During the past decades, advances in health care in treatment of various diseases have led to increasing healthcare costs. As the available resources are limited, healthcare interventions should lead to maximal health benefit with the resources available. To achieve fair resource allocation, only methods proven cost-effective should be adopted for routine use.
Ageing populations place increasing demands on health care. To control healthcare costs, we are faced with a situation where some of the preventive and rehabilitative measures of healthcare providers may eventually have to become the patients’ responsibility. One such measure to advance patient recovery could be exercise, if proven cost-effective.
Elderly people hospitalized for acute conditions can deteriorate rapidly. Not always because of the condition itself but because of bed rest. Multiprofessional interventions comprising physical training in the acutely hospitalized elderly, according to a recent literature review, reduced hospital stays and need for care in another healthcare facility (Reference de Morton, Keating and Jeffs16). The effectiveness of mere exercise, however, could not be established, possibly in part because too few studies examined exercise alone (Reference de Morton, Keating and Jeffs16).
According to a Cochrane review on exercise in the treatment of nonspecific low-back pain, exercise therapy can be slightly effective at decreasing pain and improving function in adults with chronic low back pain (Reference Hayden, van Tulder and Malmivaara26). Furthermore, there was some evidence that a graded activity program can improve absenteeism outcomes in back pain patients, but evidence for other types of exercise was unclear (Reference Hayden, van Tulder and Malmivaara26). In coronary heart disease, exercise-based cardiac rehabilitation reduced cardiac deaths (Reference Jolliffe, Rees and Taylor31), and in heart failure patients, it was found to improve exercise capacity and quality of life (Reference Rees, Taylor and Singh51).
In oncology, advances in treatment have prolonged life expectancy. With increasing life expectancy, the side-effects of treatment like osteoporosis and psychosocial impairment have, however, also become more pronounced. In alleviating these side-effects, exercise interventions may help and, according to systematic literature reviews, are effective (Reference Knols, Aaronson and Uebelhart33;Reference McNeely, Campbell and Rowe41;Reference Schmitz, Holtzman and Courneya56). McNeely et al. (Reference McNeely, Campbell and Rowe41) for example, reported that, in breast cancer patients, exercise improves health-related quality of life (HRQoL) and physical capacity and reduces fatigue. Furthermore, physical activity improves cardiorespiratory capacity and well-being in cancer patients (Reference Schmitz, Holtzman and Courneya56). Knols et al. (Reference Knols, Aaronson and Uebelhart33) reported similar results, but pointed out that the studies concerned with exercise interventions in cancer patients were of moderate quality only, and for confirmation, larger, randomized controlled studies are necessary.
Exercise interventions have also been studied in several other diseases. For instance, in patients with type 2 diabetes, exercise significantly improves glycemic control and reduces visceral adipose tissue and plasma triglycerides (Reference Thomas, Elliott and Naughton62). In obesity, exercise is associated with weight loss and it improves cardiovascular disease risk factors even if no weight is lost (Reference Shaw, Gennat, O'Rourke and Del Mar59). In postmenopausal women, exercise increases the bone mass density of the spine and thus is effective in preventing and treating osteoporosis (Reference Bonaiuti, Shea and Iovine5).
Although exercise interventions have been shown to be effective in several studies, information on their cost-effectiveness in disease treatment is sparse. In a review of cost-effectiveness of health care interventions aimed at improving physical activity, Hagberg and Lindholm (Reference Hagberg and Lindholm21) identified 26 studies published before year 2005. The majority of them investigated exercise in primary prevention. Only ten studies were concerned with exercise in the treatment of people already ill, for example with cardiac diseases, arthrosis, or diabetes. In nine of those ten studies, exercise was judged cost-effective. One study was inconclusive as to cost-effectiveness.
Optimal resource allocation should lead to maximal health benefit in the society, but much uncertainty exists in decision making in health care. The adoption of exercise interventions in health care should be based on verified cost-effectiveness, as is the case with other treatment modalities. As studies reporting on cost-effectiveness have become more common recently, it is reasonable to update, in a systematic manner, what is known about the cost-effectiveness of exercise interventions in the treatment of various medical conditions.
