A scoping review of the effectiveness of underwater treadmill exercise in clinical trials of chronic pain

2.1 Eligibility criteria

Based on the recommendation for scoping reviews to include the breadth of literature on a topic [14], we did not exclude studies based on the research approach the trial used. All clinical trials, including but not limited to randomized controlled trials (RCTs), feasibility trials, and pilot study publications of clinical trials examining the effectiveness of underwater treadmill exercise for human participants living with pain, were included. To be eligible for the scoping review, studies were required to include adult participants (i.e., individuals aged 18 or older) living with chronic pain, defined as pain lasting for 3 months or longer [15]. Though the patient population was restricted to those living with chronic pain, there were no restrictions on the chronic pain diagnoses that participants had. Articles published since the inception of the searched database until May 9, 2024, were included in the scoping review. Review eligibility was restricted to articles that specifically included underwater treadmill exercise as an active intervention in the trial. Eligibility was limited to English-language, peer-reviewed published studies with full-text available. When full-text versions of articles were not available, an attempt was made to contact the primary author via email. Published studies that examined acute pain (i.e., pain lasting less than 3 months) or exercise training recovery (i.e., delayed onset muscle soreness) were excluded.

2.2 Information sources

An initial search of EMBASE was undertaken on June 28, 2022 with an update of the search strategy being conducted on May 12, 2024, using all identified keywords and index terms [14]. The search included articles published between the inception of the searched database (e.g., 1947 for EMBASE) and May 9, 2024. Second, MEDLINE, SPORTDiscus, CINAHL, and Cochrane Reviews were systematically searched on June 28, 2022, with an update of the search strategy being conducted on May 12, 2024. The same search strategy was applied to each database. This search strategy can be found in the Appendix. In addition, a search of gray (i.e., difficult to locate) material was completed through a manual Google Scholar search by reviewing the first ten pages of the search results to identify any other potential studies that met review inclusion criteria. Finally, the reference list of identified articles included in the review was searched to identify any additional sources.

2.3 Study selection

All articles identified through the search strategy were uploaded into Covidence (Melbourne, Australia), a software application for literature reviews. Duplicates were then identified and removed by Covidence and by manual search. Of the remaining articles, titles and abstracts were screened by two independent reviewers (EG and PR) to determine if they met study eligibility criteria and were excluded if they did not meet the identified inclusion criteria. The remaining articles underwent a full-text review by two independent reviewers (SK and GD) to assess eligibility again. In both processes, a third reviewer (MM) determined eligibility when consensus on whether to include a study in the review was not reached by the independent reviewers. A PRISMA flow diagram is provided to detail the screening process (Figure 1).

Figure 1

PRISMA flow diagram of the study selection process.

2.4 Data charting process

Two reviewers (MM and PR) independently carried out data extraction from the nine included clinical trials. Microsoft Excel was used to compile data extracted from the trials. To ensure accuracy and reduce the risk of bias, data extraction was independently performed and then crosschecked by a third independent reviewer (GD). For every publication, the author(s), year of publication, country the study was completed in, type of clinical trial (e.g., RCT), number of intervention groups, intervention type(s) (e.g., underwater treadmill exercise and wait-list control), group sample sizes, gender of participants, average and standard deviation of age of participants, pain diagnosis of participants, intervention characteristics (including frequency and length of intervention time per week, duration of intervention in weeks, and perceived exertion from exercise), and outcome measures (i.e., pain-related measures, mobility, strength, joint kinematics and/or velocity, balance, and QoL), were extracted. The data charting underwent iterative amendments as the review progressed to ensure all information relevant to the research question was captured.

