Background Despite the clinical recommendation of exercise and diet for people with knee osteoarthritis (OA), there are no systematic reviews synthesising the effectiveness of combining physical activity and dietary restriction interventions on the musculoskeletal function of overweight and obese older adults with knee OA.
Objective To evaluate the effectiveness of combined physical activity and dietary restriction programmes on body weight, body mass index (BMI) and the musculoskeletal function of overweight and obese older adults with knee OA.
Information sources A detailed search strategy was applied to key electronic databases (Ovid, Embase, Web of Science andCumulative Index to Nursing and Allied Health Literature (CINAHL)) for randomised controlled trials (RCTs) published in English prior to 15 January 2017.
Participants Participants with BMI ≥25 kg/m2, aged ≥55 years of age and with radiographic evidence of knee OA.
Interventions Physical activity plus dietary restriction programmes with usual care or exercise as the comparators.
Outcome measures Primary outcome measures were body weight, BMI or musculoskeletal function. Secondary outcome measures were pain and quality of life.
Results One pilot and two definitive trials with n=794 participants were included. Two articles reporting additional data and outcome measures for one of the RCTs were identified. All included RCTs had an unclear risk of bias. Meta-analysis was only possible to evaluate mobility (6 min walk test) at 6 months and the pooled random effect 15.05 (95% CI −11.77 to 41.87) across two trials with n=155 participants did not support the combined intervention programme. Narrative synthesis showed clear differences in favour of a reduced body weight and an increased 6 min walk in the intervention group compared with control groups.
Conclusion The quality of evidence of benefit of combining exercise and dietary interventions in older overweight/obese adults with knee OA is unclear.
Trail registration number CRD42015019088 and ISRCTN, ISRCTN12906938.
- randomised controlled trials
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Statistics from Altmetric.com
Strengths and limitations of this study
This is the first systematic review of combined physical activity and dietary restriction interventions in overweight and obese older adults with knee osteoarthritis.
The protocol of this review was registered in PROSPERO and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines and the Cochrane handbook; Grading of Recommendation, Assessment, Development and Evaluation was used to evaluate the quality of the included trials.
The review included a mixed method analytical approach.
Few eligible studies were identified; however, important information is highlighted which could inform clinical practice.
Current evidence shows that the burden of chronic musculoskeletal conditions especially osteoarthritis (OA) increases with advancing age.1 OA is the most common type of arthritis affecting older adults. It is a degenerative joint disease that may affect any joint within the body causing chronic pain, functional limitation and emotional disturbance and may lead to disability and negatively affect quality of life (QOL).2–5 Knee OA is a common condition in older adults affecting about 3.64% of the global population in 2010.6 7 In the UK, there is approximately 4.7 million older adults aged 45 years or over experiencing knee OA symptoms.1 8 In addition, more than 20 million people seek treatment for knee OA in the USA.9 10 Given the increasing numbers of older adults in the population, combined with the increasing prevalence of obesity and being overweight throughout the population, it is anticipated that the incidence of knee OA will increase rapidly over the next decade.8
Unfortunately, there is no specific treatment for knee OA. Most recommendations describe three treatment modalities: non-pharmacological, pharmacological and surgical.11 12 Most knee OA evidence-based guidelines recommend non-surgical treatment13 14 and most general practitioners prefer the non-pharmacological and non-surgical interventions as the first line of treatment (recognised as ‘usual care’).11 These interventions are focused on patient education, self-management, pain reduction, function and QOL improvement, body weight reduction and exercise (either land-based or water-based).1 14–17 It is well known that obesity is an important risk factor for knee OA progression and several studies recommend obesity control for decreasing disease burden, since a decrease in body weight will lead to a reduction of joint load and inflammation.3 14 17 18 Weight reduction could be considered as a functional treatment in knee OA rehabilitation since a 12%–15% reduction compared with initial body weight has been shown to improve function and reduce pain.19 Moreover, the appropriate percentage of body weight reduction has been investigated in a systematic review and meta-analysis of five randomised controlled trials (RCTs).20 The review concluded that professional treatment of knee OA should include a weight reduction plan and patients should be encouraged to lose at least 5% of body weight over a 20-week period to achieve symptomatic relief.20
In addition to weight reduction, clinical guidelines for knee OA management and level 1 evidence recommend exercise therapy as the main intervention.20–24 Moderate intensity aerobic exercise (eg, walking) is recommended to maintain musculoskeletal function and reduce pain.20–22 However, the optimal exercise prescription for older adults is still unclear and further research is required.7 The demand for optimal exercise is increased in patients with obesity who may face more challenges and believe in the greater importance of physical activity compared with dietary intervention.25 26
Clinically combining a weight loss programme with exercise therapy may help overweight and obese older adults with knee OA to achieve a 10% loss of total body weight as well as safely relieve knee OA symptoms.3 Also, a recent RCT which included older adults has shown that a non-surgical treatment programme had longer-lasting beneficial effects, evidenced by a delayed requirement for elective total knee replacement (TKR) surgery in a secondary healthcare setting.27 Moreover, for those who are eligible for unilateral TKR, non-surgical intervention may delay their surgical intervention for several months.28 There are no systematic literature reviews synthesising the evidence of the effectiveness of combining physical activity and dietary restriction interventions on the musculoskeletal function of overweight and obese older adults with knee OA.
