- Meghan B Azad, Banting postdoctoral fellow1,
- J Gerard Coneys, core resident2,
- Anita L Kozyrskyj, associate professor 1,
- Catherine J Field, professor4,
- Clare D Ramsey, assistant professor and attending physician23,
- Allan B Becker, senior investigator and section head of allergy and immunology56,
- Carol Friesen, librarian78,
- Ahmed M Abou-Setta, research associate7,
- Ryan Zarychanski, attending physician, assistant professor, and director of knowledge synthesis237
- 1Department of Pediatrics, University of Alberta, Edmonton, AB, Canada T6G 1C9
- 2Department of Internal Medicine, University of Manitoba, Winnipeg, MB, Canada
- 3Department of Community Health Sciences, University of Manitoba, Canada
- 4Department of Agriculture, Food and Nutritional Sciences, University of Alberta
- 5Manitoba Institute of Child Health, Winnipeg, Canada
- 6Pediatrics and Child Health, University of Manitoba, Canada
- 7George & Fay Yee Centre for Healthcare Innovation, University of Manitoba/Winnipeg Regional Health Authority, Winnipeg, Canada
- 8Neil John Maclean Health Sciences Library, University of Manitoba Libraries, Winnipeg, Canada
- Correspondence to: M B Azad
- Accepted 8 October 2013
Objective To evaluate the association of probiotic supplementation during pregnancy or infancy with childhood asthma and wheeze.
Design Systematic review and meta-analysis of randomised controlled trials.
Data sources Medline, Embase, and Central (Cochrane Library) databases from inception to August 2013, plus the World Health Organization’s international clinical trials registry platform and relevant conference proceedings for the preceding five years. Included trials and relevant reviews were forward searched in Web of Science.
Review methods Two reviewers independently identified randomised controlled trials evaluating probiotics administered to mothers during pregnancy or to infants during the first year of life. The primary outcome was doctor diagnosed asthma; secondary outcomes included wheeze and lower respiratory tract infection.
Results We identified 20 eligible trials including 4866 children. Trials were heterogeneous in the type and duration of probiotic supplementation, and duration of follow-up. Only five trials conducted follow-up beyond participants’ age of 6 years (median 24 months), and none were powered to detect asthma as the primary outcome. The overall rate of doctor diagnosed asthma was 10.7%; overall rates of incident wheeze and lower respiratory tract infection were 33.3% and 13.9%, respectively. Among 3257 infants enrolled in nine trials contributing asthma data, the risk ratio of doctor diagnosed asthma in participants randomised to receive probiotics was 0.99 (95% confidence interval 0.81 to 1.21, I2=0%). The risk ratio of incident wheeze was 0.97 (0.87 to 1.09, I2=0%, 9 trials, 1949 infants). Among 1364 infants enrolled in six trials, the risk ratio of lower respiratory tract infection after probiotic supplementation was 1.26 (0.99 to 1.61, I2=0%). We adjudicated most trials to be of high (ten trials) or unclear (nine trials) risk of bias, mainly due to attrition.
Conclusions We found no evidence to support a protective association between perinatal use of probiotics and doctor diagnosed asthma or childhood wheeze. Randomised controlled trials to date have not yielded sufficient evidence to recommend probiotics for the primary prevention of these disorders. Extended follow-up of existing trials, along with further clinical and basic research, are needed to accurately define the role of probiotics in the prevention of childhood asthma.
Systematic review registration PROSPERO (CRD42013004385).
