Article Text

The relationship between rowing-related low back pain and rowing biomechanics: a systematic review
  1. Frank James Nugent1,2,
  2. Anders Vinther3,
  3. Alison McGregor4,
  4. Jane S Thornton5,
  5. Kellie Wilkie6,
  6. Fiona Wilson7
  1. 1 Physical Education and Sport Sciences Department, Faculty of Education and Health Sciences, University of Limerick, Limerick, Ireland
  2. 2 Sport and Human Performance Research Cluster, Health Research Institute, University of Limerick, Limerick, Ireland
  3. 3 Department of Physiotherapy and Occupational Therapy, Herlev Gentofte Hospital, University of Copenhagen, Copenhagen, Denmark
  4. 4 Department of Surgery and Cancer, Imperial College London, London, UK
  5. 5 Western Centre for Public Health and Family Medicine, University of Western Ontario, London, Ontario, Canada
  6. 6 Bodysystem Physiotherapy, Hobart, Tasmania, Australia
  7. 7 Discipline of Physiotherapy, School of Medicine, Trinity College Dublin, Dublin, Ireland
  1. Correspondence to Dr Frank James Nugent, Physical Education and Sport Sciences Department, Faculty of Education and Health Sciences, University of Limerick, Limerick, Ireland; frank.nugent{at}ul.ie

Abstract

Background Low back pain (LBP) is common in rowers. Understanding rowing biomechanics may help facilitate prevention and improve rehabilitation.

Objectives To define the kinematics and muscle activity of rowers and to compare with rowers with current or LBP history.

Design Systematic review.

Data sources EMBASE, MEDLINE, Cumulative Index to Nursing and Allied Health Literature, Web of Science and Scopus from inception to December 2019. Grey literature was searched.

Study eligibility criteria Experimental and non-experimental designs.

Methods Primary outcomes were kinematics and muscle activity. Modified Quality Index (QI) checklist was used.

Results 22 studies were included (429 participants). Modified QI score had a mean of 16.7/28 points (range: 15–21). Thirteen studies investigated kinematics and nine investigated muscle activity. Rowers without LBP (‘healthy’) have distinct kinematics (neutral or anterior pelvic rotation at the catch, greater hip range of motion, flatter low back spinal position at the finish) and muscle activity (trunk extensor dominant with less flexor activity). Rowers with LBP had relatively greater posterior pelvic rotation at the catch, greater hip extension at the finish and less efficient trunk muscle activity. In both groups fatigue results in increased lumbar spine flexion at the catch, which is greater on the ergometer. There is insufficient evidence to recommend one ergometer type (fixed vs dynamic) over the other to avoid LBP. Trunk asymmetries are not associated with LBP in rowers.

Conclusion Improving clinicians’ and coaches’ understanding of safe and effective rowing biomechanics, particularly of the spine, pelvis and hips may be an important strategy in reducing incidence and burden of LBP.

  • lumbar spine
  • pelvis
  • sports rehabilitation programs
  • rowing

Statistics from Altmetric.com

Background

Rowing is a cyclical endurance sport that typically involves high training volumes.1 2 Rowers from junior to international levels perform in excess of 13 hours (or 11 sessions) of training per week.1 3 This places large demands on the musculoskeletal system, particularly the lumbar spine. Low back pain (LBP) is common in rowers4–6 and although there is currently no consensus on definition of LBP for athletes, it can be defined as a symptom, not a disease that can result from several different or unknown abnormalities or diseases. It is further defined by the location of pain, typically between the lower rib margin and buttock creases and can be accompanied by pain in one or both legs in some people.7 LBP affects up to 53% of international rowers per year,8 55% of national rowers per year9 and 24% of junior rowers5 across a training season. In rowers that reported LBP, 18% missed in excess of 1 month of training10 and 53%–79% reported LBP reccurrence.8 11 Therefore, reducing initial risk of LBP is a key priority for clinicians and coaches who work with rowers in order to minimise training time lost to injury and protect long-term health.

