Clinical Rehabilitation 2009; 23: 3–14

Therapeutic aquatic exercise in the treatment of low back

pain: a systematic review
Benjamin Waller University of Jyväskylä, Finland, Johan Lambeck Faculty of Kinesiology and Rehabilitation Sciences,
Katholieke Universiteit Leuven and Daniel Daly Faculty of Kinesiology and Rehabilitation Sciences, Katholieke Universiteit
Leuven, Belgium

Received 22nd February 2008; returned for revisions 20th April 2008; revised manuscript accepted 16th August 2008.

Objective: To examine the effectiveness of therapeutic aquatic exercise in the

treatment of low back pain.
Design: A systematic review.
Methods: A search was performed of PEDro, CINAHL (ovid), PUBMED, Cochrane
Controlled Trials Register and SPORTDiscus databases to identify relevant studies
published between 1990 and 2007. Population: Adults suffering from low back pain.
Intervention: All types of therapeutic aquatic exercise. Comparison: All clinical trials
using a control group. Outcomes: Oswestry Disability Index, McGill Pain
Questionnaire, subjective assessment scale for pain (e.g. visual analogue scale) and
number of work days lost as a direct result of low back pain. Methodological quality
was assessed using the PEDro scale and the SIGN 50 assessment forms.
Results: Thirty-seven trials were found and seven were accepted into the review.
Therapeutic aquatic exercise appeared to have a beneficial effect, however, no better
than other interventions. Methodological quality was considered low in all included
studies. The heterogeneity among studies, in numbers of subjects, symptoms
durations, interventions and reporting of outcomes, precluded any extensive meta-
analysis of the results.
Conclusion: There was sufficient evidence to suggest that therapeutic aquatic
exercise is potentially beneficial to patients suffering from chronic low back pain and
pregnancy-related low back pain. There is further need for high-quality trials to
substantiate the use of therapeutic aquatic exercise in a clinical setting.

Introduction causes of disability.1 Between 75% and 85% of the

population will experience some form of low back
Low back pain is the most common cause of refer- pain during their lifetime. In the UK it has been
ral to a physical therapist and is one of the leading estimated that low back pain costs the economy
£10 688 million (more than 20 million dollars) per
year through medical costs and lost work days.2
Address for correspondence: Professor Daniel J Daly, Low back pain can be classified into three cate-
Department of Rehabilitation Science, Faculty of
Kinesiology and Rehabilitation Sciences, Katholieke
gories: acute, subacute and chronic. In most cases
Universiteit Leuven, Tervuursevest 101, 3001 Leuven, (90%) pain is resolved within 12 weeks without
Belgium. e-mail: [email protected] long-term impairment.3 Chronic low back pain
ß SAGE Publications 2009
Los Angeles, London, New Delhi and Singapore 10.1177/0269215508097856
4 B Waller et al.