Over the past half century, there has been a sharp rise in the global prevalence of asthma, particularly in children.1 About 300 million people worldwide are estimated to have asthma, and the prevalence has been increasing by 50% every decade.1 As the most common chronic disease of childhood, asthma affects roughly one in five children in the United Kingdom and United States,2 and is the leading cause of school absenteeism.3 The total annual cost of asthma to society has been estimated at €19bn (£16.2bn; $26.2bn) in Europe4 and $56bn in the US.5 Recurrent wheeze frequently precedes the diagnosis of asthma, and is estimated to occur in more than 20% of infants.6 7
The microflora hypothesis of allergic disease has been proposed to explain the rising incidence of asthma and other allergic disorders.8 Commensal gut bacteria stimulate development of the neonatal immune system; therefore, disruption of the gut microbiota during early life may contribute to immune disorders later in childhood.9 Indeed, prospective studies have shown that perturbation of the infant gut microbiota precedes development of atopic dermatitis (allergic eczema),10 11 12 which is widely regarded as the first step in the progressive “atopic march” towards allergic rhinitis and asthma.13 Moreover, early life factors that disrupt the gut microbiota (such as caesarean delivery, lack of breastfeeding, and use of antibiotics) increase the risk of asthma.14 15 16
In the light of this evidence, probiotics—live micro-organisms that, when administered in adequate amounts, confer a health benefit on the host—have been proposed for the prevention and treatment of allergic disorders including asthma.17 18 Although naturally present in fermented foods, probiotics are increasingly being produced and administered as supplements in preventative and therapeutic medicine.19 A recent meta-analysis of 14 randomised controlled trials showed that probiotic supplementation during pregnancy or infancy decreased the incidence of atopic dermatitis by 21%.20 Less clinical evidence exists for probiotics in the prevention of wheeze or asthma,21 but animal studies have shown that perinatal use of probiotics can prevent airway inflammation and hyper-reactivity.22 23 A 2007 Cochrane review of early life probiotics for prevention of allergic diseases reported no benefit for asthma prevention after probiotic supplementation,24 based on findings from three trials25 26 27 enrolling a total of 617 infants. All three existing trials have since published extended follow-up results,28 29 30 31 and six additional trials enrolling 2308 infants have published new findings on probiotics for asthma prevention.21 32 33 34 35 36 Prevention of related outcomes (wheeze or lower respiratory infection) was not covered in the 2007 Cochrane review.
The purpose of this systematic review was to identify, critically appraise, and meta-analyse data from prospective randomised trials evaluating the use of probiotic supplements for the primary prevention of asthma or childhood wheeze.
Using an a priori published protocol,37 we conducted our systematic review using methodological approaches outlined in the Cochrane Handbook for Systematic Reviewers38 and reported according to the criteria of preferred reporting items for systematic reviews and meta-analyses (PRISMA).39 A technical panel of experts from multiple fields (nutrition, paediatric asthma, research methodology) formulated the review questions, reviewed the search strategies and review methods, and provided input throughout the review process.
Populations, interventions, comparators, outcome measures, settings, and study designs
We included only randomised controlled trials of pregnant women or healthy infants under 1 year of age (web appendix, table S1). Our primary research question was “In healthy infants under 1 year of age, are probiotics (administered prenatally or postnatally) safe and protective against the development of wheeze or asthma, compared to placebo, or to no intervention?” The main outcome measure was doctor diagnosed asthma. Secondary outcomes included wheeze (incident or recurrent), asthma drug use, hospital admission for asthma, and the asthma predictive index score.40 Lower respiratory tract infections were also included as a secondary outcome because these infections generally involve wheeze and could predispose for asthma.41 Safety outcomes included gastrointestinal disturbances, allergic reaction to probiotics, and withdrawal due to perceived side effects. Table S2 presents inclusion and exclusion criteria (web appendix).
Search strategy for identification of studies
We searched Medline, Embase, and Central (Cochrane Library) databases from inception to August 2013 for relevant citations of published trials, using individualised search strategies for each database. Table S3 presents the Medline strategy (web appendix). We searched the World Health Organization’s international clinical trials registry platform and relevant conference proceedings for the preceding five years to identify relevant planned, ongoing, or recently completed but unpublished trials (American Thoracic Society; American Academy of Allergy, Asthma, and Immunology; European Respiratory Society; European Academy of Allergy and Clinical Immunology; and the American Association for Parenteral and Enteral Nutrition). We performed forward searches of included trials and relevant reviews in the Web of Science to identify additional citations, and contacted study authors to request relevant unpublished data. Reference lists of narrative and systematic reviews and of the included trials were searched for additional citations. We performed reference management in EndNote X6 (Thomson Reuters).
We used a two stage process for study screening and selection using standardised and piloted screening forms. Firstly, two reviewers (MBA and JGC) independently screened the titles and abstracts of search results to determine whether each citation met the inclusion criteria. Second, the full text versions of potentially relevant citations were reviewed independently with reference to the predetermined inclusion and exclusion criteria (web appendix, table S2). Discrepancies between the two reviewers were resolved through consensus by discussion with a third reviewer (AMA-S), as required.