The aim of this review was to evaluate the effectiveness of combined physical activity and dietary restriction programmes on the musculoskeletal function of overweight and obese older adults with body mass index (BMI) ≥25 kg/m2, aged ≥55 years of age and with radiographic evidence of knee OA.
To evaluate the effectiveness of combined physical activity and dietary restriction programmes on body weight, BMI and the musculoskeletal function of overweight and obese older adults with knee OA.
Protocol and registration
A systematic review was conducted according to a predefined protocol following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-P guidelines29 and the Cochrane handbook.30 The review was registered on PROSPERO on 01 April 2015 (CRD42015019088), and is reported in accordance with the PRISMA statement (see online supplementary data).29
Full article not written in English.
Randomised controlled trials.
Combined physical activity and dietary restriction programmes.
Usual care (including advice or physical activity alone or dietary restriction alone) or exercise (participants received an exercise programme similar to the intervention group).
Primary outcome measures: body weight, BMI, musculoskeletal function either self-reported function or objective functional performance measures, also, including mobility, joint range of motion (ROM) and muscle strength.
Secondary outcome measures: pain and QOL.
The search employed sensitive, topic-based strategies designed for each database (to 10 December 2015):
The Cochrane Library: Controlled Trials Register, NHS Economic Evaluation Database.
Cumulative Index to Nursing and Allied Health Literature (CINAHL), Embase, Medline, Web of Science.
Hand searches in key journals and lists of references.
Unpublished research and grey literature such as Open Grey.
Government, official, organisational such as UK Department of Health, WHO and National Health Service (UK).
Clinical trials registration, theses abstracts and Google scholar.
Search strategies of predefined search terms were developed and tested for applicability (ASA, and a specialist librarian from the University of Birmingham on 13 February 2015). The definitive search strategy was run by two independent researchers (ASA/AMK, 10 December 2015). Endnote X7 software was used for data management. Search results were imported and duplicates were removed. An example of the Medline Ovid search strategy is presented in table 1. The search was updated on 15 January 2017 to include studies published in 2016 by ASA/AMK and no eligible studies were identified.
The eligibility of included studies was independently assessed by two reviewers (ASA/AMK) according to the eligibility criteria. The reviewers screened the results of the search by titles and abstracts, and then full text. A study was considered to be eligible when both reviewers assessed the full text independently and found it to fulfil the eligibility criteria. A third reviewer (CAG) mediated in the case of disagreement. The inter-rater agreement was evaluated using Cohen’s Kappa measure.32
Using a standardised form (developed by ASA) based on the Cochrane Consumers and Communication Review Group’s data extraction template,33 two reviewers (ASA/AMK) extracted data independently. A third reviewer (CAG) checked for consistency and clarity.
Items reported on the data extraction form for each trial included demographic information, methodology, intervention details and all specified reported outcomes.
Risk of bias in individual trials
The internal validity of each included trial was assessed using the Cochrane risk of bias assessment tool34 recommended by PRISMA.29 All domains of the risk of bias tool were assessed independently by two reviewers (ASA/CAG). A third reviewer (ABR) mediated in the case of disagreement.