METHODS
In considering the evidence of cost-effectiveness of exercise interventions in the treatment of medical conditions, we focused on controlled studies that reported the effect of exercise on costs or health care utilization. Literature searches were performed using the Medline, Centre for Research and Dissemination, and Cochrane Library electronic databases to June 2008. In addition, some articles were identified by scanning reference lists of the included articles. Finally, we also compared the result of our search with the listing of cost-effectiveness ratios published in the Cost Effectiveness Analysis Registry (https://research.tufts-nemc.org/cear/default.aspx) (14). The detailed search strategy is presented in Table 1.
Table 1. Search Strategies for Identifying Studies Concerned with Cost-Effectiveness of Exercise Interventions in the Treatment of Various Diseases
Screening of the identified articles, based on their abstracts, was undertaken independently by at least two of three authors (ER, RPR, PR) and the selection of relevant articles agreed upon in discussion. Full-text articles obtained for closer inspection were evaluated independently by two authors, who then reached a consensus on which articles should be included in the review. Included were articles that described, in a scientifically valid manner, controlled studies reporting on exercise intervention in the treatment of established medical conditions. As the aim was to assess cost-effectiveness, only studies reporting costs or some measure of health care utilization were included. Studies examining exercise promotion were excluded as were also studies dealing with primary prevention.
The quality of the selected controlled studies was scored, on a scale from 0 to 11, as recommended by van Tulder et al. (Reference Tulder, Furlan and Bombardier73). Furthermore, all studies were judged against the criteria for economic analysis given by Drummond et al. (Reference Drummond, Sculpher and Torrance17). The criteria combine 10 main items, scored 1 (criterion met) or 0 (criterion not met), resulting in a maximum score of 10.
The information given in the tables and in the appendix was gathered independently by one of the authors and its accuracy checked by another.
RESULTS
Selected Publications
The literature search identified 874 publications. Another forty studies were identified by scanning reference lists of the included articles and consulting experts in the field of economic analysis. Furthermore, one study was identified when results of the search were compared with the listing of cost-effectiveness ratios published in the Cost Effectiveness Analysis Registry (14). One hundred fifty-one articles were retrieved for closer inspection and of them sixty-five (representing sixty-one separate studies) were deemed to fulfill the inclusion criteria (Figure 1). Excluded were uncontrolled studies, studies with no economic information, or studies in which the exercise intervention was the same in all studied groups preventing the appraisal of the cost-effectiveness of exercise itself. Furthermore, primary prevention studies, reviews, letters and editorials were excluded.
Figure 1. Flow chart showing the various steps of study selection.
Study Classification
Of the included studies, twenty-eight (represented in 31 references) dealt with musculoskeletal disorders (Reference Beaupre, Lier and Davies3;Reference Brox, Staff and Ljunggren7;Reference Cochrane, Davey and Matthes Edwards13;Reference Coupe, Veenhof and van Tulder15;Reference Geraets, Goossens and de Bruijn19;Reference Gordon, Scuffham and Battistutta20;Reference Heymans, Vet and Bongers27–Reference Johnson, Jones and Wiles30;Reference Korthals-de Bos, Hoving and van Tulder34;Reference Lewis, Hewitt and Billington37–Reference McCarthy, Mills and Pullen39;Reference Mitchell and Carmen43;Reference Moffett, Torgerson and Bell-Syer44;Reference Niemistö, Lahtinen-Suopanki and Rissanen46;Reference Niemistö, Rissanen and Sarna47;Reference Patrick, Ramsey and Spencer49;Reference Richardson, Hawkins and McCarthy53;Reference Roush, Sevier and Wilson55;Reference Sevick, Bradham and Muender58;Reference Soegaard, Christensen and Lauersen60;Reference Søgaard, Bünger and Laurberg61;Reference Thomas, Miller and Doherty63;Reference Timm64;Reference Torstensen, Ljunggren and Meen66;69;Reference