accounts for the remaining 10% of the cases and aquatic exercise is mentioned in a number of
is responsible for the majority of the associated recent low back pain guidelines,6–10 there is no
economical burden.3,4 systematic review available looking at the effects
The management of low back pain is multifa- of this treatment form and the quality of the avail-
ceted.5 A recent systematic review concluded that able literature.
exercise therapy relieves pain and increases func- Therefore the objective here was to answer
tion in patients suffering from non-specific low the following question: Is therapeutic aquatic exer-
back pain,1 a finding supported by other published cise an effective treatment for relieving low back
treatment guidelines.6–8 Exercise therapy is consid- pain?
ered a vital part of a multifaceted approach to the
treatment and prevention of low back pain.8–10
Between 51% and 72% of expectant women
suffer from pregnancy-related back and pelvic Methods
girdle pain11,12 and it is a common reason for
lost work time, early commencement of maternity Literature search
leave and decreased ability to perform activities of A literature search was performed to identify all
daily living.13 Causes are thought to be related to possible studies that could help answer the
loosening of the pelvic ligaments as the body pre- research question. PEDro, CINALH (ovid),
pares for childbirth11 and recommended treat- PUBMED, Cochrane Controlled Trials Register
ments include exercise therapy, back support, and SPORTDiscus databases were examined.
massage and education.13 The recent systematic The databases were searched using combinations
review by Stuge et al.13 on exercise in the treat- of the keywords and search limits presented, with
ment of pregnancy-related back and pelvic girdle an example for PUBMED, in Appendix 1.
pain concluded that exercise is beneficial but not
superior to other interventions such as electrother-
apy, exercise and sacroiliac belt.13
Aquatic therapy has been used for many years Inclusion criteria
in the management of musculoskeletal problems Inclusion criteria were defined using the PICO
including low back pain. Water immersion model (population, intervention, control/compar-
decreases axial loading of the spine and, through ison and outcome).
the effects of buoyancy, allows the performance of
movements that are normally difficult or impossi-  Population: People older than 18 years suffer-
ble on land.14 By utilizing the unique properties of ing from low back pain. The inclusion of all
water (buoyancy, resistance, flow and turbulence) types of low back pain was essential to identify-
a graded exercise programme from assisted to ing at which stages therapeutic aquatic exercise
resisted movements can be created to suit the might be most effective. Women during preg-
patients’ needs and function. Additionally, water nancy were included while patients post surgery
is theoretically an ideal and safe medium for preg- were excluded.
nant women to exercise because the spine and  Intervention: All types of therapeutic aquatic
pelvis are supported by buoyancy and hydrostatic exercise such as aqua-aerobics and aqua-
pressure. A meta-analysis of spa therapy and bal- jogging were included. Spa therapy and
neotherapy indicated that these treatments could balneotherapy (non-active) were excluded.
also be beneficial for reducing low back pain.15  Control/comparison: Randomized controlled
The meta-analysis indicated a positive difference and clinical non- or quasi-randomized con-
in pain (intervention verses control: visual analo- trolled trials (CCT) were included.
gue scale) after spa therapy of 26.6 mm (95% con-  Outcomes: Oswestry Disability Index, McGill
fidence interval (CI) 20.4–32.8, n ¼ 442) and after Pain Questionnaire, subjective assessment
balneotherapy of 18.8 mm (95% CI 10.3–27.3) scales for pain (e.g. visual analogue scale) and
n ¼ 134).14 Therapeutic aquatic exercises were not number of work days lost as a direct result of
included in these studies. Although therapeutic low back pain.
 Aquatic exercise for low back pain 5

Quality assessment were summed and a higher score represents better

The databases were searched and 588 studies methodological quality. A study scoring 6 from
were identified and examined. Based on titles the 10 criteria is considered to have a high meth-
those clearly deemed inappropriate or doubles odological quality and those under 6 a low-
were immediately excluded (Figure 1). The full methodological quality.17
abstracts of the remaining 37 articles were read The articles were further evaluated using the
and a final selection was made. Reference lists SIGN 50 (Scottish Intercollegiate Guidelines
from all these studies were also examined but no Network) assessment forms.18,19 The SIGN check-
additional potential studies were found. To ensure list includes three sections: the first considers inter-
accuracy the accepted studies were further read nal validity, second degree of bias and third assists
and assessed by three reviewers and comparison extracting relevant data from the study (see
of findings between two reviewers was made. Tables 1–3). There is no weightings of the answers.
In case of disagreement a third reviewer was The degree of bias was classified into three groups.
included. When further disagreement remained, a Low: all or most of the criteria have been fulfilled
senior professor or a university sports faculty therefore conclusions of the study or review were
member was consulted. Reviewers were not still unlikely to be altered. Moderate: some of the
blinded to author, institution or journal. criteria have been fulfilled, but the conclusions are
Initially methodological quality was assessed unlikely to alter. High: few or no criteria fulfilled
using the PEDro16 Scale which is based on the and the conclusions of the study are thought likely
Delphi list and has been reported to have a fair or very likely to be altered.
to good reliability for its use in systematic reviews
of randomized controlled trials in physiother-
apy.17 The scale awards each study a value from Analysis
0 to 10 based on a series of 11 criteria (the first Based on the selected studies comparisons could
criterion is not included in the final score) with be made between therapeutic aquatic exercises
each criteria having a simple yes (1)/no (0) versus (a) active land exercises or (b) no interven-
answer. For a yes to be awarded the answer tion in the management of low back pain. Where
must be clearly reported in the study. The scores possible, standardized mean differences and 95%

Potentially relevant studies

identified and screened (n = 37)

Studies excluded, non-aquatic

therapy interventions (n = 24)

Potentially appropriate clinical

trials to be included in the
systematic review (n =13)

Clinical trials withdrawn, passive

aquatic intervention (n = 6)

Clinical trails included into

systematic review (n = 7)

Figure 1 Flowchart showing selection of studies.

