Background

Rationale

Globally about 1.71 billion people suffer from a musculoskeletal condition [1]. Most adults in rehabilitation within Germany present with this diagnosis [2]. These patients have an increased risk of developing additional chronic diseases and mental health problems, which further increase the patients’ burden [3, 4]. Rehabilitation programs, aiming at reducing these patients' burden and their duration of sick leave, are often planned, and implemented by physiotherapists. Their and other therapeutic services account for 7% of the health expenditures in Germany (2021) and thus cause a financial burden for society [5].

Physiotherapeutic treatments have been used before clinical studies were conducted, but in the last decades, several interventions for specific diseases have been evaluated regarding their clinical effectiveness—although some intervention-disease-combinations remain yet unexplored. In recent years, besides their clinical effectiveness, the costs involved with a treatment have become additionally important to the stakeholders. The combined information supports their decision on whether a treatment should be implemented or de-implemented [6]. Full economic evaluations furnish this information on the anticipated costs along with the expected clinical outcomes of an intervention, enabling comparisons between various disease interventions and by this providing insight into possible cost savings. The costs presented in such analyses consider either healthcare costs alone or both healthcare costs and societal costs [6].

Several studies have already demonstrated that physiotherapists offer treatments worth the money [7, 8]. One review provides an overview of economic evaluations for treatments of neurological conditions offered by physiotherapists [9] and another review evaluates the cost-effectiveness of physical exercise, which is one of the treatment modalities offered by physiotherapists, for various health conditions [10]. However, an overview of existing economic evaluations of different physiotherapeutic treatment modalities for patients with musculoskeletal conditions is missing. Such an overview allows physiotherapists to easily identify relevant publications. Furthermore, it enables policymakers to easily identify relevant studies and researchers to plan future investigations.

Objectives

In this review, we therefore aim to:

  • provide an overview of existing full economic evaluations of interventions for patients with musculoskeletal conditions offered by physiotherapists.

  • shed light on the cost-effectiveness of physiotherapeutic interventions for specific musculoskeletal health conditions.

  • highlight for which health conditions further research is needed.

Methods

Protocol and Registration

A protocol for this review was published and additionally registered at Prospero (CRD42021276050) [11]. We followed our protocol but decided to report the findings of the identified trial and model-based economic evaluations separately. This allows us to fully account for and address the heterogeneous aspects of the two study types. Here we present our findings from the trial-based publications. We report our results following the PRISMA statement.

Eligibility Criteria

We included full economic evaluations of physiotherapeutic interventions for patients with musculoskeletal conditions. To specify our inclusion and exclusion criteria, we used the PICOS acronym (population, intervention, control group, outcome, study type).

Our population of interest suffered from a musculoskeletal condition. If the majority had another primary disease or if the participants had intellectual disabilities, we excluded the publication.

We included publications where physiotherapists provided one of the intervention/control group treatments alone. We excluded publications if the treatment of interest was offered by an interdisciplinary team, non-healthcare professionals, or mostly by a different profession. If the physiotherapeutic treatment of interest was combined with another treatment, this needed to be provided in a comparator group as well. Thus, an isolated incremental effect evaluation of a physiotherapeutic intervention needed to be possible. Economic evaluations of E-intervention were excluded.

We allowed any type of control group including wait-and-see, usual care, placebo, and alternative treatments, and excluded publications where no control group was considered.

Our outcome needed to be the result of a full health economic evaluation, including cost-effectiveness ratios, and cost-utility ratios.

Economic evaluations based on models and clinical trials were included during the screening process. We excluded Delete 'study types like' conference abstracts, reviews, books, and articles with no access and studies without results, e.g. protocols. In this publication, we additionally excluded model-based studies during the full-text screening, to reduce heterogeneity of included studies and allow a fair comparison of the study quality.

Finally, the economic evaluation needed to be published in Danish, English, or German.

Information Sources

We searched for relevant publications in the databases Medline (through PubMed), EconLit, and NHS-EED (can only be searched up to March 2015). The initial search was performed at the end of January 2022 and the final update was performed on the 8th of December 2023.

