term.
Effectiveness of laser therapy
Laser versus placebo
Two RCTs compared laser therapy to placebo.
A high-quality RCT28 (n=24) found significant improvement (defined as excellent or good results) in the placebo group versus the laser group (27 sessions, 9×3 min) at 3 weeks of follow-up in the treatment of supraspinatus tendinitis.
In another high-quality RCT29 (n=35), no significant differences in the treatment of rotator cuff tendinitis were found between the groups (laser, 830 nm: 10 min 2 times/week for 8 weeks versus placebo) on pain scores or ‘range of movement’ at 8 weeks of follow-up.
There is conflicting evidence for laser therapy versus placebo in the short term.
Laser versus ultrasound
One low-quality study25 (n=24) studied low-power (30 J/cm2) laser therapy versus ultrasound (1.5 W/cm2, pulsed 1:4) versus controls for supraspinatus tendonitis. At 3 weeks of follow-up, a significant reduction in favour of laser versus ultrasound or versus controls was found on pain (90% vs 58% vs 50%, respectively; p<0.01) and improvement of muscle force (no exact data given: p<0.01). For disability, significant differences were found in favour of laser versus controls, but not between the other treatments (no exact data were given).
There is limited evidence for the effectiveness of low-power laser therapy versus ultrasound or versus controls for supraspinatus tendinosis in the short term.
Effectiveness of pulsed electromagnetic field (PEMF)
PEMF versus placebo
A high-quality study32 (n=60) compared pulsed electromagnetic field (PEMF) with sham PEMF for SIS. Significant differences between the groups were reported on ‘no pain at the end of treatment’ (6 days) in favour of the PEMF group (RR 19.00 (95% CI 1.16 to 312.42)) and after 4–6 weeks post-treatment (RR 39.00 (95% CI 2.46 to 617.81)).
Another high-quality trial19 (n=46) compared PEMF with sham PEMF, both in combination with non-surgical treatment (ie, Codman's pendulum exercises, exercises, cold pack and meloxicam) in patients with SIS. On rest pain, the Constant scores and the Shoulder Disability Questionnaire, no significant differences between the groups were found at 3 weeks of follow-up.
There is conflicting evidence for the effectiveness of PEMF versus placebo in the short term.
SIS is a common and challenging health problem for patients and healthcare providers. The present review provided data on the evidence for the effectiveness of manual therapy and several physiotherapeutic interventions for SIS.
Moderate evidence was found in favour of patients who received exercise therapy when compared with patients on a surgical waiting list in the short term and midterm, and also when compared with patients who had placebo laser in the midterm. Unfortunately, because of the lack of detailed description of baseline characteristics and exercise protocols used (eg, intensity, duration, frequency and load), the current evidence is not fully validated and difficult to interpret in relation to use in clinical practice.
In the trial of Giombini et al,17 more benefit was found (moderate evidence) in favour of the effectiveness of hyperthermia treatment versus exercise to treat supraspinatus tendinopathy in the short term. Because these results were yielded by one small study (n=37) that included only athletes, the results must be interpreted with caution. Athletes may have a higher pain threshold. The blinding of patients may be an issue in this RCT, regarding the special group athletes are. It may be useful to study the effect of hyperthermia in a more general population suffering from SIS.
Ultrasound and laser therapy
For ultrasound and laser therapy, we found conflicting levels of evidence: conflicting evidence was found for the effectiveness of ultrasound therapy versus placebo in the short term, and no evidence in the midterm and long term. Furthermore, conflicting evidence was found for laser therapy versus placebo. In contrast to these findings, limited evidence was found in favour of laser versus ultrasound in the short term. No evidence was found for ultrasound versus acupuncture in addition to home exercises in the short term, midterm and long term. However, more research is needed before a firm conclusion can be drawn. Future trials should concentrate on results not only in the short term but also in the midterm and long term.