A number of risk factors contribute to LBP in rowers including a previous history of LBP,8 11 high training volume,8 ergometer training4 and suboptimal rowing biomechanics.12–14 In recent years, there has been growing interest in the role of biomechanical analysis and injury reduction. The biomechanical demands of the rowing stroke involve repetitive flexion and loading of the lumbar spine. Lumbar flexion significantly increases as training session duration and intensity increase, which may lead to an increased risk of LBP.15–17 Improving clinicians’ and coaches’ understanding of safe and effective rowing biomechanics, particularly of the spine, pelvis and hips, may be an important strategy in reducing the incidence and severity of LBP.18 Despite the increasing amount of research in this area,13 15 16 19–29 no consensus exists on the relationship between LBP and rowing biomechanics. The aim of this study was to systematically review the current literature in order to define the kinematics and muscle activity of rowers (with no history of LBP and hereby defined for convenience as ‘healthy’) and to compare with rowers with current or a history of LBP.

Methods

The reporting of the systematic review was guided by the Preferred Reporting Items for Systematic Reviews and Meta-analyses recommendations.30 The study was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 13 December 2017 (registration number CRD42017083654).

Study eligibility criteria

All experimental designs (randomised controlled trials and quasi-randomised trials) and non-experimental designs (cross-sectional studies, case-control studies, case series and case studies) were considered for inclusion. Studies not in the English language or where a translation was not available were excluded. Participants were rowers (sweep and scull) of all boat categories (including traditional, coastal and tour rowing) and included asymptomatic rowers and those reporting LBP. No restrictions were placed on recruitment setting, sex or activity levels of participants or duration of symptoms (for those reporting LBP). LBP in rowers was defined as condition specific, non-specific, acute and chronic. In addition, LBP included symptoms secondary to the condition such as radiculopathy (sciatica) or other forms of associated leg symptoms. Where possible, those reporting LBP were compared with a control group which was defined as an asymptomatic group or an age-matched, sex-matched and/or activity-matched group. The primary outcomes of interest were (1) kinematics and (2) muscle activity.

Sources and study selection

Electronic databases were searched from inception to December 2019. These included EMBASE, MEDLINE, Cumulative Index to Nursing and Allied Health Literature, Web of Science and Scopus. There were no restrictions placed on language or publication period. All search strategies were completed by a university librarian experienced in the methodology and are summarised in online supplementary material. Grey literature searching included searching reference lists of published studies and contacting prominent authors in the area. Three review authors (FJN, FW and AMcG) independently screened each study for inclusion based on the title, keyword and abstract. Subsequently, full-text studies were independently assessed for eligibility by review authors (FJN, FW and AMcG); disagreements between reviewers were resolved by consensus.

Supplemental material

Data extraction and management

Three review authors (FJN, FW and AMcG) extracted data independently using a customised data extraction form. This included details of participant characteristics, sample sizes, methods, interventions or observations, comparisons and outcome measures. Study authors were contacted to clarify any study characteristics or omitted data. There was no blinding to study author, institution or journal throughout.

Quality assessment

Quality assessment of the studies that met the inclusion criteria was performed using the Quality Index (QI) checklist proposed by Downs and Black.31 The QI has been shown to be a valid and reliable tool for assessing the methodological quality of experimental designs (randomised controlled trials and quasi-randomised trials) and non-experimental designs (cross-sectional studies, case-control studies, case series and case studies).31 The QI consists of 27 items that are divided into five subscales: reporting (10 items), external validity (3 items), internal validity—bias (7 items), internal validity—confounding (6 items) and power (1 item). The item concerning a power calculation was modified to whether the study had performed a sample size calculation or not.32 The modified QI had a maximum score of 28 points and each item received 1 point if the criterion was met and 0 points if not (the distribution of principal confounders item could score 2 points).32 Zero points were given if the item was unable to be quantified. Each study was assigned a quality rating of ‘excellent’ (24–28 points), ‘good’ (19–23 points), ‘fair’ (14–18 points) or ‘poor’ (<14 points).32 Only studies with a quality rating of ‘fair’ (>14 points) or above were considered for inclusion in the review.32 33 The original PROSPERO protocol planned to use the Newcastle-Ottawa Scale; however, the authors decided the modified QI was a more appropriate quality assessment tool during pilot testing.

Data synthesis

Heterogeneity was assessed subjectively based on information about the participants, interventions and outcome measurements of each study. Due to heterogeneity across studies, data were presented in tables and synthesised qualitatively using the Synthesis Without Meta-analysis guideline.34

Results

The initial search identified 130 studies for screening (after duplicates were removed). Studies were excluded if they were commentary or review studies, involved rowing groups aged <18 years or did not include biomechanical outcome measures. The full texts of 45 studies were assessed for eligibility (figure 1); 7 studies were excluded based on outcome measures, 3 studies were not available in the English language, 10 studies were excluded based on study design and 2 studies were excluded based on participants. Therefore, 23 studies were included in the quality assessment.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-analyses flow diagram of search strategy and study selection process.

