Applications, indications, and effects of passive hydrotherapy WATSU (WaterShiatsu)—A systematic review and meta-analysis

Background WATSU (portmanteau word: water and shiatsu) is a form of passive hydrotherapy in chest-deep thermoneutral water (35°C = 95°F = 308.15 K). It combines elements of myofascial stretching, joint mobilization, massage, and shiatsu and is reported to be used to address physical and mental issues. The objective of this systematic review (PROSPERO Registration No. CRD42016029347) and the meta-analyses was to assess the applications, indications, and the effects of WATSU to form a basis for further studies. Methods A search for “WATSU OR watershiatsu OR (water AND shiatsu)” was conducted without any restrictions in 32 databases. Peer reviewed original articles addressing WATSU as a stand-alone hydrotherapy were assessed for risk of bias. Quantitative data of effects on pain, physical function, and mental issues were processed in random model meta-analyses with subgroup analyses by study design. Effect sizes were expressed as Hedges's g (± 95% confidence intervals). Results Of 1,906 unique citations, 27 articles regardless of study design were assessed for risk of bias. WATSU has been applied to individuals of all ages. Indications covered acute (e.g. pregnancy related low back pain) and chronic conditions (e.g. cerebral palsy) with beneficial effects of WATSU regarding e.g. relaxation or sleep quality. Meta-analyses suggest beneficial effect sizes of WATSU on pain (overall Hedges’s g = -0.71, 95% CI = -0.91 to -0.51), physical function (overall Hedges’s g = -0.76, 95% CI = -1.08 to -0.44), and mental issues (overall Hedges’s g = -0.68, 95% CI = -1.02 to -0.35). Conclusion Various applications, indications and beneficial effects of WATSU were identified. The grade of this evidence is estimated to be low to moderate at the best. To strengthen the findings of this study, high-quality RCTs are needed.

Introduction were contacted for clarification or additional data. Since the search was performed without any restrictions, it was necessary to execute translations from the original language to English. If translations via Google Translate did not produce sufficiently comprehensible results or appeared to be misleading, colleagues with advanced skills in the concerned languages were contacted. Any disagreements in data extraction and quality assessment were solved by discussion.
Risk of bias was evaluated by two independent reviewers (AMS and CB) pursuant to the Cochrane Handbook for Systematic Reviews [36], employing its review management tool Rev-Man [37]. Items suggested by the Agency for Healthcare Research and Quality [38], and the Norwegian Knowledge Centre for the Health Services [39] were additionally incorporated into the questionnaire for a total of 19 criteria to address the diversity of study designs. Each item was scored "+" if the criterion was estimated indicating low risk of bias, "n/a" if the criterion was not applicable due to study design, "?" if information necessary to determine risk of bias was missing, and "-"if the criterion was determined to indicate high risk of bias.