 6

Table 1 Description of methodology used

B Waller et al.

Study Sjogren et al. McIlveen and Kihlstrand Schrepfer and Saggini Yozbatiran Granath
(1997)20 Robertson et al. (1999)22 Fritz (2000)23 et al. (2004)24 et al. (2004)25 et al. (2006)26
(1998)21

PEDro score 5 2 6 4 4 2 2
Randomization Sequentially Withdrawing Sealed States randomization Permuted block States randomiza- By date of birth
of trial allocated in order a marked lottery envelopes in abstract no other randomization tion in abstract no
of presentation ticket from a box details other details
to clinic
Patient blinded No No No No No No No
Therapist blinded No Not reported Not reported No Not reported Not reported Yes
Assessor blinded Yes Yes Not applicable Not applicable Not applicable Not reported Yes
Time of 6 weeks, 4 weeks, week 34 of Immediately after 7 weeks, 4 weeks Up to birth
follow-up used immediately immediately after pregnancy, 1st treatment immediately immediately
for analysis after treatment treatment 1 week after treatment after treatment
postpartum
Longer follow-up No No No No Yes 1 year No No
Outcome measures Schober, Modified VAS pain (daily), VAS pain, ODI VAS pain, Backill, VAS pain, ODI, Days of sick leave
VAS pain, Schober, Passive unvalidated medication 12-min walk associated with
ODI, Walking test, SLR, questionnaire, test, Sorensen, LBP. VAS pain
Medication Tendon days of sick SLB, sit and
reflexes, leave reach, sit up
Strength, associated test, BMI
McGill with LBP
Pain, Sensation,
ODI

VAS, visual analogue scale; ODI, Oswestry Disability Index; SLR, straight leg raise; BMI, body mass index; SLB, single leg balance; LBP, lower back pain.
Table 2 Participants and interventions used in the selected studies

Study Sjogren McIlveen and Kihlstrand et al. Schrepfer and Saggini Yozbatiran Granath
et al. (1997)20 Robertson (1999)22 Fritz (2000)23 et al. (2004)24 et al. (2004)25 et al. (2006)26
(1998)21

Subjects (N) 60 109 329 49 40 30 390

Intervention 30 45 129 24 20 15 192
Control 30 50 129 25 20 15 198
Drop-out 4 (2 from each) 14 (11 hydro) 9þ9 None None None 124 (60 ¼ water,
64 ¼ gym)
Age (years)
Intervention 58.11  11.60 57  15.2 28 40.5  11.3 43.8 39.6  6.33 29.1  4.50
Control 57.36  13.59 58  15.0 29 41.9  15.5 42.7 38.6  6.57 29.2  4.54
Duration of 46 months Not clearly N/A 590 days 412 months 43 months N/A
symptoms stated
Type of Non-specific LBP, leg pain Pregnant Acute LBP LBP, disc LBP (disc invol- Pregnant
symptoms LBP, Disc and disc women back and back/ disease vement, neuro women LBP and
degeneration disease with LBP leg pain excluded) pelvic pain
Intervention Aquatic exercise 60 min active Aquatic therapy, Deep water Aquatic therapy, Aquatic fitness, 45 min active
with lumbar aquatic 30 min aquatic walking three-stage 15 progressive aquatic therapy
spine ROM, therapy, exercise, progressive exercises and (strength,
general strength, 30 min programme cool down, flexibility and
strength and flexibility and relaxation stretching and fitness), 5 min
endurance endurance light aerobic relaxation
exercise
Control Land exercises, Waiting list Normal prenatal Deep water Body weight Land exercises, Land-based
same structure care hanging, using relief same structure exercise, same
as aquatic upper limb rehabilitation as aquatic aims as
buoyancy aids, and stretching aquatic
no weights
Treatment 2  50 min/ 2  60 min/ 1 h/week from 20 min  1 3/week for 3/week, for 1 h/week from
duration week, for week, for week 18 session 7 weeks 4 weeks week 18
6 weeks 4 weeks of pregnancy of pregnancy

ROM, range of movement; LBP, lower back pain.

Aquatic exercise for low back pain
7
 8

Table 3 Outcome of studies included

Study Sjogren et al. McIlveen and Kihlstrand et al. Schrepfer and Saggini et al. Yozbatiran et al. Granath et al.
(1997)20 Robertson (1999)22 Fritz (2000)23 (2004)24 (2004)25 (2006)26
(1998)21
B Waller et al.