3.1 Included study demographics

Our initial search identified 2,209 studies. No additional studies were identified from a perusal of study reference lists or our search of gray material. Covidence identified 314 duplicate studies, and manual search identified 1 additional duplicate study. Of the 1,894 remaining studies that were screened at the title and abstract level, 1,781 were removed because they did not meet inclusion criteria, and 113 were retained for full-text review. The full-text review yielded a total of 9 articles that met eligibility criteria for inclusion in the scoping review, with publication dates ranging from 2010 to 2023. Six of the studies were conducted in the United States [12,16,17,18,19,20], and three of the studies were conducted in Thailand [11,21,22]. Table 1 shows the characteristics of each study included in the scoping review.

3.2 Clinical trial types

The types of studies included in this scoping review were five quasi-experimental research designs [12,16,17,18,19], two randomized controlled prospective trials [20,22], one single-blind RCT [21], and one cohort pre-posttest study [11]. Of the quasi-experimental research designs, three involved a single-group double pretest posttest design in which participants were assessed before and after a control period and after the aquatic exercise period [12,16,17]. The remaining two quasi-experimental research designs were crossover designs [18,19]. In addition, Coons et al. [16,17] used the same sample in two separate studies included in this review.

3.3 Participant characteristics

Total number of participants from the 9 included studies was 337 (women = 289). The average age of participants from all of the studies was 61.2 (SD = ±2.91). The mean age in the studies ranged from 58.1 (SD = ±3.39) to 66.2 (SD = ±3.39) when participants from all intervention groups in each study were included.

3.4 Intervention types

All of the studies included walking-based underwater treadmill exercises. This intervention was often compared to either land-based exercise interventions or other alternatives such as pneumatic partial weight supported treadmill [21]. The average training frequency in the studies was three times per week, with exercise duration ranging between 10 and 45 min. Intervention length ranged between 1 and 8 weeks. Of these walking programs, most did not outline a rate of perceived exertion that was targeted during training times. A majority of the studies did not state an exercise intensity or specify a rate of perceived exertion regarding participant activity.

3.5 Key outcome measure findings

Several different outcome measures were used by the included studies to identify the effectiveness of underwater treadmill exercise. The pain was the most investigated outcome [11,12,16–22], followed by QoL [11,12,17,20–22], mobility [11,12,16,18,21,22], and strength [11,21,22], respectively. Measures of joint kinematics and/or velocity [17,18,19] and balance [12,18] were also investigated. VO_2max [18] and body weight [11,22] were also investigated. (The results of these latter outcome measures will not be reported here.)

3.6 Pain

All of the nine studies measured pain outcomes in participants with OA who participated in the study’s underwater treadmill exercise program (note that two of the included studies used the same sample of participants). Of the included studies, five were quasi-experimental research designs [12,16,18,19], two were randomized controlled prospective trials [20,22], one was a single-blind RCT [21], and one was a cohort pre-posttest study [11]. Eligible studies utilized a variety of outcome measures to report changes in pain levels after participating in the underwater treadmill exercise trial. These included the Western Ontario and McMaster Universities (WOMAC) Index of Osteoarthritis, the Knee Osteoarthritis Outcome Score (KOOS), the Knee Outcome Survey (KOS), the visual analog scale (VAS), and the Numerical Pain Rating Scale (NPRS) ranging from 0 to 12 (with 0 being “No pain” and 12 being “Agonizing pain”).

In Bressel et al. [12] (n = 18), participants completed a 4-week non-exercise control period, followed by a 6-week underwater treadmill exercise program. Participants were assessed before the control period (pretest 1), after the control period (pretest 2), and after the underwater treadmill exercise program (posttest). Participants had significantly higher KOOS and lower VAS scores (pretest 1 M = 57.1, SD = ±26.9, pretest 2 M = 50.3, SD = ±24.8, posttest M = 15.8, SD = ±12.6) after completing the underwater treadmill exercise program in comparison to their pretest two scores, reflecting a significant reduction in their reported pain.