Van Der Roer, van Tulder and van Mechelen71;Reference Williamson, Wyatt and Yein75;Reference Wright, Lloyd Davies and Williams76), fifteen with cardiology (Reference Arthur, Daniels and McKelvie1;Reference Blumenthal, Babyak and Wei4;Reference Briffa, Eckermann and Griffiths6;Reference Carlson, Johnson and Franklin9;Reference Georgiou, Chen and Appadoo18;Reference Hagerman, Tyni-Lenne and Gordon22–Reference Harada, Kawakubo and Lee25;Reference Jolly, Taylor and Lip32;Reference Miller, Cress and Johnson42;Reference Nieuwland, Berkhuysen and Veldhuisen48;Reference Reid, Dafoe and Morrin52;Reference Yu, Lau and Chau77;Reference Zhang and Sun78), four with rheumatic diseases (Reference Bakker, Hidding and Van Der Linden2;Reference Timm64;Reference Van Den Hout, de Jong and Munneke70;Reference van Tubergen, Boonen and Landewe72), four with pulmonary diseases (Reference Carrieri-Kohlman, Nguyen and Donesky-Cuenco11;Reference Clini, Foglio and Bianchi12;Reference Ries, Kaplan and Limberg54;Reference Troosters, Gosselink and Decramer68), three with urinary incontinence (Reference Ramsay, Ali and Hunter50;Reference Schnelle, Kapur and Alessi57;Reference Williams, Assassa and Gillies74), and two with vascular disorders (Reference Lee, Mehta and Ray35;Reference Treesak, Kasemsup and Treat-Jacobson67). In addition, there was one study each in the fields of oncology (Reference Gordon, Scuffham and Battistutta20), chronic fatigue (Reference McCrone, Ridsdale and Darbishire40), endocrinology (Reference Nguyen, Ackermann and Berke45), psychiatry (Reference Torres-Carbajo, Olivares and Merino65), and neurology (Reference Lemstra, Stewart and Olszynski36).
The main findings (patients, intervention, control intervention, outcomes, conclusions concerning cost-effectiveness) of the included studies are reported in Tables 2–5 and a more detailed description of the studies can be found in Appendix 1 (which can be viewed online at www.journals.cambridge.org/thc). Some of the studies were the subject of more than one study; in those cases, we combined the results of the separate articles in the tables.
Table 2. Patients, Interventions, and Outcomes in the Identified Studies Dealing with Musculoskeletal Disorders
BC, best care; CBT, cognitive behavioral therapy; CHDR, current health desirability rating; EQ-5D, = EuroQol health-related quality of life instrument; GET, graded exercise therapy; GP, general practitioner; HAQ = Health Assessment Questionnaire, HRQoL = Health-related quality of life, ICER = Incremental cost-effectiveness ratio; LOS, length of stay; N.S., statistically non-significant; PQOL, perceived quality of life scale; PT, physiotherapist; QALY, quality-adjusted life-year; QWB, quality of well-being scale; SF-12 Health Survey, SF-12 health-related quality of life instrument; UC, usual care; VAS, Visual Analogue Scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index; $CDN, Canadian dollar.
Table 3. Patients, Interventions, and Outcomes in the Identified Studies Dealing with Cardiologic Disorders
AMI, acute myocardial infarction; CABG, coronary artery bypass grafting; CAD, coronary artery disease; CRPP, cardiac rehabilitation and prevention program; ICU, intensive care unit; MP, modified protocol; QALY, quality-adjusted life-year; QOL, quality of life; PCI, percutaneous coronary intervention; SF-36, SF-36 health-related quality of life instrument.
Table 4. Patients, Interventions, and Outcomes in the Identified Studies Dealing with Rheumatologic Disorders
AUC, area under the curve; EQ-5D, EuroQol health-related quality of life instrument; IET, intensive exercise training; QALY, quality-adjusted life-year; RAPIT, long-term intensive exercise program; SF-6D, SF-6D health-related quality of life instrument; VAS, Visual Analogue Scale.
Table 5. Patients, Interventions, and Outcomes in the Identified Studies Dealing with Miscellaneous Disorders
AUD, Australian dollar; CBT, cognitive behavioral therapy; COPD, chronic obstructive pulmonary disease; DM, dyspnea self-management program; EFP, enhanced fitness program; GET, graded exercise therapy; HRQoL, health-related quality of life; PFMT, pelvic floor muscle therapy; PRP, pulmonary rehabilitation program; PTA, percutaneous transluminal angioplasty; QALY, quality-adjusted life-year; RCT, randomized controlled trial; SEP, supervised exercise program.