Data synthesis and analysis
Random-effects model meta-analyses of the reported quantitative data of WATSU were conducted. The conduction of meta-analyses was pre-specified in the PROSPERO-protocol for variables for which enough comparable data could be retrieved. The different individual study effect sizes were standardized by dividing the effect sizes by the pooled standard deviations. To correct for potential overestimation of the true effect size in small study samples, effect sizes were expressed as Hedges's g [40]. Individual study effect sizes and their corresponding 95% confidence intervals (95%CI) as well as the overall weighted estimate and its 95% confidence interval were presented in a forest plot. If more than one randomized controlled trial (RCT, plural: RCTs) was included, subgroup analyses by RCTs and non-RCTs were performed, as pre-specified in the protocol. The difference in the direction of the scales (e.g. Visual Analog Scale (VAS) versus SF-36) was adjusted. If groups in controlled trials exhibited differences in baseline outcome measurements that were suspected to introduce risk of bias (e.g. incomplete wash-out in cross design), only data of the WATSU-intervention group were considered for meta analyses of the concerned variables. When several points in time were reported (e.g. baseline, pre-and post-WATSU, follow-up), those closest to the intervention were analyzed. When two or more scales were employed to describe a domain (e.g. VAS and SF-36 for pain) in a study, the one more frequently used in the included studies was analyzed. If several values were fully reported (e.g. range of motion in several joints), the values were pooled for further processing. When several calculated values (e.g. means and standard deviations of pain in different tasks) were reported, the number of participants was split to process all available data while avoiding artificially increased precision. If values were not reported in numbers, they were estimated from charts.
Heterogeneity was assessed by Cochran's Q-test, its degrees of freedom and corresponding p-value. The degree of heterogeneity was represented as Higgins' I 2 which was calculated as a relative measure (i.e. a proportion) of between-study variability [41]. Higgins' benchmarking was used for the interpretation of the I 2 value. According to this, an I 2 between 0% and 40% indicates that the heterogeneity might not be important while an I 2 around 30% to 60% might present moderate, from 50% to 90% substantial, and from 75% to 100% considerable heterogeneity [26]. The Comprehensive Meta-Analysis 2 software (CMA-Version 2 Professional, Biostat Inc., Englewood, USA) was used for calculating the effect sizes and the pooled estimate, as well as for establishing the forest plots.
Since only eleven, ten and five studies were included in the different meta-analyses for pain, physical function and mental effects, respectively, publication bias was not statistically analyzed [36]. Level (according to the Oxford Centre for Evidence-Based Medicine, CEBM) and grade (according to the Grading of Recommendations Assessment, Development and Evaluation working group, GRADE) of evidence were reported [42,43].

Study selection
The search strategy resulted firstly in a total of 3,848 (3,758 + 90) citations. Of the 195 peer reviewed articles (109 excluded secondary articles + 59 excluded original articles with WATSU as minor component of hydrotherapeutic programs + 27 included articles), 27 studies met the eligibility criteria and were further assessed in the present review (Fig 1, Table 1).  [9]. Control groups underwent no treatment [9,52], massage [8], active hydrotherapy [51, 53], passive stretching on land and in water [10] or were not specified [50].
The peer reviewed secondary literature comprised 109 articles, and the grey literature included a total of 424 items: 38 abstracts / conference proceedings, 156 theses, 99 books / book chapters, and 131 non-scientific contexts such as magazine-articles or webpages).

Applications, indications, and general effects
Detailed applications, indications and effects of the articles that were assessed for risk of bias are depicted in Table 1. In these studies, WATSU was applied in childhood, adulthood, and advanced age. The indications included musculoskeletal disorders (e.g. fibromyalgia, rehabilitation), and neurological (e.g. cerebral palsy, hemiplegia) and mental challenges (e.g. stress, sleep disorder). In addition, four experiments with healthy participants were conducted to study physiological effects of WATSU (e.g. changes of heart rate or blood pressure).
Applications and indications of WATSU as reported in the primary peer reviewed literature were contentually comparable with the secondary peer reviewed and grey literature. In both, WATSU was mainly reported to be used for the treatment of chronic conditions (e.g. fibromyalgia, asthma, neurologic conditions, geriatric care), and it was also described as one component of palliative care (e.g. during cancer and vigil coma state) [17,18,[66][67][68][69][70][71][72][73][74][75][76]. Most mentions referred to implementation of elements of WATSU in hydrotherapeutic interventions / aquatic exercises.
Concerning WATSU as a stand-alone therapy, the grey literature frequently focused on qualitative aspects of the experience of WATSU and its potential to address psychological trauma [14,77]. A central theme was being held and carried by the therapist throughout the treatment. The therapist's facilitation of the receiver's comfort, and provision of unconditional support are being considered pivotal to the therapeutic effects of WATSU [1,12,23,78,79].
The exceptional sensorimotor experiences reported during WATSU included the perception of weightlessness, sometimes interpreted as the notion of omnipresent support, or being back in the mother's womb [80]. Moreover, WATSU was credited with facilitating emotional growth, spiritual experiences, enhanced states of awareness, altered states of consciousness, visions of vivid colors with eyes closed, the impression of flying or floating in the air, feelings of utter connectedness with all beings, and inner stillness [1,14,23,24,77,[80][81][82][83].