Data abstraction and management
Two reviewers (MBA and JGC) independently extracted data from included trials, using standardised and piloted data extraction forms. Discrepancies between the two reviewers were resolved through consensus in discussion with a third reviewer (AMA-S), as required. Extracted data included funding sources, demographics of the enrolled mothers and infants (family history of allergic disease, mode of delivery, infant sex, gestational age or age at enrolment, and breastfeeding), details of probiotic intervention (organism, daily dose, timing, duration, and method of administration), and relevant outcomes as described above. Since the nature of asthma and wheeze can change over time,42 outcome data were extracted for predefined time periods (age <3 years, 3 to <6 years, and ≥6 years). When a trial reported results for more than one time period, results from the longest follow-up were included in the main meta-analysis; results from earlier time periods were included in subgroup analyses. Data management was performed using Microsoft Excel 2007.
Assessment of methodological quality and potential risk of bias
We evaluated internal validity using the Cochrane Collaboration’s risk of bias tool,43 which assesses randomisation and allocation of participants; blinding of participants, personnel, and outcome assessors; incomplete or selective reporting; and other relevant sources of bias. If trial methodology was unclear from the published report, authors were contacted for clarification.
Measures of treatment effect
We analysed data from the included trials using Review Manager (RevMan 5.2, the Cochrane Collaboration).44 A formal meta-analysis was conducted if the data were statistically and clinically homogeneous. Pooled dichotomous data were expressed as a risk ratio, or Peto odds ratio in the event of rare outcomes.45 A risk ratio or odds ratio less than one suggests a lower rate of the outcome (for example, asthma) among participants randomised to receive probiotics than among the control group. We used the random effects model for all analyses, with the exception of the Peto odds ratio (fixed effect model). Statistical heterogeneity was explored and quantified using the I2 statistic.46 All tests of statistical inference reflect a two sided α value of 0.05.
Subgroup analyses and meta-regression
We performed subgroup analyses to determine summary effect estimates in several prespecified groups, including: the participant receiving the intervention (mother or infant), duration and timing of intervention (prenatal or postnatal), probiotic organism and dose, duration of follow-up or age at assessment, asthma risk, caesarean delivery rate, geographical area, and source of funding.37 We conducted meta-regression to evaluate differences in effect according to duration of follow-up as a continuous variable, using Comprehensive Meta Analysis (Biostat).47
Trial characteristics and study populations
Of 3011 citations identified from electronic and hand searches, we included 20 unique randomised trials enrolling a total of 4866 infants (fig 1⇓, table 1⇓). These trials were represented by 20 primary articles,21 25 26 27 32 33 34 35 36 48 49 50 51 52 53 54 55 56 57 58 four companion articles,59 60 61 62 10 extended follow-up publications28 29 30 31 63 64 65 66 67 68 plus one forthcoming report, and eight duplicate conference abstracts.69 70 71 72 73 74 75 76 All were double blind, placebo controlled trials published in peer reviewed journals between 2001 and 2013. Sixteen trials were conducted in Europe, while four trials were conducted in Australia,27 49 New Zealand,36 and Taiwan.34 Based on family history or existing allergic disease in the mother or infant (definitions provided in web appendix, table S4), 14 of 20 trials enrolled participants at high risk for asthma; the remaining six trials21 35 50 53 56 57 were conducted in unselected populations. The caesarean delivery rate in study populations ranged from 0% to 45%, and was not reported in six trials.21 26 34 48 51 52 Most trials (14 of 20) did not restrict infant feeding practices, although six trials50 53 54 56 57 58 required exclusive formula feeding for enrolment. Nearly all trials (19 of 20) reported some degree of industry support (funding, salary support, or supplied products), including seven trials involving authors employed by the industry sponsor.32 33 50 53 56 57 58
Overall, most trials were adjudicated to be of unclear (nine of 20) or high risk of bias (10 of 20); only one trial55 was considered to have a low risk of bias across all domains (web appendix, table S5; fig S1). Of 20 trials, most had adequate random sequence generation (n=18), allocation concealment (n=18), and blinding of outcome assessment (n=18). However, eight and nine trials were subject to unclear and high risk of attrition bias, respectively, owing to incomplete outcome data after substantial loss to follow-up. Three trials were at high risk of performance bias due to unblinding of study participants at extended follow-up assessments (including one trial with unpublished data).28 68