Search Strategy

We used the three main search terms ‘economic evaluation’, ‘physiotherapy’, and ‘orthopedic’ to develop a search matrix. For each of the main search terms we collected synonyms and combined them with an ‘OR’ in a search. Afterwards, we merged the three synonym searches with two ‘AND’. Utilizing asterisks reduced the number of search terms but allowed identifying relevant articles. Our search string in PubMed consequently was: ((((economic analy*) OR (cost analy*) OR (cost benefit*) OR (cost utility) OR (cost effectiveness) OR (economic evaluation))) AND (((orthopaedic rehabilitation) OR (physical rehabilitation) OR (exercise therap*) OR (conservative therap*) OR (conservative management) OR (conservative treatment) OR (exercise training) OR (physiotherap*))) AND (((muskuloskeletal) OR (chirur*) OR (orthop*) OR (osteoarthritis) OR (back pain)))). "Details on the PubMed search" are also available in Additional file 1: Table S1.

Selection Process

After removing duplicates, we applied a two-step study selection process. In the first step, we screened the title and abstracts for our inclusion and exclusion criteria. In the second step, we evaluated the full texts. Both steps were conducted by two independent reviewers at any time (LB, WF, BK). After each step, the involved authors compared their results and resolved disagreements via discussions or consulting a third reviewer (HHK) if needed.

Data Collection Process

The spreadsheet for data extraction was independently tested on three included publications by two authors (LB, BK). The authors discussed uncertainties, adjusted the labeling in the spreadsheet, and repeated the process. The authors agreed in the second round and LB proceeded with the data extraction. WF became involved in the data extraction process after she compared the data extraction results of three publications with LB, and no disagreement occurred.

Data Items

In total, we extracted the following items per publication: authors, year of publication, sample size, mean age, the proportion of female sex, health condition, location, type of economic analysis, cost perspective, economic effect measure, time horizon, study design, setting, and frequency, intensity, duration as well as the type of intervention, further control intervention, cost-effectiveness results, and finally the author’s conclusion. If the modalities of an intervention were not presented in a publication, but a reference was mentioned, we extracted the information from the cited publication. Further, in articles with several physiotherapeutic intervention or control groups, we ordered the interventions of interest as primary, secondary and so on. We prioritized basic interventions as primary, meaning interventions which could be most easily provided by physiotherapists—without further training.

Critical Appraisal of the Included Publications

We utilized the Consensus on Health Economic Criteria checklist for evaluating the quality of the included articles [12]. This tool consists of 19 questions which can be answered with ‘yes’ or ‘no’ each. Each ‘yes’ indicates good quality, whereas a ‘no’ indicates limited quality.

Two authors (LB and HHK) independently assessed the quality of three articles. After discussing uncertainties LB proceeded with the evaluation. WF got involved in the assessment after she and LB agreed on the quality of three, independently evaluated, articles.

Synthesis of Results

We provide a descriptive overview of the available literature. We categorized the economic evaluations based on the affected body parts. Afterwards, we grouped all cost-effectiveness comparisons, which are partly between two physiotherapeutic interventions, and partly between a physiotherapeutic and another intervention, of the included articles according to the four quadrants of a cost-effectiveness plane (I. more costly, more effective; II. more costly, less effective (dominated); III. less costly, less effective; IV. less costly, more effective (dominant)). The IV. quadrant is always assumed to be cost-effective, and the II. quadrant is always dominated, hence not cost-effective in the performed comparison. However, whether the interventions are cost-effective in the other two quadrants (I. and III.), depends on the stakeholder’s willingness to pay for an extra benefit in a health outcome. As an example, if a health system is not willing to pay the additional costs for the gain in health outcomes in the comparison group, our (primary) interventions in the III. quadrant would be cost-effective. However, if the healthcare system is willing to pay the additional costs to achieve the additional benefit in the health outcome, our (primary) interventions grouped in the I. quadrant would be cost-effective. Thus, to determine if interventions of the I. and III. quadrants are cost-effective, we would need to know the amount of money the healthcare system or society is willing to pay for a one-unit gain in the respected health outcome. However, this is beyond the scope of this review. For comparisons between two physiotherapeutic interventions, we defined one intervention as the primary intervention of interest. These primary physiotherapeutic interventions were more likely offered solely by physiotherapists and required least additional training/courses for the physiotherapists.