Risk of bias across trials
Risk of bias was considered high if the proportion of information from trials with high risk of bias was sufficient to affect the interpretation of the results. Risk of bias was considered unclear if most information was from studies with a low or unclear risk of bias, and low if most information was from studies with a low risk of bias.34
Following data extraction, meta-analysis was possible for one key outcome measure across trials that applied similar interventions and compared with exercise at one assessment time-point (6 months). Meta-analysis was conducted using RevMan to assess the effectiveness of a combined intervention programme of diet and exercise on mobility (6 min walk test at 6 months) using the random effects model.35 36 Ninety-five per cent Cls were reported for the summary statistics and the SD was calculated from the SEs and CIs.37 38 Data for the other outcomes were available, but meta-analyses were not possible due to different assessment points or comparators. A modified narrative synthesis was used to present these data.39 40
Synthesis of results
A mixed method analysis was required to synthesise the available data.35–40 For the meta-analysis, no raw data were available, and therefore data analyses were conducted on the final summary statistics reports. SDs were estimated from reported SE and CI for all available data.36 Heterogeneity in treatment effects was considered by computation of I2. An analysis of the quality of the interventions was undertaken as the basis for interpretation of heterogeneity.36 37 For the modified narrative synthesis, change scores were used for trials when no other data were available.36–38 Two stages of a narrative synthesis were possible to apply; these comprised the development of a preliminary synthesis of findings of included trials, and an exploration of the relationships within and between trials.39 40
Developing a preliminary synthesis
A preliminary synthesis was developed using tabulation, textual description, grouping and clusters and data transformation. Tables were designed presenting the main characteristics of the eligible studies including eligibility criteria, intervention (number of participants, goal of weight loss, intervention period, setting and brief information about exercise and diet intervention), comparator, outcome measures and the main findings. Additional tables were used to organise studies with respect to specific outcome measures (primary or secondary) and the comparator group. Results were presented as mean (SD) by converting the continuous data from SEs or CIs to SD.39 40
Exploring the relationships within and between trials
A visual representation of the relationship between study characteristics and results was used to explore the relationships within and between trials.39 40
The Grading of Recommendation, Assessment, Development and Evaluation (GRADE) approach was used to evaluate the quality of evidence included in the meta-analysis.41 42 Specific software (GRADEpro) was used.43 This approach provided a system for rating the quality of evidence and determining the strength of recommendations for clinical practice guidelines.41 42 It has five components: risk of bias, inconsistency, indirectness, imprecision and publication bias. Quality of evidence was categorised as ‘high’, ‘moderate’, ‘low’ and ‘very low’.41 42 Each RCT evaluated as ‘high’ quality evidence was modified according to five negative and two positive factors.41 42
The Cochrane risk of bias assessment tool was used for this component.34 According to the software, risk of bias was classified as not serious, serious or very serious. The quality of evidence was downgraded by one level if there was a serious limitation or by two levels if the limitation was very serious.41 42 Inconsistency was evaluated according to I² statistics. It may be considered low if I²<40%, moderate if I²=30%–60%, substantial if I²=50%–90% and considerable if I²=75%–100%.42 Inconsistency was considered as unserious if the reviewers were able to identify a plausible explanation for the heterogeneity and the quality of evidence was not downgraded.42 Otherwise, the quality of evidence was downgraded by one or two levels if inconsistency of the results was classified as serious or very serious.42 The quality of evidence was downgraded by one or two levels if there was indirectness between the study question and the applicability of the evidence.41
Imprecision of evidence was downgraded in the presence of the following conditions: first, when the boundaries of the CI crossed the no effect line (threshold is completely within the recommended effect) and second, when the criteria for optimal information size (OIS) were not met.42 The criterion for OIS was that the total number of participants included in a systematic review (calculated from a meta-analysis) was less than the number of participants generated by a conventional sample size calculation for a single adequately powered trial. Imprecision was downgraded by one level if one of these conditions was not met or by two levels if both conditions were not met.41 42
Publication of bias was undetectable or strongly suspected according to GRADE software.43 The selective outcome reporting domain of the Cochrane risk of bias assessment tool was used to evaluate the publication bias.34 42 The quality of evidence was downgraded by one level if the selective outcome reporting domain was evaluated as unclear without justification or downgraded by two levels if evaluated as high.42
Three RCTs (n=794) were included. One was a pilot trial44 and two were definitive trials: the Arthritis, Diet and Activity Promotion Trial (ADAPT)45 and the Intensive Diet and Exercise for Arthritis (IDEA).3 For the ADAPT, there was a main trial report, and two additional articles with further analyses of additional outcome measures.46 47 The trials used two comparators: an exercise programme in the pilot study and IDEA trial,3 44 while usual care (healthy lifestyle) was the comparator in ADAPT.45 All of the included trials were conducted by the same group from the USA and published in English. No relevant unpublished studies were identified. The inter-rater agreement of the study selection process was excellent with k=0.82.32 There was one disagreement requiring consultation with the third reviewer (CAG), who was asked to clarify the eligibility of articles reporting the same trials. Specifically, one pilot study by Messier et al 44 ,44 did not clarify whether it was an external or internal pilot study. The senior author was contacted twice but no response was received. The third reviewer recommended it be treated as an external pilot study as there was nothing to indicate it was an internal pilot study in the article reporting the main trial.45 The study flow diagram is presented in figure 1.48
The main characteristics of the included trials are presented in table 2.
