A systematic review of nonrandomized controlled trials on the curative effects of aquatic exercise.
Faculty of Regional Environment
Science, Tokyo University of Agriculture,
1-1-1 Sakuragaoka, Setagaya-ku,
Tokyo 156-8502, Japan
Tel/fax +81 3 5477 2587
Email h1kamiok@nodai.ac.jp
Keywords
Article abstract
BACKGROUND:
The objectives of this review were to integrate the evidence of curative effects through aquatic exercise and assess the quality of studies based on a review of nonrandomized controlled trials (nRCTs).
METHODS:
Study design was a systematic review of nonrandomized controlled trials. Trials were eligible if they were nonrandomized clinical trials. Studies included one treatment group in which aquatic exercise was applied. We searched the following databases from 2000 up to July 20, 2009: MEDLINE via PubMed, CINAHL, and Ichushi-Web.
RESULTS:
Twenty-one trials met all inclusion criteria. Languages included were English (N = 9), Japanese (N = 11), and Korean (N = 1). Target diseases were knee and/or hip osteoarthritis, poliomyelitis, chronic kidney disease, discomforts of pregnancy, cardiovascular diseases, and rotator cuff tears. Many studies on nonspecific disease (healthy participants) were included. All studies reported significant effectiveness in at least one or more outcomes. However results of evaluations with the TREND and CLEAR-NPT checklists generally showed a remarkable lack of description in the studies. Furthermore, there was the problem of heterogeneity, and we were therefore not able to perform a meta-analysis.
CONCLUSION:
Because there was insufficient evidence on aquatic exercise due to poor methodological and reporting quality and heterogeneity of nRCTs, we were unable to offer any conclusions about the effects of this intervention. However, we were able to identify problems with current nRCTs of aquatic exercise, and propose a strategy of strengthening study quality, stressing the importance of study feasibility as a future research agenda objective.
Article content
Introduction
Over the years, aquatic exercise has been known as pool therapy, hydrotherapy, and sometimes in earlier literature, as balneotherapy.1 Exercise in warm water, usually termed hydrotherapy or aquatic therapy, is a popular treatment with a pain relief effect for many patients with painful neurologic or musculoskeletal conditions.2 The warmth and buoyancy of water may block nociception by acting on thermal receptors and mechanoreceptors, thus influencing spinal segmental mechanisms.3,4 In addition, the warmth may enhance blood flow, which is thought to help in dissipating algogenic chemicals, and it may facilitate muscle relaxation. The hydrostatic effect may also relieve pain by reducing peripheral edema5 and by dampening sympathetic nervous system activity.6
Recent reports have demonstrated the effectiveness of comprehensive health education, including lifestyle education and exercise in combination with spa bathing, for male white-collar workers,7 and middle-aged and elderly people.8,9
It is well known in research design that evidence grading is highest for a systematic review (SR) with meta-analysis of randomized controlled trials (RCTs). In “the recent review (summary)10 of the SRs of RCTs”, it was reported that there were three SRs1,2,11 that included meta-analyses of RCTs on aquatic exercise. Bartels et al1 reported that aquatic exercise had some beneficial short-term effects for patients with hip and/or knee osteoarthritis. Hall et al2 reported that aquatic exercise had a small post-treatment effect in relieving pain compared with no treatment for patients with neurologic and musculoskeletal diseases, but there were no differences in pain relieving effects between aquatic and land exercise. Pittler et al11 suggested that spa exercise may be effective for treating patients with chronic low back pain. However, we did not find any SRs of RCTs in which physical (eg, cardiovascular fitness) or psychological (eg, depression) effects were the primary outcome measurements.
An RCT is initially very difficult to execute and contains etiological issues, while the design of a non-RCT (nRCT) is easy to implement compared with an RCT. Although many studies have reported the curative effects of locomotrium diseases through aquatic exercise, there have been no systematic reviews of the evidence based on nRCTs. The objective of this study was to integrate the evidence from nRCTs on the curative effects through aquatic exercise for various diseases, and to assess the quality of those trials.