Roper et al. [19] (n = 19) completed a quasi-experimental crossover design in which all participants completed three underwater treadmill walking sessions and three land treadmill walking sessions, with each session lasting 20 min. Each of the three sessions in each condition was separated by at least 24 h and completed within 1 week. Before switching exercise conditions, participants did not engage in exercise for 1 week. Using the VAS, Roper et al. [19] found that the reduction in perceived pain after underwater treadmill exercise (M = –15.4, SD = ±20.7) was significantly greater than the reduction in pain after land exercise (M = 0.1, SD = ±19.2) at posttest, p = 0.02. Similarly, Denning et al. [18], who used the same cross-over design as Roper et al. [19], found the reduction in perceived pain scores was significantly less after land exercise (M = –8.19, SD = ±12.3) compared with after underwater treadmill exercise (M = 3.36, SD = ±12.3), p = 0.01.

Using the NPRS, Kittichaikarn and Kuptniratsaikul [11] (n = 30) documented a significant 2.3-point decrease in pain from weeks 0 to 4 of underwater treadmill exercise attended three times weekly for 30 min each (pretest M = 6.4, SD = ±1.2, posttest M = 4.1, SD = ±2.0, p < 0.001). Kuptniratsaikul et al. [22] (n = 109) randomized participants into a pneumatic partial weight support treadmill group or an underwater treadmill group. Participants in both groups completed 30 min of exercise three times per week for 8 weeks. Participants demonstrated significant improvements in NPRS scores (week 0 M = 6.25, SD = ±1.39, week 8 M = 3.38, SD = ±1.98) and all dimensions of the WOMAC (week 0 M = 5.23, SD = ±1.48, week 8 M = 2.61, SD = ±1.58) for the underwater treadmill group, at levels equivalent to the pneumatic partial weight support treadmill training group. Similarly, Silvis et al. [20] (n = 61) found equal WOMAC score improvements in the underwater treadmill group (M change in score = –46.36, 95% CI [–65.32, –27.39]) compared to the land treadmill group (M change in score = –31.21, 95% CI [–50.66, –11.74]), p = 0.27, or upright cycle group (M change in score = –30.28, 95% CI [–50.37, –12.20]), p = 0.25. A similar finding for KOOS scores was obtained (M change in score = –13.05, 95% CI [–18.98, –7.11]), with a significant improvement found within the group but no difference between groups. Participants in this randomized prospective study were randomized into underwater treadmill walking, land-based treadmill walking, or upright cycling. They completed three exercise sessions per week for 8 weeks. Sessions gradually increased in duration from 12 min per session (sessions 1 and 2) to 40 min per session (sessions 12–24).

Kuptniratsaikul et al. [21] (n = 80) conducted a single-blind RCT in which participants were randomized into one of two groups. The control group preformed home exercises daily for 4 weeks, for 30 min per session, and the experimental underwater treadmill group exercised for 30 min, 3 days per week for 4 weeks at a moderate intensity. They found statistically significant improvements in pain outcomes within both groups post-intervention (M pain score change home exercise = –1.8, SD = ±1.7; M pain score change underwater treadmill = –2.3, SD = ±1.9), but no significant differences between groups (M difference = –0.53, 95% CI [–1.31, 0.26], p = 0.184).

Coons et al. [16] (n = 6) conducted an exploratory study investigating the effects of underwater treadmill walking on knee OA. The six participants involved in this study performed underwater treadmill walking at a self-selected pace that increased from 10 to 45 min during an 8-week study period, with sessions taking place three times per week. They found a significant reduction in NPRS scores from pre-control to post-intervention (M pre-control = 5.3, SD = ±2.3, M post-intervention = 2.1, SD = ±1.4, p = 0.03) and WOMAC pain subscale scores (M pre-control = 8.0, SD = ±2.8, M post-intervention = 4.8, SD = ±1.5, p = 0.04). In addition, they observed that self-selected walking speed increased as knee pain reduced during training sessions.

