Physiotherapy interventions for pain management in haemophilia: A systematic review

Approximately 35%‐50% of people with haemophilia (PWH) report living with chronic musculoskeletal pain. Although exercise based rehabilitation is effective for pain in other arthritises, there are no published guidelines for management of chronic pain in PWH. This review aims to evaluate and appraise the current evidence of effectiveness of physiotherapy interventions on (a) pain intensity, (b) quality of life (QoL) and (c) function in PWH.


| INTRODUC TI ON
Haemophilia is an inherited bleeding disorder characterized by recurrent and spontaneous bleeding into joints and muscles and fatal bleeding in the untreated state. 1,2 People with haemophilia (PWH) experience transient episodes of acute pain from an early age from musculoskeletal bleeding episodes. Despite replacement therapy, some PWH continue to have bleeding into their joints and muscles, which can lead to debilitating arthritis with chronic and recurrent pain. 3 People with haemophilia over the age of 65 had no access to regular treatment until they were in adulthood, with those currently aged in their 40's having no access to effective treatment for the majority of their childhood. 4 Consequently, many PWH have chronically painful, multi-joint haemophilic arthritis, involving elbow, knee and ankle joints. [5][6][7] Between 35% and 50% of PWH report living with chronic musculoskeletal pain, 7-10 with 40% reporting their pain is poorly managed by their healthcare provider. 8 PWH living with pain report limitations in mobility and independence, increased anxiety, poor quality of life and frustration due to restrictions in activities of daily living. 7,11,12 A recent systematic review of management of multisite osteoarthritis (OA) found that exercise interventions may have moderate benefits on pain, function and quality of life. 13 More specifically, aerobic exercise has been shown to be effective for pain management and functional improvements in rheumatoid arthritis 14 and in OA when used with mind-body interventions. 15 However, although pain is a widespread problem in haemophilia, there are no published guidelines for the physiotherapy of management of chronic arthritic joint pain in this population.

| Objective
This review aims to evaluate and appraise the current evidence of the effects of physiotherapy interventions on (a) pain intensity, (b) quality of life and (c) function in PWH.

| Protocol and registration
The protocol was registered with the International Prospective Register of Systematic reviews (PROSPERO number: CRD42018116482).
Reporting is in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 16

| Eligibility criteria
Study design for inclusion was those described as randomized controlled trials and quasi-experimental studies including controlled studies, before and after and interrupted time studies, comparing to no intervention/routine care group or between group comparison of one treatment intervention against another.
Studies describing any physiotherapy/rehabilitation/physical therapy intervention that had pain intensity, functional outcomes and health related quality of life as outcome measures were included. Studies with participants of any age with a diagnosis of mild, moderate or severe haemophilia (A or B), and/or haemophilic arthritis were included. Those with participants with a diagnosis of an inhibitor (antibody to factor VIII or IX) and co-morbidities were not excluded. There was no restriction in country or care settings for studies.
Studies that investigated joint disease or pain as a result other inherited bleeding disorders such as von Willebrand disease were excluded. Only studies published in the English language were included. Iterative refinement of the search strategy was achieved after multiple practice searches using potential search terms and associated subject headings. The university version of OVID and EBSCO search platforms maps to subject headings by default. The search strategy was discussed in detail and endorsed by the University librarian (AE-J).

| Study selection
One reviewer (PML) independently carried out the search strategy on the listed databases. Results were saved, duplicates removed and then imported to the Rayyan platform, 17 enabling two reviewers (PML, DS) to independently review titles and abstracts whilst blinded from each other. Once each reviewer had completed their check, the abstracts were unblinded. We compared those which had been accepted, rejected and were undecided by both reviewers, and discrepancies between reviewers (n = 2) were discussed and a consensus reached.
Full texts of agreed abstracts were retrieved and evaluated independently (PML, DS) to determine eligibility for inclusion in the systematic review.

| Data collection process
A data extraction proforma was developed using the Cochrane Airways group template (https://airwa ys.cochr ane.org/data-colle ction). One reviewer (PML) extracted data studies, and a second reviewer (DS) checked extracted data for accuracy. One author was contacted for further information, and data were received. 18

| Data items
Information extracted from each trial included study design, participant information, interventions, comparison interventions, outcome measures (pre-and postintervention, follow-up if available), results including pain, function and quality of life.