Risk of bias
Unclear risk of bias was observed in a substantial number of items (see Fig 5 and Fig 6). This was partially inherent in the studies' design (e.g. CRs), partially due to poor reporting. Two  trials with randomized-and one trial with non-randomized controlled design were well documented, they presented overall low risk of bias [8,9,51]. When excluding items that were not applicable due to study design, some CSs [13,54,60,62] and CRs [11,35] were also identified as quite complete in their reporting-all presenting low risk of bias within the scope of their designs.

Study selection, designs, and assessments
While WATSU has been in use since the 1980s, when it was developed, and mentioned in the primary scientific literature since the turn of the millennium, a systematic review of the literature on its applications, indications, and effects does not yet seem to exist. The studies assessed for risk of bias reflect the overall research on WATSU: carried out in all regions of the world, several languages and a broad spectrum of study designs, they assess both the physical and mental effects of WATSU. Since the search was highly sensitive, only 27 of 3,848 articles were found to meet the PICOS-criteria.
Secondary peer reviewed and grey literature on the topic were also identified and described narratively.

Applications, indications, and general effects
In the studies included in the risk of bias assessment, WATSU was applied during childhood, adulthood and advanced age. Indications ranged from pain in premature newborns, the needs of cognitively impaired children, pregnancy related complaints, stress, fibromyalgia, hemiplegia to fall prevention. Being held and gently moved by someone in warm water may lead to the perception of a "safe harbor", eliciting kinesthetic memories of the uterus and early childhood [80]. Social touch is considered a necessity during childhood and remains a resource throughout life [84][85][86][87]. Mere gentle physical contact with another human being was shown to attenuate the subjective perception of social exclusion and pain [88,89]. Touch as therapeutic agent is being discussed in the literature with reference to c-tactile fibers, unmyelinated low-threshold afferents that respond particularly to velocities and temperatures of gentle skin-stroking caress [90,91]. The idea that warm water could act as a whole-body stimulator for this type of fibers is intriguing, although not scientifically verified [92].
On the other hand, individuals might enjoy-and benefit from-company and solitude to differing degrees based on past experiences of abandonment or inundation [93,94]. In this regard, certain responder profiles related to attachment style or traumatic life events might determine the success of WATSU and even the indication [95].
During inactive immersion, thermoneutral water seems to fade out of perception after a while-possibly because humans are not provided with hygroreceptors but instead learn to identify water by experience, with cool temperature being one key indicator of wetness [96]. In addition, with ears submerged, familiar noises are missing while unusual noises (e.g. heartbeat) attract the attention. The experience of a persistent lack of solid ground, being held passively and moved for long periods of time-everything that constitutes WATSU is unfamiliar for adults. Such interference with the normal inflow and outflow of stimuli and impulses is perfectly suited to generate altered states of consciousness [97]. One such phenomenon that can easily be quantified is altered perception of time, as reported regarding meditation or the flow phenomenon [98,99]. When comparing perceived and actual duration of the WATSU sessions in their trials, Cunha et al., 2010, found overestimations (while lasting 36 ± 2 minutes, sessions were perceived to last for one hour or more by 74% of the participants) and Hora et al., 2017, underestimations (sessions lasting 40 ± 5 minutes were perceived as lasting 29.4 ± 1.9 minutes) [54, 62]. Such deviations from normal states of mind, as frequently described in the grey literature about WATSU, have been investigated in depth with respect to sensory isolation in flotation tanks, where individuals float in darkness on the surface of a thermoneutral saline solution [100][101][102]. Interestingly, despite the human contact in WATSU compared to isolation in flotation tanks, similar effects were reported.
Effect of WATSU on pain. The meta-analyses indicated a beneficial effect of WATSU on acute and chronic pain that was observed consistently. Neither important nor statistically significant heterogeneity was evident between the included trials. The effects were measured immediately pre-and post-intervention, but also after up to 16 weeks of treatment. The small effect size observed by Rambo & Filippin, 2019, could be explained by floor-effects [64].
In the subgroup analysis, the beneficial effect was pronounced in RCTs, with very low heterogeneity between the two included RCTs. Two RCTs (and one CT), well documented and with low risk of bias support this evidence, thus qualifying for level 1a according to CEBM. Considering unclear risk of bias in the other eight included studies, the small number of participants, the observed low and non-significant heterogeneity, and the effect size, the overall grade of the established evidence (according to GRADE) was estimated as low to moderate at the best.
Other systematic reviews reporting on moderate to high quality trials comparing aquatic exercise with no therapy indicated beneficial effects for musculoskeletal pain (Cohen's d = -0.37 [95% CI = -0.56 to -0.18]) and pain in fibromyalgia (Cohen's d = -0.61 [95% CI = -0.91 to -0.30]) [103,104]. Pain reduction is a well-known phenomenon in hydrotherapy and subject to explanation models and hypotheses from relaxation after sensory overflow to the potential activation of unmyelinated c-tactile fibers [4,9,105]. Pain relieving effects sizes, e.g. in primary dysmenorrhea (moderate grade of evidence, Cohen's d = -0.43 [95% CI = -0.7 to -0.15]) and labor (very low grade of evidence, raw mean difference in VAS pain (0-100) of 10.30 [95% CI = 4.69 to 15.91]) are also attributed to acupressure, one form of which is Shiatsu [106,107].