3.7 Patient-reported outcomes

One of the main patient-reported outcomes was QoL. A number of studies documented changes in QoL using subscales of the KOOS, participant satisfaction ratings, the KOS, ratings of subjective symptom improvement, the Euro-QoL-5Dimensions-5Levels (EQ-5D-5L) Survey, and the 12-Item Short Form (SF-12) Health Survey. Bressel et al. [12] used the KOOS, which, in addition to pain measures, includes items assessing function in daily living, function in sport and recreation, and knee-related QoL. All KOOS subscale scores improved post-underwater treadmill exercise program compared to pretest 2 (i.e., after the control period), with statistical improvements ranging from 30 to 49%. In the Kittichaikarn and Kuptniratsaikul [11] (n = 30) study, two-thirds of patients were “very satisfied” with the underwater treadmill exercise program, and 90% rated their symptoms as “improved” or “much improved.” Kuptniratsaikul et al. [22] (n = 109) used a more generic QoL measure, the EQ-5D-5L. They found that the underwater treadmill exercise group’s EQ-5D-5L scores improved at post-intervention testing at a level equivalent to the pneumatic partial weight support treadmill training group. Moreover, Kuptniratsaikul et al. [21] (n = 80) noted greater self-reported global improvement (on a scale of “No change,” “Improved,” and “Much improved”) in the underwater treadmill group and better self-reported patient satisfaction (on a scale of “Unsatisfied,” “Satisfied,” and “Very satisfied”) compared to the home exercise program group. Coons et al. [17] (n = 6) also identified a significant improvement in total Activities of Daily Living Scale score of the KOS when comparing pre-intervention and post-intervention measurements (M pre-intervention = 57.4, SD = ±11.0, M post-intervention = 79.5, SD = ±7.4, p = 0.008).

Although Roper et al. [19] (n = 14) did not specifically measure differences in QoL between aquatic and land treadmill exercise groups, subjective feedback from a majority of commenters indicated a preference for underwater treadmill exercise. Silvis et al. [20] (n = 61) used the SF-12 Health Survey to document effects on QoL, but no statistical differences from baseline to post-treatment were observed for any of the interventions (underwater treadmill, land treadmill, or upright cycle).

3.8 Joint kinematics and velocity

Three studies in this scoping review compared measures of gait kinematics between interventions. Denning et al. [18] (n = 19) measured stride length and rate to determine whether changes in these variables could explain potential changes in Timed-Up-and-Go (TUG) scores. Posttest gain scores (i.e., the difference between pretest and posttest scores) for these variables did not differ between aquatic and land treadmill groups, despite Denning et al. [18] finding significantly greater TUG gain scores after land versus underwater treadmill exercise.

In addition, Coons et al. [17] (n = 6) noted a significant difference in knee flexion excursion (in degrees) between post-intervention and pre-intervention values (M pre-intervention = 51.4, SD = ±8.0, M post-intervention = 55.1, SD = ±8.1, p = 0.01).

However, Roper et al. [19] (n = 14) found no statistical difference in step rate and step length gain scores between underwater treadmill and land treadmill conditions. This was also the only study to complete a comprehensive analysis of joint angular velocities, and of note, participants consisted primarily of individuals with left-sided OA. Underwater treadmill exercise resulted in statistically significant improvements in angular velocity for left knee extension during stance (p = 0.004), as well as left knee internal rotation and extension during swing (p = 0.004, p = 0.008, respectively). This is in contrast to the increases in left hip flexion and ankle abduction during stance following land exercise, the changes that were not found in the underwater treadmill exercise group. These statistically significant changes in gait kinematics were reported after just three exercise sessions.

3.9 Mobility

Three outcome measures were utilized to report changes in mobility: the 6-min walk test (6MWT), the 10 m walk test (10MWT), and the TUG. The 6MWT was the most frequently used outcome measure, being employed in three studies [11,21,22]. The TUG was only investigated by Denning et al. [18] and Coons et al. [16], and the 10MWT by Bressel et al. [12] and Coons et al. [16].