Results

Selection of Publications

Our search findings and the study selection process are visualized in the flowchart Fig. 1. In total, we identified 5141 eligible publications of which 83 met our inclusion criteria. A list of the publications excluded during the full-text screening process including an indication of the reason for exclusion is provided in Additional file 2: Table S2.

Fig. 1
figure 1

Flow chart of the study selection process

Full size image

Characteristics of the Economic Evaluations

Table 1 summarizes our included 83 articles which were based on 78 clinical trials [13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95]. Niemistö et al., Barker et al., Skargren et al. as well as Hurley et al. published two articles each building on one trial [17, 18, 51, 52, 72, 81, 82, 96]. Abbott et al. and Pinto et al. utilized data from the MOA RCT [13, 74]. All articles were published between 1994 and 2023 (Fig. 2). The majority of the 78 clinical trials were RCTs (n = 71) and 71 of the articles originated from the Western world. The UK (n = 23) and the Netherlands (n = 22) contributed more than 50% of the articles (Fig. 3). The time horizon of the economic evaluations varied between 5 days and 36 months. In Table 1 the included articles are grouped by their addressed body parts: spine (n = 39), upper limb (n = 8), lower limb (n = 30), and other conditions (n = 6). The most frequently investigated conditions were low back pain (n = 25) and hip and knee osteoarthritis (n = 16). In most included samples the mean age was between 45 and 65 and the distribution of female and male participants was between 40 to 60%. If a sample deviated from this, the mean age and female percentage were added in Table 1.

Table 1 Overview of the included trial-based economic evaluations
Full size table
Fig. 2
figure 2

Years of publication of the included publications

Full size image
Fig. 3
figure 3

Overview of the origin of included publications

Full size image

The outcomes of the included studies needed to involve a clinical outcome as well as the economic outcome costs. The latter was evaluated from a health-payer perspective in 50 cases and in 36 cases from a societal perspective (Table 1). As the numbers indicate, some studies presented costs for both perspectives. The most frequently utilized clinical outcomes were quality-adjusted life-years (QALYs), which were assessed via the EQ-5D (n = 39) and the SF6 or SF12 (n = 9). Disease-specific disability scores were assessed second most often (n = 9), e.g. via the Roland-Morris Disability Questionnaire. Finally, pain intensity was used in five publications. The remaining publications used individual outcome measures such as bothersomeness of symptoms, a stair climbing task, and blood pressure (Table 1).

In Table 2, we present details on the interventions and highlight our defined primary intervention. The articles involved 150 comparisons between physiotherapeutic interventions and comparators. Figures 4 and 5 display the results in terms of differences in costs and effects grouped qualitatively according to the four quadrants of the cost-effectiveness plane. Eighty-one comparisons involved one treatment provided by a physiotherapist versus another non-physiotherapeutic intervention (Fig. 4), while 69 of the comparisons were between two physiotherapeutic treatments (Fig. 5).

Table 2 Overview of interventions provided in the included trial-based economic evaluations
Full size table
Fig. 4
figure 4

Cost-effectiveness plane of any physiotherapeutic intervention versus another non-physiotherapeutic intervention. BH, behavior; ED, education; EX, exercises; H, home; IN, injection; MD, medical doctor; MT, manual therapy; MX, mixed; OT, others; PT, physiotherapy; SU, surgery; b, booster session; eb, evidence-based; g, group; i, individual; p, placebo; s, stratified care; u, usual; w, written; 2×, twice, *physiotherapeutic intervention, #no significant differences in the health outcome, ‘no significant differences in the costs

Full size image
Fig. 5
figure 5

Cost-effectiveness plane of a primary physiotherapeutic intervention vs another physiotherapeutic intervention. BH, behavior; ED, education; EX, exercises; H, home; IN, injection; MD, medical doctor; MT, manual therapy; MX, mixed; OT, others; PT, physiotherapy; SU, surgery; b, booster session; eb, evidence-based; g, group; i, individual; p, placebo; s, stratified care; u, usual; w, written; 2×, twice; *physiotherapeutic intervention, #no significant differences in the health outcome, ‘no significant differences in the costs