| Risk of bias in individual studies
The Cochrane Risk of Bias assessment tool was used to assess included papers and was carried out independently by two authors (PML, DS). Criteria of unclear, low or high risk of bias were assigned against selection bias, performance bias, detection bias, attrition bias, reporting bias and any other identified bias.

| Methods of analysis
Cochrane collaboration software (RevMan version 5.3) 19 was used to collate and analyse study data.
Mean change from baseline to follow-up and standard deviation of mean difference (MD) was calculated for input into RevMan. Using a fixed effects model, mean differences ± 95% confidence interval (CI) per intervention were calculated. Studies were grouped into (a) physiotherapy intervention vs no intervention and (b) physiotherapy intervention A vs physiotherapy intervention B.
A narrative synthesis of the evidence was completed including the use of the GRADE approach in grading evidence quality. 20 The GRADE system uses eight criteria against which to assess the quality of evidence as either high, moderate, low or very low. They are

| Additional analysis
We were unable to undertake meta-analysis due to heterogeneity of the included studies.

| Study selection
The search strategy identified 417 citations. Following removal of duplicates, 309 remained. Review of abstracts resulted in removal of 290, with further 10 being removed after full text review ( Figure 2).

| Study characteristics
Nine studies were identified ( Table 1). The number of participants per study ranged from 9 22 to 40. 23 The number of participants across all studies was 235. Of these, 60 were children (aged 9-13) and 175 were adults (aged 26-58). Severity of haemophilia was not specified for 70 participants. Of the remaining 165, 93 were identified as having a diagnosis of severe haemophilia, 50 moderate, 17 mild and 5 mild/moderate. One hundred and forty-nine participants were on prophylaxis and 46 were on-demand. Treatment regime was not stated for 40 people. Following GRADE assessment, all nine studies were rated as low/very low for quality of evidence.

| Interventions
Study intervention periods ranged from 3 to 15 weeks. Four trials had a RCT design. 18,[23][24][25] Four studies compared one physiotherapy intervention against another: passive joint mobilizations and exercise vs manual therapy and exercises in adults with haemophilic ankle arthropathy 22 ; high intensity laser therapy (HILT) and exercise vs pulsed electromagnetic field and exercise in treatment of knee haemarthrosis in children 26 ; home exercise programme and self-monitoring vs home exercise alone for haemophilic in adults with knee and ankle arthropathy 27 ; and HILT and exercise vs placebo HILT and exercise in haemophilic arthropathy of the knee in children. 28 Three studies in adults compared two physiotherapy intervention arms against a control group; manual therapy and exercise against patient education and exercise in haemophilic ankle F I G U R E 2 Flow chart of trial identification and selection for inclusion in review [Colour figure can be viewed at wileyonlinelibrary.com] arthropathy, 24 with the same study design replicated for elbow arthropathy. 18 A third study investigated hydrotherapy against land-based exercise with a control group in haemophilic knee arthropathy. 23 Two studies in adults compared one physiotherapy intervention with a control group; patient education and home exercises verses control on elbow, knee and ankle haemophilic arthropathy 25 ; and fascial therapy vs control on knee and ankle haemophilic arthropathy. 29 One study performed the intervention 3 sessions per week for 3 weeks, 29 with another not stating how many sessions were performed over 4 weeks. 23 Two studies performed the intervention for 2 sessions per week for 6 weeks, 18,22 and one study encouraged participants to do exercises 10 times a day for 8 weeks. 27 Another performed 2 sessions per week over 12 weeks 24 with another doing 1 session every 2 weeks for 12 weeks. 18 The participants in two studies received 3 sessions per week for 12 weeks 26 and another one session every 2 weeks for 15 weeks. 25

| Risk of bias
All studies had an overall risk of bias (see Figure 3). Assessment of risk of bias found agreement between study authors was moderate (Cohen's K 0.51).