Classified by the country of origin, most studies (n = 16) came from the United Kingdom. There were thirteen studies from the United States, eleven from the Netherlands, five from Canada, three from Australia, and two studies from both Finland and Norway. There was one study each from Denmark, Sweden, Belgium, Italy, Spain, Germany, Hong Kong, China, and Japan.
Study Quality
The quality scores ranged from 3 to 9 (median, 6) for the studies on musculosceletal disorders, from 2 to 9 (median, 6) for the cardiovascular studies, from 5 to 8 (median, 5.5) for the rheumatology studies, and from 3 to 8 (median, 5) for the studies dealing with miscellaneous disorders.
The scores for the economic analyses ranged from 4 to 10 (median, 8) for the studies on musculosceletal disorders, from 4 to 10 (median, 6) for the cardiovascular studies, from 7 to 9 (median, 9) for the rheumatology studies, and from 3 to 9 (median, 5.5) for the studies dealing with miscellaneous disorders.
Evidence for Cost-Effectiveness of Exercise by Type of Medical Condition
Musculoskeletal Disorders. Twenty-eight studies investigated exercise in the treatment of musculoskeletal disorders (Table 2). Except for one study, all the studies were randomized controlled trials.
BACK PAIN
There were fourteen studies dealing with back pain patients, the majority with chronic back pain. Nine of them reported that exercise saved costs (Reference Carr, Klaber Moffett and Howarth10;Reference Johnson, Jones and Wiles30;Reference Lewis, Hewitt and Billington37;Reference Mitchell and Carmen43;Reference Moffett, Torgerson and Bell-Syer44;Reference Timm64;Reference Torstensen, Ljunggren and Meen66;69;Reference Wright, Lloyd Davies and Williams76), whereas in three the exercise intervention was found not to be cost-effective.(Reference Niemistö, Lahtinen-Suopanki and Rissanen46;Reference Niemistö, Rissanen and Sarna47;Reference Soegaard, Christensen and Lauersen60;Reference Søgaard, Bünger and Laurberg61;Reference Van Der Roer, van Tulder and van Mechelen71 In one study on low back pain patients, costs between the three studied modalities (exercise, bed rest, normal activities) did not differ but patients in the normal activity group had better recovery (Reference Malmivaara, Hakkinen and Aro38). Heymans et al. (Reference Heymans, Vet and Bongers27) compared high- and low intensity back schools to usual care. The low-intensity back school was the most effective and cost-effective alternative.
In four studies, exercise was compared to usual care and in one study to physician consultation. In three of them, exercise was reported to be more cost-effective than usual care (Reference Mitchell and Carmen43;Reference Moffett, Torgerson and Bell-Syer44;Reference Wright, Lloyd Davies and Williams76). In the study by Johnson et al. (Reference Johnson, Jones and Wiles30), the reduction in pain and disability produced by the intervention was small and nonsignificant. Wright et al. (Reference Wright, Lloyd Davies and Williams76) found a back program with exercise (consisting of exercise, manipulation, joint and tissue mobilization, and so on) to lead to earlier return to work and thus to be cost-saving compared to usual care. By contrast, manipulative treatment together with stabilizing exercises was less cost-effective than physician consultation alone regarding both healthcare use and work absenteeism (Reference Niemistö, Lahtinen-Suopanki and Rissanen46;Reference Niemistö, Rissanen and Sarna47).
Compared to self-exercise, medical exercise therapy and conventional physiotherapy were both reported to be clinically superior and cost-saving in chronic low back pain patients (Reference Torstensen, Ljunggren and Meen66).
In a large UK trial, best care in general practice was compared with best care combined with exercise, manipulation, or both (69). Spinal manipulation was found to be a cost-effective addition to best care and appeared to give better value for money than manipulation followed by exercise.