Methods
Criteria for considering studies included in this study
Studies were eligible if they were nRCTs and included one treatment group in which curative aquatic exercise was applied. Any type of aquatic exercise for cure and not for sports (eg, swimming) was permitted. The use of medication, alternative therapies, or lifestyle changes was described, and had to have been comparable in the group studies. There was no restriction on language.
Search methods for identification of studies
We searched the following databases from 2000 up to July 20, 2009: MEDLINE via PubMed, CINAHL, Web of Science, and Ichushi-Web (in Japanese). The International Committee of Medical Journal Editors (ICMJE) recommended uniform requirements for manuscripts submitted to biomedical journals in 1993. We selected articles published on and after 2000 because it appeared that the ICMJE recommendation had been adopted by the relevant researchers and had strengthened the quality of reports.
All searches were performed by two specific searchers (hospital librarians) who were qualif ied in medical information handling, and who were sophisticated in clinical trial research.
Search strategies
The search strategies contained the following elements and terms for all databases:
-
I: Search “aquatic exercise” or “water exercise”
-
II: Search “water gymnastic” or “water aerobics” or “pool exercise” or “pool therapy” or “aerobic aquatics” or aquatics
-
III: Search “exercise therapy”[MeSH] and “water”[MeSH]
-
IV: Search “water-based exercise”[All Fields] or “water-based training”[All Fields] or “aquatic therapy”[All Fields] or “aquatic physical therapy”[All Fields] or “water training”[All Fields] or “water-gymnastics”[All Fields]
-
V: Search I or II or III or IV Limits: Publication Date from January 1, 2000 to 2009
-
VI: Search I or II or III or IV Limits: Publication Date from January 1, 2000 to 2009, Randomized Controlled Trial
-
VII: Search V not VI.
Only keywords about intervention were used for the searches. First, titles and abstracts of identified published articles were reviewed to determine the relevance of the articles. Next, references in relevant studies and identified nRCTs were screened.
2000 is the year the CONSORT Statement became available on the Internet. The CONSORT Statement was created in the mid-1990s for improving the quality of RCTs. Because of the impact of the Internet, the quality of RCTs has improved since 2000.
Reference checking, hand-searching and others
We did not check the references of included studies, perform any hand-searching, or contact any institutions, societies, or specialists known to have expertise in aquatic exercise, or authors of included studies to identify any additional published or unpublished data.
Selection of trials
To make the final selection of studies for the review, all criteria were applied independently by two authors (JK and NS) to the full text of articles that had passed the first eligibility screening (Figure 1). Disagreements and uncertainties were resolved by discussion between the review authors.
Studies were selected when 1) the design was an nRCT and 2) one of the interventions was a form of aquatic exercise. Curative effects were used as a primary outcome measure. Trials that were excluded are presented with reasons for exclusion (Appendix 1).
Quality assessment and summary of studies
To ensure that variation was not caused by systematic errors in the study design or execution, two review authors (MK and HK) independently assessed the quality of articles. A full quality appraisal of these papers was made using the TREND statement checklist12 and CLEAR-NPT checklist,13 developed to assess the methodological quality of nRCTs and nonpharmacological trials, respectively. Disagreements and uncertainties were resolved by discussion between the review authors.
For meta-analysis preparation, the target objects and main outcomes in each study were examined. We found that there were various kinds of target diseases in the studies reviewed: healthy young students, middle-aged or elderly people, or people with a certain disease. In addition, the studies were heterogeneous, and the main outcomes varied. Moreover, the quality of most studies was low according to the checklist results, and such low-quality studies were excluded from the analysis based on the Cochrane Review.1 We could not perform a meta-analysis since no variable was eligible.
One review author (HK) selected the summary from each of the structured abstracts.
Benefit, harm, and withdrawals
The GRADE Working Group14 reported that the balance between benefit and harm, quality of evidence, applicability, and the certainty of the baseline risk were all considered in judgments about the strength of recommendations. Adverse events, withdrawals, and the cost for intervention were especially important information for researchers and users of clinical practice guidelines, and we presented this information with the description of each article.