Quality assessment

The 23 studies were assessed using the modified QI checklist by two reviewers (FJN and FW). The initial QI scores achieved a kappa value of 0.7. After email correspondence and a meeting to discuss the disagreements, a kappa value of 1.0 (perfect agreement) was achieved. Twenty studies had a quality rating of ‘fair’ (14–18 points) and two studies had a quality rating of ‘good’ (19–23 points) (table 1). One study had a quality rating of ‘poor’ (<14 points) and thus was excluded from the review. Therefore, 22 studies were included (figure 1). The QI score of the 22 studies had a mean of 16.7 points (range: 15–21) of a maximum of 28 possible points (table 1). Across the 22 studies, the highest subscale scores achieved were reporting and internal validity—bias. The lowest subscale scores were external validity, internal validity—confounding and power (sample size calculation). Only two studies provided a sample size calculation.13 35

Table 1

Summary of modified QI checklist scores (n=22)

Characteristics of included studies

Thirteen studies were a cross-sectional design13 15 16 19 21 22 25–27 36–39 and nine studies were a case-control design.12 14 28 29 35 40–43 The studies involved a total of 429 participants (308 males and 121 females), which included novice rowers,13 38 club rowers,12 13 21 university rowers22 27 29 36 37 40 43 and elite rowers.13–16 19 25 26 28 35 41 LBP was defined using the Oswestry Disability Index,12 37 Fear Avoidance Behaviour Questionnaire37 or a specific definition28 29 35 40 41 such as ‘a history of LBP in the past year which required them to miss rowing for a period of at least 1 week’,28 ‘LBP which had required non-surgical intervention and had resulted in time off training’41 or ‘a pain or ache in the lumbosacral region with or without radiation to the buttocks’.35 Out of the 429 participants. 9 had current LBP,35 41 92 had a history of LBP12 16 27–29 35 37 40 41 43 and 328 were healthy (no LBP).13 16 29 37 38 Two studies compared healthy rowers (no LBP) with rowers with current LBP,35 41 8 studies compared healthy rowers with rowers with a history of LBP12 28 29 35 37 40 41 43 and 14 studies were on healthy rowers only.13–16 19 21 22 25–27 36 38 39 42 One study out of the 22 studies assessed biomechanical outcome measures on the water in a rowing boat15 while the remaining studies assessed outcome measures using laboratory-based equipment (eg, rowing ergometer, electromyography (EMG) and so on).12–14 16 19 21 22 25–29 35–43 Table 2 provides a summary of the characteristics of included studies.

Table 2

Characteristics of included studies (n=22)

Kinematic studies

Out of the 22 included studies, 13 investigated kinematics during rowing,13 15 16 19 21 22 25–29 35 38 which was primarily on the rowing ergometer13 15 16 19 22 25–29 38 (table 3). Seven studies measured rowing kinematics using electromagnetic motion measuring devices such as the Flock of Birds13 16 22 25–27 or 3-Space Fastrak.21 Three studies used motion capture systems for two-dimensional analysis36 38 or three-dimensional analysis using the Motion Analysis Corp system.39 Five studies used range of motion (ROM) tests or devices such as the goniometer12 15 40 and inclinometer.19 40 Seven studies incorporated a load cell in order to provide additional kinematic data.13 16 22 25 27 28 38

Table 3

Summary of findings for kinematic studies (n=13)

Across studies it was reported that kinematics were influenced by fatigue13 15 16 19 26 27 which resulted in an increase in both sagittal15 16 26 27 and frontal plane motion19 of the lumbar spine as rowing duration and intensity increased. It was consistently noted that the increase in lumbar spine motion was accompanied by a decrease in hip ROM and less anterior pelvic rotation at the catch26 27 35 and an increase in hip ROM and posterior pelvic rotation at the finish position.26 35 One study compared the ergometer with rowing in a boat and reported that fatigue was associated with an increase in sagittal motion of the spine which was significantly greater on the ergometer.15

Kinematics of ‘healthy’ rowers (no LBP)