In the pilot trial by Messier et al,44 participants were randomised into two groups, a combined intervention and control group.44 The control group received an exercise programme similar to the intervention group. Messier et al45 randomised participants into four groups: combined intervention, exercise, diet and a control group. The control group received health education plus telephone contact to obtain information on pain, medication use, illness and hospitalisation.45 Messier et al3 randomised participants into three groups: combined intervention, diet group and exercise group. The exercise-alone group was the control. Duration of the trial was 6 months for the pilot trial44 and 18 months for ADAPT45 and IDEA.3
All participants were community dwelling, obese older adults with radiographic evidence of knee OA. A total of 794 participants aged 55 years or older were randomised into the included studies. One hundred and fifty-five participants were included in the meta-analysis.
The pilot trial44 and two definitive trials3 45 were conducted by the same group from Wake Forest University, Winston-Salem, North Carolina, USA. The goal of weight loss varied from 6.8 kg over 6 months to 10%–15% of total body weight over 18 months of intervention. Outcomes were recorded at three time-points for the pilot trial (baseline, 3 months and 6 months) and for the two definitive trials (baseline, 6 months and 18 months). Exercise duration and frequency were similar in all included trials (1 hour/three times per week). Exercise types were aerobic exercise and resistance training. Principles from group dynamics and social-cognitive theory were used for behavioural treatment in the diet group in IDEA.3 The diet sessions were graded from intensive (facilitating behavioural changes by using self-regulatory skills) to transition stage (assisting participants who not reached their weight loss goals in establishing new goals) and maintenance stage (assisting patients who had reached their weight loss goals to maintenance of their weight loss). For the intensive weight loss trial, the daily caloric intake was adjusted according to the rate of weight change between intervention visits (low fat and high vegetable diet). The initial diet plan provided an energy-intake deficit of 800–1000 kcal/day, as predicted by an energy expenditure (estimated resting metabolism×1.2 activity factor), of at least 1200 kcal for men and 1100 kcal for women.3
Due to few eligible studies, analysis was based on all of the outcomes of interest (body weight and BMI as well as musculoskeletal function), irrespective of whether they were specified as the primary or secondary outcome in the included trials (see below):
Messier et al44: no primary or secondary outcomes were specified.
Messier et al45: the ADAPT primary outcome was self-reported physical function measured using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) Secondary outcomes included weight loss, 6 min walk distance, stair-climb time, WOMAC pain and stiffness scores and joint space width.45
Messier et al3: in the IDEA trial, the primary outcomes were knee joint compressive force and plasma interleukin (IL)-6 concentration. Secondary outcome measures included WOMAC pain, WOMAC function, gait speed, 6 min walk test, QOL, body weight, height, BMI and body composition.3
Risk of bias within trials
Substantial inter-reviewer agreement was achieved on the risk of bias assessment (k=0.73).32 All of the included trials were evaluated as unclear risk of bias.34 Most of the key domains were assessed as unclear risk of bias within each trial (table 3).
Risk of bias across trials
Risk of bias across trials was evaluated as unclear,34 only component 5 (selective outcome reporting) was evaluated as low risk of bias for all studies. For the ‘blinding of participants, personnel and outcome assessor’ component, all trials were evaluated as having unclear risk of bias as no strategies were reported to address the issue of outcome assessor unblinding. Also, for the ‘other sources of bias’ components, all trials were evaluated with unclear risk of bias due to unclear reporting.
Results of individual trials and synthesis of results
Meta-analysis was possible for only one outcome measure at one assessment time-point. Meta-analysis was used to assess the effect of the combined intervention programme compared with exercise on the 6 min walk test (metres) after 6 months of intervention. Only two trials3 44 with unclear risk of bias with n=155 participants were available for meta-analysis. The pooled random effects (15.05, 95% CI −11.77 to 41.87) did not support a combined intervention effect (figure 2).