Outcome Both interventions Improvement in Less pain VAS at No improvement Both groups No difference Less sick days in
produced ODI score 1 week in either interven- showed intra- between groups. aquatic therapy
improvements (P ¼ 0.04) in postpartum tion group group VAS and Both groups group (P ¼ 0.03)
in pain scores aquatic therapy (P ¼ 0.034) and Backill showed and less pain
(VAS) with no group less sick improvements. improvements experienced
difference days taken in No difference in ODI, Walk (P ¼ 0.04)
between groups aquatic therapy between test, sit up test,
group (P ¼ 0.09) interventions spinal flexibility
and trunk
strength*
Improvement in Pain score 27% of group 502 (34%) Less Pain score (VAS) Based on figure: Pain score No subjects
active aquatic (VAS) improved ODI sick days due þ4.2 mm (9.1%) 5 pt decrease in (VAS) þ35.3mm required sick
exercise group þ13.5 mm by þ10 patients to LBP taken. a 10 pt pain 64.7%* ODI leave due to
24.4%* vs. 8% in no Insufficient scale with 2 pt þ19.34 (48%)* LBP. Insufficient
ODI (8.7%)* treatment data for pain regressions at data for pain
scores 1 year scores
follow-up. No
regression
in weight
reduction group
Standard mean VAS –0.02 Insufficient data Insufficient data VAS 0.28 Insufficient data VAS 0.35 Insufficient data
difference (0.52, 0.49) (0.28, 0.84) (1.07,
(95% CI) ODI 0.10 0.37) ODI 0.03
(0.40, 0.61) (0.75, 0.69)
Control of Good Good Poor Good Poor Good Poor
co-interventions
Risk of bias Moderate Moderate Moderate High Moderate Moderate High
Intention-to-treat No No No No No No No
analysis
Control of Good Good Poor Good Poor Good Poor
co-interventions

*Significant with a P-value50.05.

ODI, Oswestry Disability Index; VAS, visual analogue scale.
 Aquatic exercise for low back pain 9

confidence intervals were calculated using the initial postintervention assessment.24 Based on
Cochrane Collaboration Review Manager 5 pro- the information gathered using the SIGN 50
gram, version 5.0.11. The heterogeneity among assessment guidelines, bias was considered moder-
studies, in numbers of subjects, symptom dura- ate (in 5 out of 7 studies) or high (in 2 out of
tions and especially interventions and outcome 7 studies) (Table 3).
measures along with inconsistent reporting of The study participants (in total n ¼ 1007) are
results, precluded any extensive meta-analysis. described in Table 2, including mean ages, symp-
toms and duration of low back pain and sample
size. In addition this table also presents the
Results interventions used. Only one study included
people with acute and subacute low back
pain,23 three studies examined people with
After the initial database search and selection chronic low back pain, 20,24,25 and in one study
based on title and keywords, a total of 37 studies the duration of symptoms was unclear. In stu-
were found. Based on reading of the full abstracts dies including pregnant women, low back pain
24 studies were then eliminated due to non-aquatic was classified as pregnancy-related low back or
interventions. The abstracts from all 13 remaining pelvic pain.22,26 The overall age range was 18–74
articles were then further examined and six addi- with mean age per study never above 60 years.
tional articles were excluded as the intervention The age ranges and duration/type of symptoms
was deemed passive (Figure 1). The remaining varied widely among studies. This fact as well as
seven articles20–26 were accepted into the review. unclear reporting prevented any further analysis
These included two studies pertaining to preg- of small cohort groups. Interventions all differed
nancy-related low back pain and the effect of
in content as well as duration (1–21 sessions)
aquatic exercise compared with normal prenatal
with the exception of the two pregnancy-related
advice.22,26 Two studies comparing aquatic exer-
studies where the treatments appeared to
cise to land exercise,20,25 two comparing active
be almost identical (1  week from gestation
aquatic therapy to static traction techniques23,24
week 18). Sjogren et al.20 and Yozbatiran
and one comparing aquatic exercise to no inter-
et al.25 attempted to reproduce the water train-
vention,21 all in the management of low back pain.
ing on dry land with the control group.