| Sequence generation
One study rated low risk as it described the use of a random number generation table for each participant. 27 Six studies were rated as unclear risk as sequence generation methods were not described. One paper rated high risk as participants were chosen for inclusion based on geographical location. 29

| Allocation concealment
Two studies had a low risk with both describing opaque envelopes being distributed by someone unrelated to the study. 18,25 Six studies were rated unclear due to lack of detail on methods of concealment.
One study rated high risk as participants were selected based on geography. 29

| Blinding
Blinding of the participants was not possible in any of the included studies, and none of personnel were blinded in any study. Five studies used blinded evaluators to assess outcomes 18,22,[24][25][26] and were rated low risk. Two studies rated unclear as they did not state if outcome assessment was blinded, 23,28 and two rated high risk as outcome assessment was completed by the same individuals delivering the intervention. 27,29

| Incomplete outcome data
Four studies rated as low risk of attrition bias because each stated that all participants completed the intervention. 18,22,25,27 Five rated unclear as although they did not report dropouts, they also did not explicitly state that all had completed the intervention. 18,24,26,28,29 One study was rated high risk as although the authors reported three dropouts, they did not specify from which group they came. 23

| Selective reporting
Three studies were rated high risk of selective reporting bias. One study failed to report on changes to bleeding frequency even though this was an inclusion criteria for the study. 26 Another describes an improvement in joint health with the Haemophilia Joint Health Score but include no data to support this 28 and another does not report on all of the elbow joints included in their study. 18 The six other studies had unclear risk of selective reporting bias.
No study was determined to have any source of other potential bias and therefore were rated as low.

| Data synthesis
Outcome measures for the nine trials are presented in Table 1.
Although there were multiple outcomes measured across the trials,

| PHYS I OTHER APY VS NO INTERVENTI ON
Five studies were included in this comparison ( Figure 4). 18,[23][24][25]29 TA B L E 1 Summary of included studies with GRADE assessment 15 wk study period Educational sessions every 2 wk for 60 min alongside home exercise programme: Control group advised to continue with the same daily professional and sporting routines that they had been following Education/HEP: Theory: anatomy/biomechanics elbow, knee and ankle joints, haematoma management, exercise theory, joint bleeds, synovitis and arthropathy, proprioception, physical activity and sport.    Only difference is the participants in the intervention arm could review their progress on their monitors, whereas the control arm group could not. Home exercise programme-guidance about strengthening knee extensors, static stretching for knee flexors and standing balance training. Advice on promotion of physical activity given by physio to improve knee function. Knee extension strength training, static stretches and balance training Advice on leading an active life and doing non-contact sports were recommended for improving physical activity. ** physio recommended the exercise most appropriate to the physical condition of each patient to be done 10 times per day Self-monitoring: Participants were equipped with display activity monitors and feedback system via internet and mobile phone. When participants accessed the server to data input-feedback results appeared with time in form of graphs and tables.

| Pain
All five trials included assessment of pain intensity using the visual analogue scale (VAS), with improvement in pain reported as a decrease in the VAS score. All were conducted on adults over 18 years of age, and apart from one study, 29 all were RCT's.

Elbow
There was no clear benefit on pain intensity when using manual

| Quality of life
Only one study 25

| Function
None of the studies measured function as a outcome of intervention.

| PHYS I OTHER APY INTERVENTI ON A VS PHYS I OTHER APY INTERVENTI ON B
Seven studies were included in this comparison ( Figure 5). 18

Pain (knee and ankle combined)
It is not clear if there any beneficial effect on knee and ankle pain of a self-monitoring home exercise programme compared to an exercise programme alone, MD 0.62 VAS (95% CI −0.37 to 1.61).

| Quality of Life
Only one study 22

| Function
Three studies included a measure of function as an outcome measure of intervention.