Table 2⇓ presents details of the probiotic interventions administered in each trial. All trials evaluated probiotic supplements, rather than consumer food products. Supplements were delivered orally by various methods, including capsules; oil droplets; and suspensions in water, milk, or infant formula. One trial49 exclusively evaluated prenatal maternal supplementation, 10 exclusively evaluated postnatal infant supplementation, while nine evaluated a combination of prenatal and postnatal supplementation. The total duration of intervention ranged from one to 25 months (median 6.3 months). Various probiotic organisms were tested in isolation or in combination, including four Bifidobacterium species (B bifidum, B longum, two strains of B breve, and four strains of B lactis), and six Lactobacillus species (L acidophilus, L casei, L lactis, L reuteri, two strains of L paracasei, and three strains of L rhamnosus). Six trials21 32 33 48 50 54 evaluated combinations of multiple probiotic organisms, and five32 50 56 57 58 evaluated probiotics in combination with prebiotics (selectively fermented compounds that facilitate changes in the composition or activity of the gut microbiota to confer benefits on host health).77 The daily dose of probiotics ranged from 108 to 1011 colony forming units, and was not quantifiable in six trials using supplemented infant formulas that were fed without restraint.50 53 54 56 57 58 Compliance was assessed by a variety of methods, including maternal interview or daily diaries, counting of unused supplements, and faecal analysis.
Among the included trials, the duration of follow-up ranged from four months to eight years, and the median age at final assessment was 24 months (table 1). Nine trials reported clinical asthma diagnosis, 11 reported wheezing outcomes, and six reported lower respiratory tract infections; table S6 provides individual study definitions for these outcomes (web appendix). One trial reported the asthma predictive index score,49 two reported asthma drug use,54 58 and none reported admission to hospital for asthma. Adverse events were inconsistently reported.
Primary outcome: asthma
Nine trials including 3257 children contributed asthma data for meta-analysis (fig 2⇓). Incidence of doctor diagnosed asthma at final assessment was 11.2% among patients randomised to receive probiotics and 10.2% among those receiving placebo (risk ratio 0.99, 95% confidence interval 0.81 to 1.21, I2=0%). Results were similar when expressed as a Peto odds ratio for rare events.
Secondary outcomes: wheeze, lower respiratory tract infection, and adverse events
Nine trials including 1949 children contributed incident wheeze data for meta-analysis (fig 3⇓). Incident wheeze at final assessment was similar after supplementation with probiotics or placebo (35.0% v 31.1%; risk ratio 0.97, 95% confidence interval 0.87 to 1.09, I2=0%). Three trials reported recurrent wheezing, and meta-analysis was not pursued owing to considerable statistical heterogeneity (I2=83%). Two of these trials29 55 reported an increased risk of recurrent wheeze after probiotic supplementation, whereas the third trial58 reported a decreased risk (web appendix, fig S2).
Six trials including 1364 children contributed data on lower respiratory tract infections. The incidence of lower respiratory tract infection was 14.5% among children randomised to receive probiotics, and 13.2% among those who received placebo (risk ratio 1.26, 95% confidence interval 0.99 to 1.61, I2=0%). Notably, four of six trials documented lower respiratory tract infections non-systematically as adverse events,48 50 52 53 rather than as primary or secondary outcomes. Excluding these four trials, the pooled risk ratio of lower respiratory tract infection associated with probiotics was 1.11 (0.70 to 1.76, I2=35%; fig 4⇓).
Most trials did not systematically screen for or report the incidence of relevant safety outcomes, including severe gastrointestinal disturbances or allergic reactions (web appendix, table S7). The Peto odds ratio associated with withdrawal due to perceived side effects was 1.45 (95% confidence interval 0.66 to 3.17, I2=0%; eight trials, 2732 children; web appendix, fig S3).
We evaluated the efficacy of probiotics for prevention of asthma in children according to predefined subgroups (table 3⇓). Subgroup analyses by participant type (mother, infant, or both), timing of intervention (prenatal, postnatal, or both), or duration of intervention (≤6 or >6 months) did not show significant differences. However, individual subgroups were subject to type II errors, owing to small sample sizes. Similarly, we observed no statistical differences according to baseline asthma risk, probiotic dose or organism, caesarean delivery rate, feeding restrictions, geographical area, risk of bias, or industry authorship. Differences were not observed according to duration of follow-up, whether classified according to predefined strata (table 3) or assessed as a continuous variable by meta-regression (web appendix, fig S4). We found no significant differences across subgroups for incident wheeze (table 3); subgroup analyses were not pursued for recurrent wheeze or lower respiratory tract infection because of the small number of trials reporting these outcomes.