Effect of WATSU on physical function
The meta-analysis indicated a beneficial effect of WATSU on physical function during chronic conditions and in healthy individuals with substantial and statistically significant heterogeneity among the included trials. Fetal growth between measurements might have contributed to the small effect in the 2015 Schitter et al. study [9].
The subgroup analysis confirmed the effect with very low heterogeneity among the three included RCTs. Six well-documented studies, all with low risk of bias, are included in this analysis (among them two RCTs and one CT), thus overall qualifying for evidence level 1a according to CEBM. Considering unclear risk of bias in the other four included studies (among them another RCT), the small number of included participants, the observed overall heterogeneity, and the effect size, the grade of the established evidence (according to GRADE) was overall estimated as low to moderate at the best.
There is moderate evidence indicating beneficial effects for physical function (Cohen's d = 0.32 [95% CI = 0.13 to 0.51) in musculoskeletal conditions comparing aquatic exercise with no therapy [103]. In the present meta-analysis, the construct of physical function was connected to lower muscular tone and stiffness, translating to less spasm and increased range of motion. These can be relevant preconditions for successful active exercises. In addition, passive proprioceptive training (as which WATSU could be interpreted) was observed to be surprisingly effective for motor learning, when compared with active exercise and visual demonstration [108].

Mental effects of WATSU
The meta-analysis indicated beneficial mental effects of WATSU during chronic conditions and in healthy individuals that was observed consistently with neither important nor significant heterogeneity. The beneficial effect was confirmed by the RCT included in this analysis. One RCT and one CT, both well documented and with low risk of bias were included in this analysis, thus overall qualifying for evidence level 1b according to CEBM. Considering the unclear risk of bias in the other three included studies, the small number of included participants, the effect size, and the low heterogeneity, the grade of the established evidence (according to GRADE) was estimated as low to moderate at the best.
While feelings of wellbeing are often central in qualitative descriptions of WATSU, this does not seem to be fully transferable to the effect of WATSU on actual pathologies such as depression (a common parameter in this analysis) [9,21,35,54,62,66]. Interestingly, Maczkowiak et al., 2007, reported very large effect sizes of WATSU related to BDI as a result of the combination of reception and in turn administration of WATSU by the study participants, who were clinically depressed [15]. One contributing factor to this result might be increased release of endogenous oxytocin (peptide hormone and neuropeptide which plays a role in social bonding) due to physical contact in combination with received signals of trust [109].
A systematic review on the topic did not report significant effect sizes of aquatic exercise on symptoms of depression in patients with fibromyalgia (Cohens' d = -0.19 [95% CI = -0.88 to 0.50]) [104].