| D ISCUSS I ON
This review presents the current evidence of trials utilizing a va- Pain is an issue that many PWH state is one of their major concerns, 30 yet no study included in this review defined pain as a specific inclusion criteria or ascertained if the presence of pain was of concern to participants. This may indicate assumptions made on the presence of pain based only on having haemophilia and/or arthropathy. Across many of the studies, the small differences in pre-and postintervention pain (VAS) highlight only a small change after intervention. It is unclear if this is due to a low pain VAS report preintervention (ie they had less/minimal pain upon starting the intervention) or a lack of effect of the intervention.
Only two studies included quality of life assessment, 22,25 and three included an assessment of function. [26][27][28] The minimal evalua- intervention, thereby making it difficult to evaluate them in their practical effectiveness or to identify how they may be exerting their effects.
Education showed little positive effect when used in conjunction with exercise or manual therapy. It is not clear from the studies how the teaching curriculum was developed. Including health education without consideration for potential behaviour change models of action limits evaluation of how any education provision may be having its effect. 32 Previous systematic reviews share some similarities to this review. However, the quality of some of those reviews, as well as their results and recommendations, differs to those presented here.
Although described as a systematic review, an evaluation of exercise and sport in the treatment of PWH 33 lacked clear inclusion criteria as well as systematic analysis or comparison of data. In a narrative review of physical exercise, pain and musculoskeletal function in PWH Schäfer and colleagues 34 described the data for intervention effectiveness and concluded that exercise promoted a reduction in pain, improved ROM and strength in PWH. However, mean change from baseline together with confidence intervals was not reported or compared. They also reported low risk of bias in three studies that we assessed as having high risk, 22,23,27 although this may be due to different assessment tools being used.
A recent Cochrane Review on Exercise for haemophilia 35 was well conducted and included a broader range of outcomes in their analysis than this review. Similar to us, they noted major issues on study quality and stated that although exercise was likely safe, they urged caution with results as they stand.
Two further systematic reviews focussed on the treatment of chronic haemophilic ankle arthropathy 36 and physiotherapy in the treatment of haemophilic arthropathy. 37 The focus of the analysis in both was on the physiotherapy intervention itself rather than comparing the effects of those interventions on specific identified outcome measures. This makes it difficult to infer efficacy of any one specific intervention on measures such as pain and function. In contrast to our findings, the second review reported good study homogeneity, but it was not clear how this was evaluated.
Overall, the methodology and reporting quality of many of the included trials were poor. No study rated as high when being assessed for overall risk of bias. Many failed to provide details on randomization or participant allocation as well as appearing to omit some data in their overall results. High degree of trial heterogeneity was identified for both participants (severity of haemophilia, age range, location and number joints affected) as well as interventions (varied time frames for delivery, intervention components and outcome measures). Thus meta-analysis was not possible. Participant numbers were low for all included trials, and four of the trials were randomized pilot studies. 18,22,24,29 Although overall safety from physiotherapy interventions was No studies appear to have involved PWH in the trial design, nor evaluated any qualitative measure of participation in such trials. As a rare disease, many PWH can be considered experts not just in their condition, but also in potentially identifying what matters to them in respect of rehabilitation interventions.
A strength of this review is the process of using two blinded reviewers throughout the process. Unlike many of the previous similar systematic reviews, we analysed the data to produce confidence intervals (CI's) and mean difference (MD) figures, allowing a good visual representation of effectiveness. A limitation is that we were unable to proceed to complete a meta-analysis of the data from any of the included studies. This precludes any clear recommendations for the use of physiotherapy interventions in the management of pain in haemophilia.
Better design of trials is required and should include PWH in the process. Specific and defined inclusion criteria relating to haemophilia disease severity, as well as pain as a self-reported symptom, are needed to better assess efficacy of any interventions.
The current use of only VAS in measuring pain intensity requires further scrutiny. Pain as a multi-modal, personal, lived experience is poorly evaluated when measuring intensity alone. 38 Further trials need to focus on how interventions may be designed to improve the physical, social, psychological and functional ramifications of a lifelived with pain.

| CON CLUS ION
This systematic review highlights that there is currently an unclear demonstration of evidence for the use of physiotherapy interventions for pain management in people with haemophilia. LASER with exercise and hydrotherapy/land-based exercise appears to have some positive effect on knee pain in PWH. However, caution must be taken with this recommendation due to poor quality reporting and high risk of bias in both trials. It is not possible to make recommendations on any other physiotherapy intervention in the management of pain in haemophilia. Improved trial design and methodology will allow this emerging body of research to be effectively collated and compared, to further develop effective interventions for pain in haemophilia.

ACK N OWLED G EM ENTS
We wish to express our thanks to librarian Anna El-Jouzi for her help and guidance in the development of the search strategy.