Full size image

Critical Appraisal of the Included Articles

More than two-thirds of the included 83 trials-based economic evaluations were of high quality with sum-yes-scores of 17 (n = 13), 18 (n = 16), or 19 (n = 31) on the Consensus on Health Economic Criteria checklist (Additional file 3: Table S3). The lowest score was 11 which was observed in two articles. Scores of 12, 13, and 14 were observed in two articles each. The scores 15, and 16 were reached by five and ten articles, respectively. The three items “Is a well-defined research question posed in answerable form?”, “Is the actual perspective chosen appropriate?”, and “Are all outcomes measured appropriately?” were evaluated with a ‘Yes’ for all the articles. The lowest sum of positive evaluation was observed for the item “Does the article indicate that there was no potential conflict of interest of study researchers and funders?” (n = 58). Limitations were also observed in 23 articles when it comes to the performance of incremental analyses of the costs and outcomes of alternatives. Fifty-nine publications met the criteria regarding sensitivity analyses to account for uncertain variable values. Nonetheless, the overall article quality was high.

Results of Syntheses Grouped by Body Parts

Spine

Of our 39 articles dealing with spine-related conditions four address the back in general [17, 18, 71, 84], 25 deal with the low back [14,15,16, 21, 26,27,28,29, 32, 38, 39, 43, 45, 46, 50, 55, 57, 59, 72, 73, 77, 83, 87, 93, 95], six with the neck [24, 62,63,64, 68, 94] and four include mixed patients [34, 67, 81, 82] (Tables 1, 2).

The four articles on general back conditions [17, 18, 71, 84] built on three clinical trials. The sessions varied between 1 and 36 over a time horizon of 1 to 24 weeks. Most interventions were conducted 1–2 times per week. The duration if indicated varied between 60 and 90 min. The intervention and control groups contained individual and group-based exercise therapy, manual therapy, and education as well as usual care (Table 2). Group exercise therapy was deemed cost-effective over usual care [71] and interpatient exchange group meetings were cost-effective over increasing the frequency of traditional therapy according to the authors [84]. The articles from Barker et al. considered three interventions provided by physiotherapists [17, 18]. The findings of all cost-effectiveness comparisons can be found in Figs. 4 and 5.

The 25 articles on low back pain were built on 24 clinical trials [14,15,16, 21, 26,27,28,29, 32, 38, 39, 43, 45, 46, 50, 55, 57, 59, 72, 73, 77, 83, 87, 93, 95]. Niemistö et al. published two articles based on the same clinical trial [72, 73]. Patients received between 1 and 46 sessions lasting between 20 and 120 min, spread out over a maximum of 30 weeks. Sessions were most often offered 1 to 3 times a week. The treatments in the intervention and control groups included mostly group and individual exercise therapy, individual physiotherapeutic treatments, and education. Additional manual therapy, McKenzie-based treatments, behavioral graded activity, behavior change techniques, a walking program, visits at the workplace, yoga, usual care, chiropractic treatment, and a massage chair were provided (Table 2). Fritz et al. compared individual physical therapy following evidence-based guidelines to physical therapy not following those guidelines and found that following evidence-based guidelines is cost-effective [38]. The articles on low back pain included a total of 67 comparisons of interventions of which 34 were between a physiotherapeutic intervention and another treatment (Figs. 4, 5).

Of the 6 articles on neck conditions all addressed neck pain [24, 62,63,64, 68, 94]. One to 24 sessions were offered. They lasted between 20 and 60 min and were spread over up to 12 weeks. Most treatments were offered 1–2 times per week. The intervention contained exercises, cognitive behavioral treatment, behavior-graded activity, manual therapy, individual therapy, and advice. The control group received exercises, usual care involving physiotherapeutic treatments, and written information (Table 2). The articles involve 12 treatment comparisons. Five highlighted better outcomes at higher costs for the physiotherapeutic intervention, and three were dominant. In the four dominated investigations three of the comparison interventions were offered by physiotherapists (Figs. 4, 5).