A group exercise program was more cost-effective than individual physiotherapy (Reference Carr, Klaber Moffett and Howarth10). Likewise, manual therapy together with spinal stabilization exercises was more cost-effective when given in groups as compared to individual treatment (Reference Lewis, Hewitt and Billington37).
OSTEOARTHRITIS
In osteoarthritis patients exercise interventions were found to be cost-effective in only three (Reference Cochrane, Davey and Matthes Edwards13;Reference McCarthy, Mills and Pullen39;Reference Richardson, Hawkins and McCarthy53;Reference Sevick, Bradham and Muender58) of the eight included studies. (References 39 and 53 represent the same study.) In the other five (Reference Beaupre, Lier and Davies3;Reference Coupe, Veenhof and van Tulder15;Reference Hopman-Rock and Westhoff28;Reference Patrick, Ramsey and Spencer49;Reference Williamson, Wyatt and Yein75, the effect of exercise was modest compared to its costs.
KNEE PAIN
In all three studies dealing with chronic knee pain, exercise appeared to be an effective intervention (Reference Hurley, Walsh and Mitchell29;Reference Roush, Sevier and Wilson55;Reference Thomas, Miller and Doherty63). However, only the studies by Roush et al. (Reference Roush, Sevier and Wilson55) and Hurley et al. (Reference Hurley, Walsh and Mitchell29) found that exercise is likely to be cost-effective compared to the control intervention.
SHOULDER COMPLAINTS
Graded exercise therapy was more effective than usual care in patients with chronic shoulder complaints. It also reduced direct healthcare costs, but total costs during the 1-year follow-up were higher due to higher costs of the intervention (Reference Geraets, Goossens and de Bruijn19).
In treatment-resistant rotator cuff disease, both a supervised exercise regimen and arthroscopic surgery were more effective than placebo soft laser treatment. The improvement in the surgery group was greater, but the differences between the two groups were not significant or clinically important, and the costs of the exercise regimen were lower (Reference Brox, Staff and Ljunggren7).
NECK PAIN
In patients with neck pain, manual therapy was more cost-effective than physiotherapy comprising of individualized exercise therapy or usual care (Reference Korthals-de Bos, Hoving and van Tulder34).
Cardiovascular Diseases. The second largest group of studies covered the use of exercise interventions in the treatment of patients with cardiovascular diseases (Table 3). The majority (eleven of fifteen) of them (Reference Arthur, Daniels and McKelvie1;Reference Blumenthal, Babyak and Wei4;Reference Briffa, Eckermann and Griffiths6;Reference Carlson, Johnson and Franklin9;Reference Hall, Wiseman and King23;Reference Hambrecht, Walther and Möbius-Winkler24;Reference Jolly, Taylor and Lip32;Reference Nieuwland, Berkhuysen and Veldhuisen48;Reference Reid, Dafoe and Morrin52;Reference Yu, Lau and Chau77;Reference Zhang and Sun78) dealt with coronary artery disease (CAD), two with chronic heart failure (18,22), one with essential hypertension (Reference Harada, Kawakubo and Lee25), and one with hypertension in dialysis patients (Reference Miller, Cress and Johnson42). Most of the studies (eleven of fifteen) were randomized controlled trials.
CORONARY ARTERY DISEASE
In CAD patients, three studies comparing an exercise intervention with usual care considered the exercise intervention to be cost-effective (Reference Arthur, Daniels and McKelvie1;Reference Briffa, Eckermann and Griffiths6;Reference Yu, Lau and Chau77). Furthermore, one study reported exercise to be more cost-effective than stent angioplasty (Reference Hambrecht, Walther and Möbius-Winkler24). One study comparing formal rehabilitation after myocardial infarction to early return to normal activities without rehabilitation, however, found the latter more cost-effective in low-risk patients (Reference Hall, Wiseman and King23). In patients exhibiting mental stress-induced ischemia, stress management was shown to be beneficial over exercise and was associated with lower medical costs (Reference Blumenthal, Babyak and Wei4). A high-frequency exercise program, when compared to a low-frequency program led to slightly better quality of life (QoL) at approximately double cost but no conclusion or results on cost-effectiveness were reported (Reference Nieuwland, Berkhuysen and Veldhuisen48).