Synthesis of results
Modified narrative synthesis
With respect to the guidelines for a narrative synthesis, only two elements were possible to apply: developing a preliminary synthesis and exploring the relationships within and between studies.39 40
Developing a preliminary synthesis
Tabulation was used to present primary and secondary outcome measures that were not included in the meta-analysis: body weight, knee ROM, physical function, mobility, pain and QOL. Studies including a comparison of the combined intervention programme with an exercise intervention are presented in tables 4-5.
Tabulation was used to compare the effect of the combined intervention programme compared with usual care (healthy lifestyle) on primary and secondary outcome measures: body weight, physical function, mobility, pain and QOL (tables 6-7).
Exploring the relationships within and between studies
Characteristics of the trials are presented in table 2. All included trials were conducted by the same research group. The eligibility criteria were very similar across studies. The exercise intervention included strengthening and aerobic exercise. The intervention frequency (1 hour/three times per week) was the same across the included trials. Further details about the design of the trials are presented above.
Tables 4 and 6 show differences between the intervention group and the control group (despite the comparator) with respect to body weight and the 6 min walk distance. These differences were consistent with the results from the included trials. The diet plus exercise group in the pilot study44 lost weight compared with the control group (p=0.01) after 6 months of intervention44 and this was also the case with respect to the longer duration intervention trial (18 months) in which the intervention group lost significantly (p <0.001) more weight than the exercise group.3 However, in ADAPT45 both groups (intervention and healthy lifestyle) lost weight (p<0.05) after 18 months of intervention,45 although there was a significant difference in the 6 min walk result in favour of the diet plus exercise group (p<0.05).45 Also, there was a significant difference (p=0.005) in the 6 min walk between the intervention and exercise groups in the IDEA trial.3
No further analyses were possible owing to the lack of reported information and low number of included trials.
Grading the quality of evidence
A summary assessment was undertaken to draw conclusions about the overall quality of evidence for the combined intervention on mobility using GRADE software.43 Both trials included in the meta-analysis3 44 were evaluated as ‘high’ quality evidence before being downgraded as they were RCTs, before being modified according to five negative and two positive factors.41 42 The quality of evidence for a combined intervention programme of physical activity and diet on walking distance (metres) within 6 min after a period of 6 months of intervention was evaluated as moderate (table 8).
Summary of evidence
This is the first systematic review and mixed method analysis investigating the effectiveness of combining dietary restriction and physical activity interventions for musculoskeletal function in overweight/obese older adults with knee OA. One pilot trial44 and two definitive trials3 45 (794 participants) conducted by the same research group (Wake Forest University) were included. The intervention programme was compared with exercise training in one definitive trial (IDEA)3 and the pilot trial,44 while usual care was the comparator in the ADAPT.45 Two additional articles46 47 which reported further outcomes of the ADAPT were identified.45
Data syntheses of this review were conducted using both meta-analysis and modified narrative synthesis. Although visual inspection of the tables of results indicated that the combined programme enhanced body weight reduction, and improved mobility, there was moderate evidence for no effect. Changes of BMI scores were not reported in the included studies. Meta-analysis was possible for only the 6 min walk test at 6 months and was not possible for the other outcome measures due to the inconsistency of assessment points or the comparator. The pooled random effect of two trials3 44 with 155 participants did not support the combined intervention programme (15.05, 95% CI −11.77 to 41.87) (although with a total effect of 15 m deemed not clinically significant according to previous literature).49 50 Although the meta-analyses showed substantial heterogeneity I²=50%, this was classified as not serious using the GRADE evaluation tool41 42 as it was assessed as likely to be due to high variability in both the sample size and effect size. Clinical heterogeneity across trials was limited to comparator and duration. Overall, the quality of evidence was downgraded to moderate due to imprecision of the results according to GRADE.43 All included trials were reported as having an unclear risk of bias, which was mainly due to unclear reporting of some information.34 For instance, both the ‘blinding of participants, personnel and outcome assessor’ and the ‘other sources of bias’ component were evaluated as unclear for all trials.