Two of the four articles with a mixed spine population originated from the same study. These studies did not differentiate between neck and back pain [34, 67, 81, 82]. Participants received up to 8 sessions. One article focused on physiotherapeutic care in comparison to traditional medical referral care, another compared the McKenzie treatment concept to the individual solution-finding approach, and the remaining two, from the same study, compared physiotherapeutic to chiropractic care (Table 2). Considering only the effectiveness, all articles favored the intervention offered by physiotherapists, although not significantly in the articles by Skargren et al. [81,82]. In the article of Manca et al. the psychotherapeutic treatments involved higher costs [67]; dominance of the physiotherapeutic treatment was found in the remaining articles [34, 81, 82] (Figs. 4, 5).

Upper Limb

The seven papers addressing the upper limb dealt with carpal tunnel syndrome [37], epicondylitis lateralis [61, 86], and shoulder complaints [23, 41, 53] involving a rotator cuff disorder [48]. The number of intervention and control group sessions varied between 3 and 18. Patients were treated 1–2 times a week for 20 to 60 min. All but three articles involved one intervention and one control group. The exceptions involved three different intervention groups. The provided interventions contained manual therapy, education, exercises, and physiotherapeutic treatments such as friction, ultrasound, and massage. The control groups received surgery, injections, braces, and usual care. One study also included a wait-and-see approach (Table 2). The articles involve 14 treatment comparisons. The physiotherapeutic treatment for carpal tunnel syndrome was dominant [37]. The remaining comparison of physiotherapeutic interventions, but one, may as well be cost-effective, depending on the willingness to pay (Figs. 4, 5).

Lower Limb

The included 30 articles built on 28 clinical trials dealing with the lower limb. Of these trials, four addressed the hip [40, 42, 56, 88], nineteen the knee [20, 22, 35, 47, 49, 51, 52, 54, 58, 60, 69, 70, 75, 78, 80, 85, 88, 92], and five a mixed population [13, 25, 31, 36, 66, 74]. Nineteen of the clinical trials dealt with osteoarthritis including the last therapy option of a joint replacement [13, 22, 25, 31, 36, 40, 47, 49, 56, 58, 60, 69, 70, 75, 76, 78, 80, 85, 88].

Among the hip-related clinical trials, three dealt with osteoarthritis and one addressed the femoroacetabular impingement syndrome [42]. Patients participated in 1 to 12 sessions. The duration of a session ranged from 30 to 60 min. The frequency varied between twice a day in the inpatient setting and one time in 2 weeks. The treatments offered contained individual as well as group exercises and education. The control groups received the usual care or an arthroscopy. Besides that, Fusco et al. compared the cost-effectiveness of a treatment with or without limiting hip motion [40] (Table 2). They conclude that not limiting hip motions after a hip replacement is dominant. Similarly, Griffin et al. and Juhakoski et al. highlight dominance of the physiotherapeutic treatment [42, 56]. However, Tan et al. found, that the physiotherapeutic intervention involved cost savings but was also less effective [89] (Figs. 4, 5).

The 19 knee complaint clinical trials bear 20 included articles. Two articles built on one clinical trial dealing with chronic knee pain [64, 65]. One article deals with patellofemoral pain [89], one with ACL tears [35], two with meniscal tears [90, 92], and two with knee pain [20, 54]. The remaining twelve articles investigated treatments for knee osteoarthritis. Patients participated in 1 to 18 sessions spread out mostly over 12 weeks (minimum 6 weeks, maximum 12 months). Inpatients were treated up to twice a day and all patients were treated at least once a month. The intervention for knee pain patients lasted the longest—24 months. Interventions contained mainly individual as well as group-based exercise therapy and education, but manual therapy was also offered in some clinical trials. The control groups received usual care, injections, surgery, education, and physiotherapeutic treatments (Table 2). In total the articles include 40 treatment comparisons (Figs. 4, 5).