One study reported that a modified, reduced cost cardiac rehabilitation program is more cost-effective than traditional rehabilitation (Reference Carlson, Johnson and Franklin9). The distribution of the exercise intervention over 12 months, as opposed to 3 months, did not affect cost-effectiveness (Reference Reid, Dafoe and Morrin52) and home-based rehabilitation did not differ in a significant manner from center-based rehabilitation regarding effectiveness or costs (Reference Jolly, Taylor and Lip32).
One study reported that a gradual increase in moving around and walking led to shorter hospital stay and lower hospital costs compared to absolute bed rest in myocardial infarction patients (Reference Zhang and Sun78). Absolute bed rest in myocardial infarction, however, is currently an obsolete treatment modality in western countries.
HEART FAILURE
In heart failure patients, exercise interventions appeared to be cost-effective compared to usual care (Reference Georgiou, Chen and Appadoo18;Reference Hagerman, Tyni-Lenne and Gordon22).
HYPERTENSION
In hypertensive patients, drug therapy was clearly less costly per mmHg reduction in systolic blood pressure than exercise therapy (Reference Harada, Kawakubo and Lee25). In hemodialysis patients with hypertension, a greater reduction in antihypertensive medication was seen in the exercise group (Reference Miller, Cress and Johnson42).
Rheumatic Diseases. In three of the four studies on rheumatology, exercise was considered cost-effective (Table 4). Both of the studies on ankylosing spondylitis showed positive cost-effectiveness results (Reference Bakker, Hidding and Van Der Linden2;Reference van Tubergen, Boonen and Landewe72), but in rheumatoid arthritis, patient costs and quality-adjusted life-years (QALYs) were in favor of usual care (Reference Van Den Hout, de Jong and Munneke70). Intensive exercise training was more effective than usual care in patients with arthritis admitted to hospital because of disease activity flare or for elective joint replacement (Reference Bulthuis, Mohammad and Braakman-Jansen8). The cost-effectiveness results, however, were contradictory depending on how QALYs were determined.
Miscellaneous Disorders. Results concerning the cost-effectiveness of exercise interventions in chronic obstructive pulmonary disease (COPD) patients were equivocal with two studies showing some positive effects (Reference Carrieri-Kohlman, Nguyen and Donesky-Cuenco11;Reference Troosters, Gosselink and Decramer68), whereas a third one (Reference Ries, Kaplan and Limberg54) found a significant advantage from exercise (Table 5).
In the treatment of urinary incontinence, results of the three included studies were either negative or inconclusive regarding cost-effectiveness (Reference Ramsay, Ali and Hunter50;Reference Schnelle, Kapur and Alessi57;Reference Williams, Assassa and Gillies74).
Positive cost-effectiveness results were also reported in two studies dealing with peripheral arterial disease (Reference Lee, Mehta and Ray35;Reference Treesak, Kasemsup and Treat-Jacobson67), one study with breast cancer patients (Reference Gordon, Scuffham and Battistutta20), one study in diabetes (Reference Nguyen, Ackermann and Berke45), and one study with schizophrenic patients (Reference Torres-Carbajo, Olivares and Merino65).
Overview of Outcomes
The studied exercise interventions in musculoskeletal disorders were deemed to be cost-effective in 54 percent (fifteen of twenty-eight) of the cases, in cardiology in 60 percent (nine of fifteen) of the cases, and in rheumatic diseases in 75 percent (three of four) of the cases. Furthermore, there was some evidence that exercise interventions might be cost-effective in the treatment of intermittent claudication (two of two studies), breast cancer patients (one of one), diabetes (one of one), and schizophrenia (one of one).
DISCUSSION
We conducted a systematic review on the evidence of cost-effectiveness of exercise interventions in various diseases and critically examined the study quality. Our findings provide a basis for decision makers when considering which exercise interventions should be adopted for routine use. Several studies have shown the efficacy of various types of exercise interventions; their cost-effectiveness, however, remains poorly documented. As healthcare resources are limited, it is important that only cost-effective interventions are used.