Results
Study characteristics
The literature searches included 402 potentially relevant articles (Figure 1). Abstracts from those articles were assessed and 88 papers were retrieved for further evaluation (checks for relevant literature). Sixty-seven publications were excluded because they did not meet the eligibility criteria (see Appendix 1). Twenty-one trials15–35 met all inclusion criteria (Table 1). The languages of the eligible publications were English (N = 9), Japanese (N = 11), and Korean (N = 1). Target diseases were knee and/or hip osteoarthritis,19,24,28 poliomyelitis,15 chronic kidney disease,21 discomforts of pregnancy,30 cardiovascular diseases,33 and rotator cuff tears.35 Many studies16–18,20–23,25–27,29,31,32,34 on nonspecific disease (healthy participants) were included (Table 2). All studies reported significant effectiveness in one or more outcomes. In particular, many studies reported that aquatic exercise had a significant effect on pain relief and outcome measurements for locomotor diseases.15,19,24,28,35 These intervention periods ranged from 2 weeks to 12 months. These reflected the difficulty of maintaining long-term participation in each intervention trial. Whatever the case, the long-term effects are not clear.
Withdrawals and adverse events
Withdrawals (dropouts) were reported in five studies,24,28,29,32,34 and adverse events were reported in four studies (Table 2). There were three studies15,19,35 that reported ‘nothing’ on adverse events, and one study28 reported a slipping accident on the poolside (details of the injury were unclear). Other studies did not provide information on withdrawals or adverse events.
Intervention costs
A description of intervention costs was included in only one trial,30 but the summary of that trial did not describe the costs (Table 2).
Quality assessment
We evaluated 21 items from the TREND checklist in more detail (Table 3). This assessment evaluated the quality of how the main findings of the study were summarized in the written report. A lack of description was noteworthy for the studies in general. The items for which the description rate was less than 30% were as follows: “information on how units were allocated to interventions (23.8%)”; “how sample size was determined and, when applicable, explanation of any interim analyses and stopping rules (23.8%)”; “method used to assign units to study conditions, including details of any restriction (19.0%)”; “inclusion of aspects employed to help minimize potential bias induced due to non-randomization (4.8%)”; “whether or not participants, those administering the interventions, and those assessing the outcomes were blinded to study condition assignment; if so, statement regarding how the blinding was accomplished and how it was assessed (14.3%)”; “if the unit of analysis differs from the unit of assignment, the analytical method used to account for this (9.5%)”; “statistical methods used for additional analyses, such as subgroup analyses and adjusted analysis (9.5%)”; “methods for imputing missing data, if used (14.3%)”; “flow of participants through each stage of the study: enrollment, assignment, allocation and intervention exposure, follow-up, analysis (19.0%)”; “dates defining the periods of recruitment and follow-up (14.3%)”; “baseline comparisons of those lost to follow-up and those retained, overall and by study condition (9.5%)”; “comparison between study population at baseline and target population of interest (4.8%)”; “indication of whether the analysis strategy was ‘intention to treat’ or, if not, description of how noncompliers were treated in the analyses (9.5%)”; “for each primary and secondary outcome, a summary of results for each study condition, and the estimated effect (9.5%)”; “inclusion of results from testing prespecified causal pathways through which the intervention was intended to operate, if any (14.3%)”; “summary of other analyses performed, including subgroup or restricted analyses, indicating which are prespecified or exploratory (4.8%)”; and “summary of all important adverse events or unintended effects in each study condition (19.0%)”.
Table 4 presents an assessment of the evaluation of study quality by use of the CLEAR-NPT checklist. This tool mainly evaluated the quality of the study conduct. The description rate for the details of the intervention was high (81%), but the rate was low in other primary items. The primary items for which the executive rate was less than 30% were as follows: “Were participants adequately blinded? (4.8%)”; “Were care providers or persons adequately blinded? (4.8%)”; “Were outcome assessors adequately blinded to assess the primary outcomes? (9.5%)”; and “Were the main outcomes analyzed according to the intention-to-treat principle? (14.3%)”.