Of the five trials on mixed lower limb conditions, we included five articles dealing with hip and knee osteoarthritis [13, 25, 31, 36, 74], and one article on ankle fractures [66]. The patients participated in 6 to 18 sessions spread out over 3 to 12 weeks. Inpatients again received treatments up to twice a day, but most interventions were offered 1–2 times a week. The sessions lasted between 50 and 75 min and contained individual physiotherapeutic treatments, manual therapy, and individual and group-based exercises. The control groups received information, usual care, and individual physiotherapeutic treatments (Table 2). Three articles favored a physiotherapeutic treatment. The remaining articles compared different physiotherapeutic treatments with each other. It was highlighted by the authors, that exercise therapy (individual and group) and education were more cost-effective than usual physiotherapeutic care [25], and that individual physiotherapy alone is cost-effective compared to individual physiotherapy combined with manual therapy [66] (Figs. 4, 5).

Other Conditions

Other conditions evaluated in cost-effectiveness studies included the complex regional pain syndrome [19], musculoskeletal problems [33], rheumatoid arthritis [91], and mixed patients [44, 65, 79] consisting of older adults, sedentary adults, and orthopedic outpatients. Patients participated in 5 to 60 sessions over a maximum of 24 months. The interventions contained individual pain exposure therapy, physiotherapeutic care, education, manual therapy, and group exercises. The control groups received usual care and exercises (Table 2). In each of the six articles only one cost-effectiveness comparison was performed (Figs. 4, 5).

Summary

The provided details on the delivered treatments varied widely between articles. A type of treatment was mentioned in all, however three articles only named “physiotherapy” as treatment without further explanations. Information on the frequency of the treatment was given in 67 articles; 16 provided no information. Similarly, the intensity was mentioned in 53 articles and missing in 30 articles. However, 28 of the 53 articles providing information on the intensity only mention that it was increased during the intervention or tailored to the patient without further explanation. Finally, the dose was mentioned in 55 and lacking in 28 articles (Table 2).

Due to the limited description of the interventions and the heterogeneity of treatment combinations in the intervention and control groups, summarizing comparisons based on interventions was largely restricted.

Nonetheless, in the 81 comparisons between a physiotherapeutic intervention and any other intervention we found the following insights [13, 14, 19, 20, 23, 26, 29, 30, 33,34,35, 37, 39, 41, 42, 45,46,47,48,49, 51,52,53, 55,56,57, 59, 61,62,63, 65, 67, 68, 72, 74, 76, 79,80,81,82, 84,85,86,87,88,89,90,91,92, 95,96,97]: 27 comparisons were performed between a physiotherapeutic intervention (without care provided by a medical doctor) and an intervention involving care from a medical doctor (Fig. 4) [13, 23, 26, 33, 34, 41, 46, 47, 49, 51, 52, 56, 62, 65, 72,73,74, 88, 91]. Of those about 50% (n = 13) were dominant [13, 34, 46, 47, 49, 51, 56, 62, 72, 74], two were dominated [52, 91], one involved lower costs and lower effects [88], and 11 had better outcomes but involved higher costs [13, 23, 26, 33, 41, 52, 74, 96]. Seven comparisons involved surgery in the control group [35, 37, 42, 85, 92], most of these (n = 6) were more expensive than the physiotherapeutic intervention [35, 37, 42, 85, 92]. The clinical outcome of the physiotherapeutic interventions was better in three of the comparisons [37, 42, 92]. Additionally, we found three comparisons between an intervention provided by a physiotherapist with one provided by a chiropractor [29, 81, 82]. Two originate from the same study and found physiotherapeutic care to be dominant, however, in the third comparison physiotherapeutic care was dominated. Injections were compared to solely other treatments in five comparisons stemming from three studies [30, 48, 53]. In three of these comparisons, physiotherapeutic care was more expensive but led also to better effects [30, 53]. In the remaining two comparisons (from the same study) physiotherapeutic care was cheaper but also less effective (Fig. 4).