Strengths and limitations
Strengths of this systematic review are the sensitive and very comprehensive search in 32 databases without any filters or language restrictions, and the consideration of secondary and grey literature. WATSU is a rather young therapy and embracing languages other than English proved to be very useful considering that only 29 of the 86 retrieved peer reviewed primary articles about WATSU and only 11 of the 27 articles assessed for risk of bias were written in English. The use of Google Translate enabled the inclusion of all accessible data. In addition, a pool of native speakers stood by to assist in case of ambiguity.
Google Scholar provided all but two of the 27 articles that were assessed for risk of bias. Moreover, 10 of these 27 articles were exclusively found by Google Scholar and in no other database. Therefore, Google Scholar was a valuable help to complete this search.
The use of Google Scholar for systematic reviews has been and is being discussed in the literature [110][111][112][113]. While clearly dominating other search engines employed in this review, Google Scholar presented other forms of incompleteness: e.g. defining a timeframe or filtering out patents caused the exclusion of interesting articles. Also not all articles listed as search results could actually be retrieved-on the one hand because Google Scholar assumed the researchers to actually be robots, on the other hand because only 100 pages (= 1,000 results) could be accessed, no matter the amount of estimated results displayed on top of the screen (e.g. "approximately 1,170 results" ended up as exactly 1,000). Even below this number the last pages with results could not be retrieved (therefore filtered "approximately 880 results" turned to suspicious 800). However, PubMed also reduced its 8 initial results for "WATSU" to 4 for no conceivable reason, once the filter was set on "humans". Therefore, the question must be raised, whether the term "systematic review"-suggesting thorough retrieval through "exhaustive, comprehensive searching" [114]-can be justified under these conditions at all. This, however, is not a weakness specific to the systematic review reported here but holds true for any systematic literature search.
Due to limited number of included studies, publication bias was not statistically analyzed [36]. On a beginner's level, motion sequences of WATSU are taught to help students by providing a framework-well educated and experienced practitioners, however, will abandon these sequences and follow the "free flow" where individually-oriented interaction between receiver and therapist is central [23]. Thus, a scientific approach to WATSU is challenging, because standardized procedures are contrary to some of its core principles-a strict execution of predefined motion sequences might be counterproductive for the effect of this form of therapy, in any case it would contradict the underlying idea [1]. Beyond that, the general conditions of the administration of WATSU varied greatly in the retrieved studies: water temperatures varied considerably when WATSU was applied as stand-alone therapy, and in articles that used WATSU to warm up or cool down, water temperatures were even reported to be as low as 25˚C (77˚F, 298.15 K) [115]. This might be due to pragmatic reasons, e.g. the availability of suitable pools with convenient depth and reasonable water temperature. However, 35˚C are clearly defined and recommended as optimal by WATSU's originator and the Worldwide Aquatic Bodywork Association (WABA) because water at this temperature is thermoneutral [1,[5][6][7]. Thermoneutrality is defined as the range of temperature that allows one hour of resting without changes in central body temperature. This condition was measured in water to be 35-35.5˚C in men in winter in Rochester, New York, at room temperatures between 25 and 28˚C (humidity not reported) [5]. The question remains whether this can be generalized, since adaptive thermal comfort was reported to change with seasonal outdoor temperatures [116]. Moreover, biological factors such as age, gender or ethnicity might be influential, as well as cultural habits (e.g. heating and cooling of indoor environments), and clients might become accustomed and conditioned to certain water and air temperatures over the course of their treatment series. Also, the interrelation of humidity and room temperatures could be of importance and might be worthwhile to report on in WATSU studies, as introduced by Chun et al., 2006, andChon et al., 2009 [10, 13, 117]. Shortening WATSU sessions to 30 minutes due to receivers' chilling in 32.7˚C water temperature after 45 minutes was reported in the grey literature [66]. Consequently, multiple questions arise: whether an intervention at too low a temperature could or should still be considered "WATSU", how essential the temperature (water and room) really is, what factors determine the "right" temperature and how temperature influences the treatment (from the first moment on, from when on).
Furthermore, timeframes of WATSU sessions when applied as stand-alone therapy varied in the assessed literature. A guiding rationale with respect to physiological changes (such as e.g. the relaxation response) that have been observed, or are assumed to have occurred or to have been satisfied after a certain period of time, seems to be lacking [118]. Also, guidelines concerning the duration of treatment series at given indications, or the frequency of WATSU sessions remain unclear. If there were a time necessary to integrate impressions, or to regenerate between one session and the next, this latency is not known. Dose-response issues are also a concern identified as a result of the meta-analyses, as they combine mere pre-and post-intervention values (short-term effects) with values assessed before and after entire treatment series. Only Ramirez et al., 2019, reported data from a longer (three-month) follow-up [51]. More studies with subsequent verification of the effectiveness and sustainability of the effects would be desirable.
In six studies, the topic of "side effects" was addressed, once positively (in a rehabilitation setting, an injured joint was once swollen after treatment). In general, one gets the impression that WATSU has virtually no negative or undesirable side effects, however, it will be essential to explore negative side effects and adverse events, as the applicability of WATSU in medical settings depends on this aspect.
This systematic review is limited by the quality of findings, as only seven of the studies assessed for risk of bias had control groups. The assessment revealed that not only the studies' designs, but also poor reporting impeded accurate judgement in several studies. In subgroup analyses, however, the effect sizes of the meta-analyses were confirmed consistently with minimal heterogeneity, indicating that the inclusion of trials with lower methodological rigor led to an increase of noise without major over-or underestimation of the effect size as observed in RCTs. The weighted effect sizes provided in the current article warrant attempts to reproduce the reported results and may support future researchers in designing adequately powered RCTs about the effectiveness of WATSU.