Compared with 6MWT scores, Kuptniratsaikul et al. [22] (n = 109) found that significant improvements were reported in post-intervention compared to pre-intervention for both the underwater treadmill and pneumatic partial weight supported treadmill training groups (p < 0.001 for both), though there were not any statistically significant differences identified when 6MWT post-intervention values were compared between the underwater treadmill group and the pneumatic partial weight supported treadmill training group. Kuptniratsaikul et al. [21] (n = 80) also found significant improvements in distance walked during the 6MWT at the end of the underwater treadmill intervention compared to baseline, but no significant differences were found between the underwater treadmill and at-home exercise group, p = 0.34. Kittichaikarn and Kuptniratsaikul [11] (n = 30), whose study did not include a comparison group, nonetheless found significant improvements in walk distance (i.e., 34.9 m) between baseline and the end of the underwater treadmill intervention, p = 0.002. Denning et al. [18] (n = 19) investigated TUG performance scores in both underwater and land-based treadmill groups. After computing the gain scores (i.e., the difference between pre-intervention and post-intervention values), a statistically significant improvement in TUG performance scores was found for the underwater treadmill group (M = 0.83, SD = ±2.85), p < 0.05), but no significant improvement was noted for the land-based treadmill group. In fact, an increase in score time was noted (M = –0.55, SD = ±1.38). When examining 10 MWT scores in underwater treadmill participants, Bressel et al. [12] (n = 18) found a 12% reduction in posttest times compared to pretest two times (i.e., after the control period; p = 0.008).

Coons et al. [16] (n = 6) found no significant differences in measurements of 10MWT (M pre-control = 1.11, SD = ±0.25, M post-intervention = 1.18, SD = ±0.27) and TUG scores (M pre-control = 9.7, SD = ±2.5, M post-intervention = 9.6, SD = ±2.6). However, they observed an increase in self-selected walking speed during the exercise intervention as the 8-week study progressed.

3.10 Balance

Two outcome measures were used to assess balance: the TUG and computerized dynamic posturography (CDP). The TUG was classified by Denning et al. [18] (n = 19) as assessing both mobility and balance; their TUG results are noted above. Bressel et al. [12] (n = 18) employed CDP using the SMART EquiTest system. This system measured balance using three tests: a sensory organization test, a motor control test, and a limits of stability (LOS) test. The researchers found that all specific balance measures, including sensory organization, motor control, and LOS, except for the LOS directional score, were significantly improved after the 6-week underwater treadmill exercise intervention [12].

3.11 Strength

Quadriceps strength was measured in three studies using hand-held dynamometers and each reported improvements after completion of the underwater treadmill exercise intervention. However, Kuptniratsaikul et al. [22] (n = 109) reported that the pneumatic partial weight supported treadmill group demonstrated greater improvement in quadriceps strength as compared to the underwater treadmill group (p < 0.05), whereas Kuptniratsaikul et al. [21] (n = 80) reported no statistically significant differences between the underwater treadmill group and the at-home exercise group (p = 0.088). Although no control group was utilized, Kittichaikarn and Kuptniratsaikul [11] (n = 30) also reported a statistically significant improvement (p < 0.001) in quadriceps strength using their study design of testing quadriceps strengthen pre- and post-intervention.

4.1 Future research and practice

This scoping review provided valuable insights into the effectiveness of underwater treadmill exercise for chronic pain. The results of the review suggest that participation in underwater treadmill training can lead to improvements in pain, mobility, balance, strength, and QoL in patients with chronic pain over time. However, only three RCTs were available in the literature and included in this review, making definitive conclusions about the effectiveness of underwater treadmill exercise compared to other interventions difficult to make. This highlights the need for a greater number of RCTs comparing the efficacy of underwater treadmill exercise to wait-list control groups or other intervention groups. Additionally, the low number of men participants in the aggregated clinical trials (n = 48) and the lack of gender-diverse individuals included in the trials point to a need to increase recruitment of these participants in future clinical trials. This will allow for data to be disaggregated by gender/sex and for researchers to determine if the efficacy of this intervention differs by gender/sex. Additionally, based on the evidence available within the literature, further research is warranted that investigates an optimal protocol for underwater treadmill exercise, the long-term effects of this exercise modality on physical and QoL outcomes, and includes additional chronic pain populations other than individuals with primarily knee OA, such as back pain, fibromyalgia, and other forms of arthritis.