Among the 69 comparisons between two physiotherapeutic interventions [13,14,15,16,17,18, 21, 22, 24,25,26,27,28, 31, 32, 38, 40, 43, 44, 48, 50, 52, 54, 57, 58, 60, 62,63,64, 66, 69,70,71, 74, 75, 77, 78, 80, 83, 86, 93, 94], we found 57 mentioning exercises as a treatment modality [13,14,15,16,17,18, 21, 22, 25,26,27,28, 31, 32, 48, 54, 57, 58, 60, 63, 64, 69,70,71, 74, 75, 78, 80, 83, 94, 93], however, it should be noted that if usual physiotherapy was provided this could also include exercises. Nonetheless, since it was not explicitly mentioned we do not consider “usual physiotherapy” as any intervention involving exercises. Of those comparisons 22 mentioned exercises for both comparators. Twenty-four of the comparisons involved manual therapy as part of an intervention, four in both groups [13, 16,17,18, 24, 26, 62,63,64, 66, 74, 75, 94]. Again, it should be noted that individual physiotherapy could have included manual therapies as well but this was not mentioned. Overall, there was no clear trend regarding the cost and the clinical outcomes observed. Having a closer look at comparisons between group-based and individual physiotherapeutic interventions, we found no clear trend regarding cost-effectiveness. This can partly be explained by the mixed treatments that were involved in these interventions as well, hindering clear comparisons.

Discussion

Key Results

Several good quality cost-effectiveness evaluations of physiotherapeutic interventions for patients with musculoskeletal conditions exist. Low back and knee conditions are frequently evaluated, however, for conditions addressing other joints none to few studies are available, here further research is needed. Unfortunately, the description of investigated interventions is often limited in detail and the combination of treatments varies widely. This restricted the ability to fairly compare different treatments.

In the comparisons between a physiotherapeutic intervention and those provided by other health professionals, a minor indication of physiotherapy was found to be cost-effective. Of the 42 comparisons between physiotherapeutic care and care provided by a chiropractor or a medical doctor involving surgeries and injections, we found that 18 were dominant and only four were dominated. For the 14 comparisons with higher costs and better effects, as well as for the 6 with a lower effect and lower costs the willingness to pay is crucial for deciding if the treatment should be considered cost-effective or not. The identification of which physiotherapeutic interventions are cost-effective was hindered by clear descriptions of the provided interventions and similar comparisons of treatment combinations.

To the best of our knowledge, this is the first review summarizing the findings of cost-effectiveness evaluations of physiotherapeutic interventions for musculoskeletal conditions. The earliest full-economic evaluation of a physiotherapeutic intervention for a musculoskeletal condition was published in 1994. However, since 2003 at least one new trial-based economic evaluation article has been released each year. This long inclusion period should be noticed when evaluating the relevance of a specific recommended treatment. The physiotherapeutic care and also the comparison treatments could have been enhanced during this time, meaning careful consideration of the (albeit often limited) intervention details is necessary before implementation.

Similarly, when utilizing and comparing the individual study findings of this review the underlying context of the study, the clinical outcome measures, and the findings of the Health Economic Criteria checklist should be considered. Different healthcare systems, as well as the culture of clinicians and patients could influence the cost-effectiveness. As an example, while in some countries like the Netherlands, Great Britain, and Sweden the patients may proceed directly to the physiotherapist, in other countries like Germany they have to consult a medical doctor first—this of course affects the overall cost-effectiveness and should be considered when studies from different countries are compared. Similarly, summarizing and comparing studies that utilized different health outcome measures is challenging. Of course, it is most important that the clinical outcomes are relevant to the patient, however, if different health outcome measures are utilized the comparison of the studies is limited, as the cost-effectiveness might change with a different health outcome. Finally, attention should be given to the evaluation of the Health Economic Criteria checklist of the individual studies/comparisons, which we do not indicate in the cost-effectiveness plane.

The availability of economic evaluation studies in our work mirrors the prevalence of the conditions, which is in line with previous reviews [9, 98]. We found that osteoarthritis was frequently studied. Similarly, a review of orthopedic surgery interventions found that joint arthroplasty, which is the last treatment option for patients with hip and knee osteoarthritis, was commonly investigated in related cost-effectiveness analyses [98]. Additionally, like our work, a review on physical exercise found most cost-effectiveness studies involved back conditions, osteoarthritis, and knee pain [10]. Single studies were also found for the musculoskeletal disorders of the shoulder and the neck [10].