Healthy elite rowers have distinct kinematics during the rowing stroke, particularly at the catch and finish phase.13 25 26 35 At the catch, rowers hold neutral or anterior pelvic rotation (figure 2A).25 27 35 At the finish, rowers hold a flat low back spinal position25 26 and have full knee extension.13 16 In addition, healthy elite rowers have greater hip ROM.13 35 A 2-year study involving seven senior national team rowers by McGregor et al 25 found changes to kinematics over time as a result of a trunk strengthening programme performed twice per week, in combination with normal rowing training. Peak force at the handle increased by 40–80 N (p<0.01) and stroke length increased by 15–19 cm (p<0.0001). Kinematic variables also improved at the catch, with increased anterior pelvic rotation (p<0.001) and rowers may achieve anterior pelvic rotation earlier during the recovery phase of the stroke (p=0.09).

Figure 2

The catch phase of the stroke in a rower with neutral or anterior pelvic rotation (A) and the catch phase of the stroke in a rower with posterior pelvic rotation (B).

Kinematics of rowers with LBP

Studies that specifically examined rowers with LBP reported conflicting results regarding the influence of LBP on kinematics; some demonstrated that those with current or a history of LBP have greater ROM through their lumbar spine and less hip ROM (at the catch)35 and others found no difference between groups.28 29 McGregor et al 35 found rowers with LBP had significantly less ROM at the L5/S1 level (in the catch position: 2.8°±5.5° in the LBP group vs 4.8°±1.2° in the history of LBP group vs 7.5°±1.3° in the healthy group; p<0.05) and at the L1/L2 level (p<0.01)35 (figure 2B). At the finish, rowers with LBP were in relatively greater posterior pelvic rotation (p<0.05) with relatively greater hip and trunk extension than the healthy group (p<0.05).

Muscle activity studies

Nine studies primarily investigated muscle activity (eg, via EMG, ultrasound imaging and so on)12 14 35–37 39 40 42 43 (table 4). Five studies measured muscle activity using EMG,12 14 36 39 43 one study used MRI41 and one study used ultrasound imaging.37 Four studies used isokinetic tests (eg, Cybex Norm Isokinetic Testing System)14 42 and isometric tests (eg, Back Analysis System)14 40 42 43 to measure muscle activity.

Table 4

Summary of findings for muscle activity studies (n=9)

Muscle activity of ‘healthy’ rowers (no LBP)

Studies demonstrated that muscle activity in healthy rowers is dominated by the extensor group of the trunk,12 39 42 with relatively less trunk flexor activity12 39 and little co-activation of the trunk flexors and extensors.39 Flexor activity was notable in the transition from the late drive to recovery phase.12 39 Asymmetries in trunk muscle function and cross-sectional area were common and of no significance to performance or history of LBP.14 37 Gill et al 37 found there were no significant differences in relative thickness of the transversus abdominis, internal oblique or external oblique based on gender (p=0.460, 95% CI=0.915 to 0.538, respectively), experience in rowing (p=0.154, 95% CI=0.622 to 0.743, respectively) or history of LBP (p=0.075, 95% CI=0.602 to 0.982, respectively). Only one study36 compared fixed and dynamic ergometers and found no significant difference in trunk muscle activity. No studies examined trunk muscle function in a boat.

Muscle activity of rowers with LBP

Martinez-Valdes et al 12 investigated the erector spinae (ES) activity of 18 competitive rowers (8 rowers with a history of LBP and 10 healthy rowers) using high-density surface electromyography (HDEMG). Martinez-Valdes et al 12 found that the magnitude of activation and distribution of ES activity were altered in rowers with a recent history of LBP (p<0.001). Rowers with a LBP history reduced ES recruitment to a smaller portion of the muscle group sited caudally compared with healthy rowers who extended recruitment out further into the muscle group. McGregor et al 41 investigated the cross-sectional area of the muscles acting directly on the lumbar spine (the ES, multifidus (MF) and iliopsoas (IP)) using an MRI scanner during simulated rowing in 22 elite oarsmen (5 rowers with current LBP, 13 with a history of LBP and 4 healthy rowers). McGregor et al 41 found that the cross-sectional area of the ES, MF and IP was significantly greater in those with current LBP or a history of LBP compared with healthy rowers.