Summary, implications, and conclusions
In the literature retrieved in the present systematic review, WATSU was applied from childhood to advanced age. Indications included physical conditions (musculoskeletal, neurological) and mental issues (trauma, stress, depression, consciousness, emotional growth, spiritual experiences).
The implemented meta-analyses suggested beneficial effects of WATSU on pain, physical function, and mental issues. The level of this evidence is 1a (pain and physical function) and 1b (mental effects). Its quality according to GRADE was estimated to be low to moderate at the best since the number of study participants was small, and while the risk of bias was low in some of the analyzed studies, it was unclear in others due to study design or poor reporting. Nonetheless, considering the effect sizes and only one reported incident of negative side effects, WATSU can be confidently recommended for wellness purposes and cautiously for clinical applications in relation to the above-mentioned effects.
Investigations concerning the verification of the obtained results by experts could add further evidence. However, methodologically sound RCTs based on good clinical practice following recommended reporting guidelines to underpin frequency and magnitude of mentioned short-and long-term effects of WATSU will be needed in the future to determine WATSU's proper position within the healthcare system. It is suggested that dose response, and responder profiles among patients with different pathologies as well as healthy individuals be considered. Dose-response issues are a concern identified as a result of the meta-analyses, as there are reports of mere pre-and post-intervention values (short-term effects) and values assessed pretreatment series and post-treatment series. Furthermore, varying water temperatures and durations of both individual sessions as well as treatment series might depict practical reality, however, on behalf of comparability, gold standard definitions for WATSU (ideal setting, ideal timeframe) should be established.
The presented meta-analyses are suited to serve future researchers for designing trials and sample size calculations to further investigate the effect of WATSU on pain, physical function, and mental issues.