Examining the effect of properties of water, such as resistance, buoyancy, and water temperature, on underwater treadmill exercise outcomes may also be beneficial. Moreover, studies should determine and outline the most beneficial duration, frequency, and intensity of underwater treadmill exercise, as this will allow this intervention to be applied outside of clinical trial settings and into community and therapy pools. Prospective research studies investigating the long-term outcomes of this exercise modality are also important to assess the impact of this intervention on QoL, physical fitness, and chronic pain over an extended period of time. Such studies would help identify appropriate treatment frequencies and follow-up sessions needed for patients to optimize their rehabilitation.

4.2 Strengths and limitations

This scoping review is the first to identify research examining the effectiveness of underwater treadmill exercise in reducing chronic pain in adults. By doing so, it highlights the positive outcomes of this intervention and sheds light on the areas that require more research, which may promote the utilization of this exercise modality in future research and practice, giving clinicians another tool with which to treat chronic pain. Furthermore, this review can be used to help clinicians compare the utility of underwater treadmill exercise to research the utility of other forms of aquatic therapy. In addition, several of the included studies did not record adverse events within their protocols. However, of those that did, two recorded no adverse events in the underwater treadmill intervention [12,20]. A subsequent two studies recorded adverse events, which included joint pain, muscle pain, nausea and/or vomiting, dyspepsia, and vertigo. However, the number of adverse events was not significantly different in regards to occurrence between groups [19,20]. It was also observed by Kuptniratsaikul et al. [21] that there were more adverse events in the control group than in the underwater treadmill intervention group. This lends support to the contention that underwater treadmill interventions are effective and safe for patients in supervised settings.

There are several limitations to this scoping review. The included studies lack variation in their pain samples. In the nine studies, all included participants were diagnosed with OA. Additionally, the evidence obtained from all studies was primarily low-grade evidence. Five of the 9 studies had a sample size of less than 20 participants [12,16,17,18,19], and out of the 9 included studies, none incorporated a follow-up assessment of participants. Designing a longitudinal study and increasing the sample size to observe the long-term effects of underwater treadmill exercise will be required in future studies. Despite the apparent utility of underwater treadmill exercise for at least OA chronic pain, a limitation of this form of exercise modality is its cost and the limited availability of underwater treadmills. Scarce access to the device may impede individuals’ ability to participate in and continue with underwater treadmill exercise programs. Additionally, the control groups in all studies were not consistent. Four out of nine studies used pretest results as a “control group” [11,12,16,17] and others used either land treadmill training or other forms of exercise. A greater number of RCTs are needed to determine the effectiveness of underwater treadmill exercise relative to other interventions and no intervention. In addition, many of the included studies had short intervention periods and did not report on long-term outcomes, which may limit the impact and generalizability of the results found. Furthermore, all the studies included in this scoping review that used the VAS to measure changes in pain did not meet the MCID (23 mm) for the VAS [25]. This may indicate that underwater treadmill training reduces pain perception, but not to a level that is clinically important for patients. Further RCTS of higher-quality evidence will need to be conducted to confirm whether this is the case. Finally, several of the included studies only used 3-level satisfaction anchors to evaluate self-reported treatment efficacy, with only 1 indicating the percentage of participants that self-reported improvement [11]. It should be noted that pre-existing studies have identified that patients are influenced by their current health status when making improvement ratings [26]. Theoretically, this means that healthier individuals, regardless of diagnosis, have a greater tendency to report more satisfied improvement ratings. As such, 3-level satisfaction anchors may have biased some of the data found in the included studies.