The number of included articles was higher in the present review than in a previous literature review on cost-effectiveness of exercise interventions by Hagberg and Lindholm (Reference Hagberg and Lindholm21). This may in part be because of our relatively loose inclusion criteria. In addition to studies reporting pure cost-effectiveness data, we also included studies reporting on healthcare service utilization. Furthermore, studies with varying kinds of exercise interventions and control interventions were included. Another explanation is the fact that the number of studies investigating exercise and reporting on cost-effectiveness or healthcare utilization data have increased since the review by Hagberg and Lindholm (Reference Hagberg and Lindholm21), which covered studies published before year 2005. Twenty-eight of the studies included in this review were published in year 2005 or later.
There was great heterogeneity in the study settings of the included studies, and in many of them, exercise was combined with other interventions, most often with education or advice. There was also great variance in what exercise was compared to, ranging from usual care to surgical procedures and educative measures. Also, several studies compared different exercise or physiotherapy interventions of varying intensity. As the populations and interventions studied varied greatly, it was not possible to combine the results to draw conclusions about the effectiveness of exercise interventions in specific patient populations. Furthermore, the fact that, in many cases exercise was combined with advice or education prevents solid conclusions about the effectiveness of exercise per se. Consequently, in the future randomized controlled trials studying pure exercise interventions are needed to define the effect and cost-effectiveness of mere physical training.
Due to the small number of studies dealing with specific conditions, and the fact that reporting was in some cases incomplete, it is difficult to draw conclusions about the effect of patient characteristics (e.g., age, gender, previous physical activity, severity of illness), or the characteristics of the exercise interventions (intensity and workload, adherence to physical training) to the outcomes regarding health and costs. In future studies, it would be important to identify those patients that are expected to benefit the most from exercise interventions.
The overall quality of the included studies varied widely and was, on average, only mediocre. There was no systematic difference regarding quality between the different disease groups. The economic quality of the studies judged against the criteria by Drummond et al. (Reference Drummond, Sculpher and Torrance17) tended to reach somewhat higher scores than the overall quality and, was judged, on average, to be fair to good in studies dealing with musculoskeletal disorders and rheumatology. There was also much variance in how the outcomes were reported, which precludes the use of meta-analysis for combining the results of different studies. Only a few studies reported outcomes as cost/QALY, which enables the comparison of different kinds of interventions in health care. In prospective studies, it would be useful to use the same instruments to measure the outcomes, which would allow pooling of the results of different studies together.
CONCLUSION
The number of studies assessing cost-effectiveness of exercise interventions in various diseases is still rather limited, and the results show large variation. The results suggest that some exercise interventions, however, can be cost-effective. In rheumatology, the percentage of studies reporting positive cost-effectiveness outcomes was the highest, but the number of studies was low. Most convincing evidence for cost-effectiveness was found for rehabilitation of cardiac patients and back-pain patients; however, even in these cases, the evidence was partly contradictory.
SUPPLEMENTARY MATERIAL
Appendix 1: www.journals.cambridge.org/thc
CONTACT INFORMATION
Eija Roine, MD (eija.roine@helsinki.fi), Resident Doctor, Department of Oncology, Helsinki University Central Hospital, P.O. Box 180, 00029 HUS, Helsinki, Finland
Risto Roine, MD, PhD (risto.p.roine@hus.fi) Adjunct Professor, Chief Physician, Administration/Research and Development, Helsinki and Uusimaa Hospital District, P.O. Box 705, 00029 HUS, Helsinki, Finland
Pirjo Räsänen, PhD (pirjo.rasanen@thl.fi), Senior Researcher, Finnish Office for Health Technology Assessment, Finnish Office for Health Technology Assessment, Ilkka Vuori, MD, PhD (ilkka.vuori@uta.fi), Senior Consultant, Harri Sintonen, PhD (harri.sintonen@helsinki.fi), Professor, Department of Public Health, University of Helsinki, Research Professor, Finnish Office for Health Technology Assessment, The National Institute for Health and Welfare, P.O. Box 30, 00271 Helsinki, Finland
Tiina Saarto, MD, PhD (tiina.saarto@hus.fi), Chief, Department of Oncology, Helsinki University Central Hospital, Department of Oncology, P.O. Box 180, 00029 HUS, Helsinki, Finland