Discussion
Overall evidence and quality assessment
There were only 21 nRCTs about aquatic exercise published from 2000 to July 20, 2009, possibly due to the increase in studies with an RCT design. In previous SRs1–3 of RCTs, the authors actually collected many articles that were published in 2000. Pooling of the data from nRCTs was not performed because of the heterogeneity of the studies, multiple outcome measurements, and overall poor presentation. We used the TREND and CLEAR-NPT checklists as quality assessments. There were serious problems with the conduct and reporting of the target articles. Our summaries detected omissions of description and nonfulfillments, including detailed information on participants, sample size, assessors, care (health care) providers, blinding, and analysis methods. Furthermore, the description of adverse events and withdrawals were generally insufficient. In the Cochrane Review,1 there are strict eligibility criteria for a meta-analysis, and for each article, heterogeneity and low quality of reporting must first be excluded. Because there was insufficient evidence in studies of aquatic exercise, due to poor methodological and reporting quality and heterogeneity, we are unable to offer any conclusions about the effects of aquatic exercise based on an SR.
Characteristics of articles
The relevant articles represented several studies of locomotorium diseases (N = 4), compared with few studies of respiratory, circulatory, and psychiatric diseases. This characteristic was apparent in the review article10 of SRs of RCTs. In that review, there were many articles (N = 15) that targeted the health enhancement effect instead of a specific disease, the primary and/or secondary outcome measurements and target participants varied, and a consistent trend was not detected.
We were interested in comparing not only nonintervention as the control, but also land exercise. There were five reports that defined land exercise as the control group, knee osteoarthritis19,28 and rotater cuff tears35 as the target disease; the remaining reports22,31 were about healthy people. Two studies of knee osteoarthritis did not describe intervention effects. The recent meta-analysis2 of RCTs showed no differences between aquatic exercise and land exercise for the pain-relieving effects on neurologic or musculoskeletal disease (P = 0.56; weighted standardized mean difference (SMD), 0.11; 95% confidence interval [CI]: −0.27–0.50; N = 103). Presently, we can not suggest that aquatic exercise is more effective than land exercise for pain relief in locomotorium diseases.
Future research agenda
There were no studies to clarify the effects on serum lipids, body composition, or blood pressure as the main outcome for patients with metabolic syndrome (MS). A joint scientific statement36 about MS was issued recently, and many studies based on those criteria or outcomes should be started soon. Studies of aquatic exercise using prevention or cure as the outcome are also expected.
Table 5 shows the future research agenda for aquatic exercise. In advanced nations and areas, it appears that there is interest in studies about mental health as well as MS. Researchers should use the respective checklists for research design and intervention method, which would lead to improvement in the quality of the study, and contribute to the accumulation of evidence. Suitable comparisons are necessary to explain why aquatic exercise is better than other types of dynamic exercise. Aquatic exercise needs a valuable resource (hot water pool), which can not be overlooked in the study feasibility. A recent study37 suggested that public health is moving toward the goal of implementing evidence-based interventions, but the feasibility of possible interventions and determining whether comprehensive and multilevel evaluations should be justified to accomplish it.
Study limitations
This study was based on the PRISMA statement,38 except for the meta-analysis. However, there were several limitations to the study. Some selection criteria were common to the studies, as described above; however, bias remained due to differences in eligibility for participation in each study. Publication bias was also a limitation. Since we did not limit our search to English, we found 11 articles (52%) published in Japanese. Furthermore, we did not check the references of hand-searches and did not contact institutions, societies, specialists known to have expertise in aquatic exercise, and authors of included studies to identify any additional published or unpublished data.
In terms of quality assessment, disagreements and uncertainties were resolved by discussion between two authors; discussions with a third expert and contact with authors for the purpose of clarification were not allowed.
Conclusion
Because there was insufficient evidence on aquatic exercise due to poor methodological and reporting quality, and heterogeneity of nRCTs, we were unable to offer any conclusions about the effects of this type of intervention. However, we were able to identify the problems with current nRCTs of aquatic exercise, and propose a strategy of strengthening study quality and stressing the importance of study feasibility as a future research agenda objective.