In this systematic review and meta-analysis of randomised controlled trials, we found no evidence to support a protective association between probiotic supplementation during pregnancy or early life, and subsequent development of childhood asthma or wheeze. Although inadequately reported, probiotic supplementation could be associated with clinically relevant increases in lower respiratory tract infections.
Comparison of results with other studies
Our review provides a timely update to the 2007 Cochrane review of probiotics for prevention of allergic diseases.24 The Cochrane review reported no benefit for the prevention of asthma after probiotic supplementation, based on findings from three trials enrolling a total of 617 infants.25 26 27 Our review evaluates updated long term findings from these three original trials,28 29 30 31 and adds results from six new trials (2308 infants) reporting on probiotics for asthma prevention.21 32 33 34 35 36 Furthermore, we have evaluated 11 additional probiotic trials (1976 infants) reporting asthma related outcomes (wheeze or lower respiratory infection) that were not analysed in the 2007 Cochrane review. After systematic evaluation of these new and extended trial results, involving over 4000 additional children, we conclude that there is still insufficient evidence to recommend probiotics for the primary prevention of asthma. Our findings further identify several unanswered questions and highlight opportunities for future research.
Opportunities for future research
Despite widespread enthusiasm for evaluating probiotics to prevent allergic disease17 and recognition of asthma as a major allergic disease in childhood,1 78 relatively few randomised trials have formally tested the use of probiotics for asthma prevention. Only nine probiotic trials have reported asthma diagnosis, and none were powered for asthma detection as the primary outcome. Long term follow-up is essential for asthma prevention studies because diagnosis is challenging before age 6 years.79 80 However, among the 20 trials included in our review, the median age at last follow-up was 24 months, and only five trials reported outcomes at or beyond 6 years of age (including one trial with unpublished data).28 30 67 68 Moreover, long term follow-up was frequently subject to high or unclear risk of bias owing to attrition or the unblinding of participants. Extended follow-up data from other established trials are highly anticipated, including planned adolescent assessments by Kalliomaki26 and Kukkonen32 and colleagues. Because of the paucity of long term follow-up data among probiotic trials, we also evaluated wheeze as an early presentation of asthma. However, only a minority of wheezing infants will ultimately develop asthma later in childhood.81
Thus, extended follow-up of existing studies, combined with novel, respiratory focused trials,82 83 84 will be necessary to define the role of probiotics for asthma prevention. Owing to the dynamic nature of the gut microbiota, trials evaluating prolonged probiotic supplementation (beyond the first year of life) may also be needed. As West and colleagues have shown,67 probiotics are transient colonisers of the intestine, indicating that prolonged supplementation may be required to achieve durable benefit.
Our findings also highlight a need to further consider the effect of probiotics on the incidence and severity of recurrent wheeze and lower respiratory tract infections. Recurrent wheeze during infancy is considered a better predictor of asthma than incident wheeze,40 yet only three of 10 trials documenting wheeze reported variable measures of recurrence,27 55 58 and statistical heterogeneity precluded meta-analysis. Reporting of lower respiratory tract infections was similarly incomplete and variably defined. In six trials reporting this outcome, we observed a trend toward increased infections in children who received probiotics. Future trials should systematically define and capture recurrent wheeze, along with lower respiratory tract infections and other relevant safety outcomes.
Our subgroup analyses indicated that the effect of probiotics was similar regardless of the timing of intervention (prenatal v postnatal v both) or the participant receiving the intervention (mother v infant v both). The efficacy of specific probiotic organisms was difficult to assess because of the large variety of strains, combinations, and doses tested, and requires further investigation. Different probiotic organisms probably have distinct effects on the gut microbiota and host physiology. A recent animal study found that four Lactobacillus species had markedly different immunomodulatory effects,85 and at least one clinical study has shown strain specific anti-allergic effects.36 Further basic and clinical research is also needed to characterise the mechanisms by which probiotics influence asthma development, including how specific organisms colonise the gut, modify the resident microbiota, and ultimately affect host immunity and health. Such knowledge will help optimise the selection of probiotic organisms and the design of intervention regimens for future study.