The present review has some limitations. First, there is a lack of an unambiguous definition for SIS. Therefore, we used a wide-ranging search strategy, although it is based on the description of the SIS definition in the CANS model. Although SIS can be defined as a symptomatic irritation of the rotator cuff and subacromial bursa in the limited subacromial space, the diagnostic criteria and aetiology of SIS remain debatable. We used the criteria of Furlan et al11 to assess the methodological quality of the recent and additional RCTs. The Furlan criteria are recommended by the Cochrane Back Review Group (CBRG) and are used widely in systematic reviews of musculoskeletal disorders.7 ,9 ,38–40 The methodological quality assessment of the RCTs included in the reviews and our methodological quality assessment of the recent and additional RCTs differ greatly, for example, we defined a study as ‘high quality’ when the study scored ≥50% on the quality assessment. There is empirical evidence from a methodological study conducted with data from the CBRG that a compliance threshold of less than 50% of the criteria is associated with bias.41 Therefore, we also used a cut-off point of 50%. However, in the Cochrane reviews, no clear definition is given as to whether the study is considered to be of high or low quality. Even though the advised threshold for methodological quality was chosen arbitrarily, a threshold of, for example, 40% would have minimally altered the conclusions of this review: only one study25 would be upgraded to high quality. When 60% is taken as a cut-off point, the conclusions would weaken on exercise therapy20 ,21 ,34 and manual therapy plus self-training.22 A risk of bias can be introduced by categorising some of the studies as high-quality studies. Second, a study with a small sample size (ie, n=14)31 can be underpowered in numbers but categorised as high-quality studies.17 ,30 ,31 Another weakness of some included studies33 ,42 is the wide range of CIs reported, although a significant difference is found. Outcome measures may not be that strong and might even have no clinical significance. The dosage, frequency and methods of the therapies differ between studies when compared with each other. Therefore, the results must be viewed with caution. Despite this, we decided to include these smaller studies, in order to make this review more complete and to provide the next step for future, larger studies. We used a best-evidence synthesis as data synthesis. Although classification on the levels of evidence is arbitrary, conducting a best-evidence synthesis is often used throughout the literature.13 ,14 ,43
In conclusion, only one RCT on manual therapy was included in this review. Limited evidence for the effectiveness of manual therapy as an add-on therapy to self-training was found. All other studies included in this review concentrated on physiotherapy. Conflicting levels of evidence were found for the effectiveness of laser therapy and ultrasound. For mobilisation as an add-on therapy to exercises, conflicting evidence for effectiveness was found for the short term. Further, although we found the best results for hyperthermia compared to exercise therapy or ultrasound in the short term (moderate evidence), hyperthermia and exercise therapy were found to be more effective in comparison to controls or placebo in the short term (moderate evidence). For the effectiveness of hyperthermia, no midterm or long-term results were studied. In the midterm, exercise therapy gave the best results (moderate evidence) compared to placebo or controls, whereas for the long term no evidence was found. However, because of the lack of a detailed description of the exercise protocols used, the current evidence is difficult to interpret in relation to use in clinical practice. Therefore, future RCTs studying exercise therapy should also concentrate on the (long-term) effect of different exercise protocols, including the intensity, duration, frequency and load of the exercises.
What is known about the subject?
Different non-surgical interventions to treat the subacromial impingement syndrome (SIS), ranging from oral medication and physiotherapy to corticosteroid injections or extracorporeal shock wave therapy, are applied in clinical practice.
What this study adds to existing knowledge?
This study provides an evidence-based overview of the effectiveness of physiotherapeutic interventions and manual therapy to treat SIS.
Moderate evidence was found for the effectiveness of hyperthermia compared to exercise therapy or ultrasound in the short term. Hyperthermia and exercise therapy were more effective in comparison to controls or placebo in the short term (moderate evidence).
In the midterm, exercise therapy gave the best results (moderate evidence) compared to placebo or controls. For other interventions, conflicting, limited or no evidence was found.
In future RCTs, more detailed descriptions of the exercise protocols used are needed in order to interpret the results in relation to the use in clinical practice.
We thank M S Randsdorp for her participation in the quality assessment.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online appendix
Contributors LG was involved in the literature search, data collection, data interpretation, methodological quality assessment and writing of the article. EMH and BWK contributed to the data interpretation and critical revision of the article for important intellectual content. RvdS and WDR were involved in the data interpretation and writing of the article. BMAH was involved in the literature search, data collection, data interpretation and critical revision of the article for important intellectual content, as well as approval of the version to be published. All authors discussed the results and commented on the manuscript.
Funding This research was funded by Fonds Nuts Ohra.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.