Discussion

The aim of this study was to systematically review the current literature in order to define the kinematics and muscle activity of healthy rowers (no LBP) and to compare healthy rowers with rowers with current or a history of LBP. The main findings of this review suggest that healthy rowers have distinct kinematics—neutral or anterior pelvic rotation at the catch,25 27 35 flat low back spinal position at the finish,25 26 full knee extension at the finish13 16 and greater hip ROM13 35; and muscle activity—trunk extensor dominant12 39 42 with relatively less trunk flexor activity.12 39 In contrast, rowers with LBP exhibit relatively greater posterior pelvic rotation at the catch and finish,35 greater extension throughout the hips at the finish35 and muscle activity in the ES muscles may be inefficient.12 41 In both groups, fatigue can result in greater lumbar spine flexion at the catch.13 15 16 19 26 27

The number of participants in the studies with current LBP (9 participants)35 41 or a history of LBP (92 out of 429 participants)12 16 27–29 35 37 40 41 43 was low and this should be considered when interpreting the findings. This may reflect the ethical difficulties associated with asking a rower to participate in an activity which was associated with onset of LBP and one that may aggravate pain. Most studies focused on characterising ‘normal’ LP biomechanics during rowing and under conditions which induced fatigue. Understanding ‘normal’ kinematics and muscle activity provides a baseline for comparison, so even in the absence of studies examining LBP it was important to characterise healthy biomechanics.44 The findings may not be generalisable to all populations as most of the studies were on elite rowers13 15 16 19 25–28 35 39 41 42 and the majority of participants were male (317 out of 429 participants). The definition of LBP varied across studies12 28 29 35 37 40 41 and a more consistent definition and method for recording LBP is required. The majority of rowing related injuries (including LBP) are overuse injuries, therefore methods for recording injuries should be focused on prevalence of pain, dysfunction and reduced performance (eg, the Oslo Sports Trauma Research Centre Overuse Injury Questionnaire).45

Fatigue was found to result in greater lumbar spine flexion at the catch in both healthy rowers and rowers with LBP.13 15 16 19 26 27 Cyclical loading into flexion has been found to be associated with increased risk of LBP, notably when combined with fatigue.46 This causes an increased inflammatory cytokine response (especially in spinal ligaments)47 and a reduced proprioceptive response by desensitisation of the mechanoreceptors.48 Greater lumbar flexion and spinal creep (increasing sagittal motion) is observed on the ergometer compared with the water,15 however, there is insufficient evidence to recommend one ergometer type (fixed vs dynamic) over the other to avoid LBP.36A number of recent studies have investigated force production on fixed versus dynamic ergometers49–51 and the primary biomechanical differences are the mass of the moving parts. On the fixed ergometer, both the mass of the rower and the seat are moving, whereas with most dynamic ergometers the mass of the flywheel and/or foot plate is moving. When the fixed ergometer is placed on slides, in order to convert it to dynamic, the mass of the ergometer is free to move.52 Increased peak force/power production at the handle in fixed ergometers has consistently been observed when compared with dynamic ergometers in investigations of rowing at identical exercise intensities between 70% and 100% of 2000 m race pace.49–51 This could be hypothesised to affect the overall loading of the rower including the forces acting on the lumbar spine. A recently published cross-sectional study53 prospectively monitored 153 Australian national team rowers over eight seasons and found a reduction in LBP prevalence when fixed ergometers were replaced by dynamic ergometers, however no biomechanical outcome measures were explored.

Trunk asymmetries do not appear to be associated with LBP in rowers14 37 making it difficult to recommend a programme to address these imbalances. However, training trunk muscle strength and endurance is likely to have a role in the prevention of LBP due to the spinal fatigue and increased posterior pelvic motion noticed during increased rowing loads and among rowers with a history of LBP.13 15 16 19 25–27 35 A focus should be placed on training the posterior trunk muscles across the entire spine and the anterior trunk muscles to absorb braking forces in the last part of the drive phase.12 25 39 42 It is important to note that rowers with LBP or a history of LBP displayed greater ES, MF and IP cross-sectional area than healthy rowers when holding rowing-specific positions and examined with MRI.41 This may indicate a level of strength is important but that greater cross-sectional area41 and muscle recruitment12 may predispose to LBP or may be a result of LBP. This can only be determined by longitudinal studies following healthy rowers to the point of an acute LBP episode and through to recovery.