Results from the trial by Messier et al44 indicated no statistically significant differences across groups with regard to self-reported performance measures of physical function, knee pain scores, knee strength and biomechanical measures (synovial fluid, keratan sulfate and level of IL-1) after 6 months of intervention. Findings from Messier et al45 indicated a statistically significant benefit of the combined intervention in terms of self-reported physical function, 6 min walk test, stair climb and knee pain. The findings from Messier et al3 indicated a significant improvement in the 6 min walk test and walking speed in the intervention group. Moreover, there was a significant reduction (p<0.05) of body weight among the intervention groups in all trials. In the current review, the finding of no effect of a combined intervention programme may be due to the very low number of included trials (and participants) but probably is not due to low compliance. Compliance of the diet and exercise group to the exercise programme at 6 months was higher in the pilot study44 compared with the IDEA trial.3 In the pilot study,44 compliance (ratio of the number of exercise sessions attended to the total number of the exercise sessions prescribed with the exercise programme) was 82.6% for the exercise group and 94.7% for diet plus exercise group. For the IDEA trial,3 399/454 participants (88%) completed the study; compliance of the diet and exercise group was 70% at 6 months and 58% at 18 months with no adverse events and no significant differences between groups.
In addition to diet and exercise, two current trials MEDIC128 and MEDIC227 have reported that a multimodal approach of education, neuromuscular exercise, insoles and, if indicated, a dietary weight loss programme and pain medication are effective for adults and older adults with moderate-to-severe knee OA. These studies were not included in this review due to the wide age range across participants. MEDIC128 included 9 participants and MEDIC227 included 12 participants below the age of 55 years and there was no subgroup analysis of older participants. In MEDIC1,28 the participants were eligible for TKR and were randomised to non-surgical and surgical treatment followed by the intervention programme. Both interventions showed substantial improvement, but the surgical treatment resulted in greater pain relief and functional improvement after 12 months compared with non-surgical treatment alone. However, only 26% of the patients who were assigned to receive non-surgical treatment alone underwent TKR in the following year.28 In MEDIC2,27 participants had radiographic confirmation of OA (Kellgren-Lawrence grade ≥1), but were not eligible for a TKR. The 12-week non-surgical treatment programme consisted of individualised progressed neuromuscular exercise, patient education, insoles, dietary advice and prescription of pain medication if indicated, while usual care comprised two leaflets with information and advice on knee OA and recommended treatments. This non-surgical treatment programme was found to be more effective with respect to pain, activities of daily living and QOL at 12 months compared with usual care, although it was not possible to determine which of the components within this multi-intervention programme were most effective and whether the intervention as a whole would be equally effective in older patients with OA.27
The main limitation of this review is that only few eligible studies were identified. Thus, the optimal components of dietary and exercise interventions in terms of type, duration and quantity suitable for this population are still unclear. Future studies are required in this field to optimise outcome measures and methods of delivering a programme at an acceptable cost, prior to a future adequately powered definitive trial.
Based on current evidence synthesised in this review, it is hard to judge the effectiveness of a combined programme of diet and physical activity due to the low number of included trials and participants and the quality of available evidence. Only moderate quality evidence was available to investigate the intervention programmes. However, the narrative synthesis suggests that interventions with a focus on reduction of body weight and/or improved mobility are worthy of further evaluation. Further adequately powered RCTs testing the effects of a combined intervention against each component individually are required to optimise diet and exercise interventions using a multimodal approach.
The authors would like to thank Lynn Harris, Library Services, University of Birmingham for her contribution to the development of the search strategies.
Contributors ASA led the review and contributed to all stages, including development of the search strategy, running the searches, assessing trials eligibility, data extraction and synthesis (including risk of bias and Grading of Recommendation, Assessment, Development and Evaluation analysis), and preparation of the manuscript. CAG and ABR provided expertise in systematic review methodology and risk of bias analysis. AMK assisted with assessment of eligibility of included studies and data extraction. All authors read and provided feedback on the preparation of this manuscript according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2009 checklist and approved the final manuscript.
Funding This review is a part of a PhD programme undertaken by ASA, and is fully funded by the King Saud University, Kingdom of Saudi Arabia, Riyadh and the Saudi Arabian Cultural Bureau, United Kingdom, London.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no additional unpublished data from the review.
Correction notice This paper has been amended since it was published Online First. Owing to a scripting error, some of the publisher names in the references were replaced with 'BMJ Publishing Group'. This only affected the full text version, not the PDF. We have since corrected these errors and the correct publishers have been inserted into the references.