That review [10] on physical exercise in the treatment of various diseases overlaps partly with our study. It includes 28 studies on musculoskeletal conditions of which we included 12 as well. However, the investigators focused on exercises as an intervention while we focused on treatments delivered by physiotherapists.

Interestingly, exercise was the most studied physiotherapeutic treatment in our work. Unfortunately, the treatments like exercise considered in our review were often combined with other treatments such as education, which often leaves the effectiveness of one specific treatment modality open. Furthermore, the treatments were rarely described sufficiently. For exercises e.g. they often lacked at least one dimension of frequency, intensity, time, and type of exercises e.g. aerobic vs. anaerobic or group vs individual therapy. Some articles even mentioned physiotherapy as provided treatment only. However, physiotherapy is not a treatment but a profession [99]. Consequently, several of the described interventions lack details, which leaves practitioners with the intention to implement cost-effective treatments with uncertainty and limits the ability to compare between different physiotherapeutic treatments.

The lack of provided details of the provided physiotherapeutic intervention is present in several studies and in clinical practice. Some initiatives aim to improve the documentation of provided treatments [100, 101]. However, they often apply to one specific treatment only such as exercises or the McKenzie treatment method [102]. Therefore, their applicability to other physiotherapeutic treatments is limited and therapists without special training sometimes do not understand the documentation. Hence, a specific but detailed documentation system for the provision of physiotherapeutic treatments is yet to the best of our knowledge missing.

Interestingly the provided treatments were similar across different body parts. This might suggest that the ideal treatment for a painful joint does not depend on the location. However, studies with more detailed treatment descriptions and with mixed population groups need to prove this hypothesis.

Future Research

The findings of our study indicate that there is a lack of economic evaluations for musculoskeletal conditions affecting joints other than the back and knee. Furthermore, conditions other than osteoarthritis such as fractures were rarely investigated and need further attention. Finally, systematic reviews of economic evaluations for physiotherapeutic treatments of back and knee complaints may be indicated, if not already covered by available reviews [103,104,105,106]. Additionally, a systematic review of the cost-effectiveness of preventive interventions could be interesting, since our focus was on studies dealing with patients with musculoskeletal conditions, thus excluding primary preventions.

Strength and Limitations

The major limitation of our study relates to the investigated interventions. First, defining the inclusion and exclusion criteria for an intervention was challenging. We aimed to include physiotherapeutic interventions, however, there is a large body of treatments such as tai chi, yoga, behavior change techniques, etc. which can be offered by physiotherapists but require further training. This training is open to non-health professionals as well. Some treatments e.g. behavior change techniques can also be offered by other health professionals who may be better qualified and thus more likely to provide a specific treatment. Furthermore, we decided to take a rather narrow approach regarding the inclusion criteria to ensure that the intervention was offered by physiotherapists. Second, we excluded conference abstracts, reviews, and model-based studies, which might involve the exclusion of relevant publications. Finally, our systematic search could have been broadened by additional search terms and the involvement of an additional database. This could have identified additional relevant publications since two of the included publications were only identified through the recommendation of experts in the field. The major strength of this overview is its transparency. Besides publishing a protocol and providing details on the conduct of our study and using the PRISMA statement, we also provide a list of the excluded full-text publications. The qualitative completeness of this review is underpinned by the quality evaluation of the individual articles and the provision of a cost-effectiveness plane. Consequently, we provide insights on the availability and quality of available articles and in this way highlight knowledge gaps in the literature.

Conclusions

Several high-quality trial-based economic evaluations of physiotherapeutic interventions for patients with musculoskeletal disorders exist and demonstrate cost-effectiveness. However, most articles address low back and knee conditions, and evaluations concerning other joints are limited. Finally, the documentation of provided interventions needs improvement to enable clinicians and stakeholders to fairly compare and finally to implement cost-effective treatments.