It is also impossible to know whether the kinematics observed in rowers with a history of LBP are a risk factor or develop due to having had an episode of pain. Regardless of this, aiming for the kinematic profiles seen commonly in healthy rowers seems to be a reasonable proposal (figure 2A). Ensuring rowers have large ranges of hip motion and ensuring they are taught the motion of anterior pelvic rotation and holding this position at the catch may influence risk of LBP. Training the rower to open the body from the hips during the drive phase of the stroke and ending in a flat low back position may also influence risk. A common observation of rowers with LBP was greater posterior pelvic rotation that resulted in lumbar spine flexion at the catch (figure 2B). It should be noted that these are observations from comparison of a LBP group with a healthy group so are assumptions which require confirmation with longitudinal studies or risk factors.

Only 1 out of the 22 studies assessed biomechanical outcome measures on the water in a single scull rowing boat15 while the remaining studies assessed outcome measures using laboratory-based equipment such as a rowing ergometer with a load cell,13 16 22 25–27 38 motion capture systems36 38 39 or EMG.12 14 36 39 43 There are numerous difficulties associated with biomechanical assessment in aquatic sports, however significant differences in spinal kinematics were found between the ergometer and rowing boat.15 Further studies are required in sweep rowing boats as significant differences were found between pelvic axial rotation and lateral bending of the spine in sweep rowing versus sculling on a modified ergometer (p<0.05).21 The methodology for identifying the catch and finish phase of the rowing ergometer stroke using a load cell varied across studies.13 16 22 25–27 38 The catch phase is where the initial loading of the spine, pelvis and hips occur, thus consistent methodologies are important to further understanding this loading pattern. Similar differences in motion capture set up and EMG systems were found. Surface EMG was used across the majority of studies12 14 36 38 39 43 but has been shown to have large variability and low reliability.54 One study used HDEMG which consists of grids of 10 electrodes and has been shown to increase reliability and sensitivity of amplitude estimates, and measures of spatial distribution of muscle activity.55 56

Future research

Future prospective studies evaluating rowers’ biomechanics as part of a longitudinal LBP risk assessment programme are warranted. Presentation of results should include reporting of magnitude of effect between groups. Studies should be adequately powered, study rowers of differing expertise and focus on water-based rowing.

Conclusions

This review included 20 studies of ‘fair’ quality and two of ‘good’ quality. Two studies compared healthy rowers (no LBP) with rowers with current LBP, 8 studies compared healthy rowers with rowers with a history of LBP and 14 studies were on healthy rowers only. The findings of this review suggest that healthy rowers have distinct kinematics which are neutral or anterior pelvic rotation at the catch, flat low back spinal position at the finish, full knee extension at the finish and greater hip ROM. In healthy rowers, muscle activity is dominated by the trunk extensors during the entire drive phase with flexor activity during the transition from the late drive to recovery phase. In contrast, rowers with current or a history of LBP exhibit posterior pelvic rotation at the catch and finish, greater extension throughout the hips at the finish and muscle activity in the ES muscles may be inefficient. In both groups, fatigue can result in greater lumbar spine flexion at the catch. Caution is advised when interpreting these findings as the number of studies involving rowers with LBP is very low and causality cannot be inferred.

What are the new findings?

  • A focus on achieving greater hip range of motion and neutral or anterior pelvic rotation at the catch may be an important strategy to reduce the risk of low back pain (LBP).

  • Fatigue as a result of increased rowing duration and/or intensity results in increased spinal flexion at the catch; this is more pronounced on the ergometer compared with rowing on water.

  • Training the trunk muscles to withhold forces over long durations and with increased load may be a consideration for the risk reduction of LBP in rowing.

  • There is insufficient evidence to recommend one ergometer type (fixed vs dynamic) over the other to avoid LBP.

  • Trunk asymmetries do not appear to be associated with LBP in rowers.

  • The extensor muscles dominate the rowing stroke and this should be considered in rehabilitation and training.

Acknowledgments

The authors would like to thank Toby Heaton, Clare Ardern, Alex Wolf, Caroline McManus and Mads Haubro for their contributions to this study.

References

Supplementary materials

  • Supplementary Data

    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.

Footnotes

  • Twitter @FrankNugent10, @janesthornton, @fionawilsonf

  • Contributors All authors were involved with the original design of the study. Data analysis was performed by FN, AMcG and FW. Drafting and approving of the manuscript was performed by FN, AV, AMcG, KW, JT and FW.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Request Permissions

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