Finally, identifying infant populations most likely to benefit from probiotics is highly desirable. For example, one trial has shown that probiotics were protective against IgE associated allergic disease in infants delivered by caesarean section (whose gut microbiota is disrupted86), but not in their vaginally delivered counterparts.64 With trial level data, we could not identify differential efficacy according to caesarean delivery rate, but this and other microbiota disrupting exposures (such as formula feeding and antibiotic treatment) warrant further study as possible indications for probiotic supplementation in the prevention of asthma and wheeze.
Strengths and weaknesses of the study
The strengths of this review included the completeness of the search strategy, which reviewed multiple citation databases, trial registries, and conference proceedings. By omitting outcome related search terms, we identified trials that were not primarily focused on asthma or allergic disease, but nevertheless reported relevant outcomes.50 53 56 57 We focused on patient centred outcomes and evaluated efficacy in the context of relevant safety outcomes and adverse events. Finally, we used an a priori published protocol and followed established methodological guidelines in the conduct and reporting of this review. Limitations include pooling data from trials conducted in distinct populations (for example, infants at high risk for asthma, or unselected populations) receiving different probiotic formulations (various organisms with a 1000 times range in daily dose) through varying regimens (prenatal or postnatal supplementation, for 1-25 months). Subgroup analyses were susceptible to type II errors owing to relatively small sample sizes.
We found no evidence to support a protective association between perinatal administration of probiotics, and doctor diagnosed asthma or childhood wheeze. There is currently insufficient evidence to recommend probiotics for the primary prevention of these disorders, and further research is warranted to explore the potential association between probiotic supplementation and increased risk of lower respiratory tract infection. Extended follow-up of existing trials, along with further clinical and basic research, are needed to accurately define the role of probiotics in the prevention of childhood asthma.
What is already known on this topic
Asthma is the most common chronic disease of childhood, and is frequently preceded by wheeze
Increases in asthma prevalence have been partly attributed to disruption of the commensal gut microbiota in early life
Perinatal probiotics have been shown to prevent atopic dermatitis, but uncertainty remains regarding their effectiveness in asthma prevention
What this study adds
We found no evidence to support a protective association between perinatal probiotics and childhood asthma or wheeze. Although inadequately reported, probiotic supplementation could be associated with increases in lower respiratory tract infections
Additional basic research and adequately powered long term clinical trials are needed to fully define the role of probiotics in the prevention of asthma
Probiotics cannot be recommended for primary prevention of childhood asthma or wheeze at this time
Cite this as: BMJ 2013;347:f6471
We thank the George and Fay Yee Centre for Healthcare Innovation for infrastructure support, and the following authors for providing unpublished data: G T Rijkers (University College Roosevelt Academy, The Netherlands) D M W Gorissen (University Medical Centre Utrecht, The Netherlands), J Hol (St Elisabeth Hospital, Curacao), P Steenhout (Nestec), K Kukkonen and E Savilahti (Helsinki University Central Hospital, Finland), and K Wickens (University of Otago, New Zealand).
Contributors: MBA, AMA-S, and RZ conceived and designed the study with input from ABB, CJF, CDR, and ALK. CF provided guidance on search strategies. MBA and JGC screened and reviewed studies, extracted data, and assessed the quality of included trials. MBA analysed the data and all authors contributed to its interpretation. MBA drafted the manuscript and all authors participated in the revision process and have approved this submission for publication. MBA is guarantor, had full access to all of the data in the study, and can take responsibility for the integrity of the data and the accuracy of the data analysis.
Funding: No specific funding was obtained for this study. MBA is a Canadian Institutes of Health Research Banting postdoctoral fellow, and recipient of a Parker B Francis research opportunity award and Alberta Innovates Health Solutions incentive award. RZ is a recipient of a randomised controlled trial mentorship award from the Canadian Institute of Health Research. ALK is the research chair in maternal-child health and the environment at the Women and Children’s Health Research Institute. None of the funders influenced the conduct of this research.
Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Ethical approval: As this study did not involve the collection of new data, but rather the analysis of data from previously published research, no ethics approval was sought.
Data sharing: The RevMan file used to generate summary effect measures will be made available on request.
MBA, as guarantor, affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned and registered have been explained.
This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) 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/3.0/.