Methodological quality
Table 2 presents the methodology used in each Outcomes
study. Only one of the seven studies taken in The primary outcome of each study, as well as
this review scored 6 using the PEDro scale.22 All possible bias in the results, is given in Table 3.
studies included claimed to randomly assign parti- Intention to treat was not reported in any of the
cipants to the treatment or control group, however studies. In both the pregnancy-related back and
only three,21,22,24 used true randomization techni- pelvic pain studies significant benefits were
ques and only one of these used computerized demonstrated in both reduced number of sick
randomization. Two of the studies used quasi- days related to low back pain (34%,
randomization techniques20,26 and in the remain- P ¼ 0.09)22 and lower visual analogue scale
ing two papers, the method of randomization was pain score (P ¼ 0.034)22 and (P ¼ 0.04)26 in the
not reported.23,25 In no studies were patients aquatic exercise groups. In other low back pain
blinded to the treatment. Evaluator and therapist groups there was no significant difference (see
blinding was often poorly reported. The outcome Table 3) in effect between therapeutic aquatic
measures most commonly included were the visual and land exercises with mean effect sizes (95%
analogue scale for pain (6 out of 7) and Oswestry CI) of 0.02 (0.52, 0.49)20 and 0.35 (1.07,
Disability Index (4 out of 7), but there was no 0.37)25 for pain scores and 0.10 (0.40, 0.61)20
single outcome measure used in all the studies. and 0.03 (0.75, 0.69).25 The meta-analysis of
Only one study included a follow-up after the these did not provide additional information.
10 B Waller et al.

Both the experimental interventions and control role of therapeutic aquatic exercise in the manage-
interventions showed significant improvements ment of acute low back pain can be currently
compared with baseline measurements. Active made. None of the studies indicated any negative
aquatic therapy also improved the Oswestry effects. Drop-out rates were comparable if both
score (P ¼ 0.04) compared with no treatment groups received some kind of treatment.
after four weeks of intervention, with no signifi- The results indicate that the effect of therapeutic
cant changes in symptoms occurring in the con- aquatic exercise is comparable to that of spa ther-
trol group. No data concerning the size of the apy and balneotherapy. The mean change in visual
changes were reported.21 Schrepfer and Fritz23 analogue scale pain scores in three studies for the
compared the effect of one 20-minute session group participating in therapeutic aquatic therapy
of aqua-jogging with the same duration of could be calculated. Improvements of 4.2 mm
static aquatic lumbar traction. Their results (9.1%),23 13.5 mm (24.4%)20 and 35.3 mm
showed no significant pain relief as measured (64.7%)25 were reported. These improvements
with the visual analogue scale pain scale for appear to be similar to those reported by Pittler
the patients in either group (0.28 (95% CI et al.15 in the review of spa- and balneotherapy,
0.28, 0.84)). Saggini et al.24 found a significant suggesting that the effects might be similar.
decrease in pain (5 points on a 10-point scale) However due to methodological and numerical
and reduction of medication intake after seven differences direct comparison between the two
weeks of treatment for both a progressive aqua- types of interventions is hazardous.
tic exercise programme and a programme of The first comparison examined here was ther-
weight relief treatment and stretching. At one apeutic aquatic exercise verses no intervention,
year follow-up the aquatic intervention group for which only one study of low quality (2 out
had regressed somewhat while no regression of 10 in the PEDro scale) was included.21 The
was found in the weight relief treatment group. results indicated that aquatic exercise resulted in
Both improvements were still significant. None a significant improvement in function (P ¼ 0.04)
of the studies indicated a negative effect of as measured by the Oswestry Disability Index
active aquatic therapy in the treatment of low but not in any direct measurements of function.
back pain. This study did not report the descriptive data
from the outcome measures, thus preventing
comparison of the size of the change related to
the intervention. These authors did set standards
Discussion for clinically relevant improvement in the mea-
sures they use and pointed out that these stan-
This study indicates that therapeutic aquatic exer- dards were most often met in the aquatic
cise appears to be a safe and effective treatment intervention group even when mean changes
modality for patients who are suffering from did not reach statistical significance. The bias
chronic low back pain and women suffering in this study was considered high as the patients
from pregnancy-related low back pain. Six of the had already been referred to aquatic therapy by
studies20,21,22,24–26 showed that therapeutic aquatic an experienced clinician and therefore were
exercise produced a statistically significant benefit already presumed to benefit from aquatic
for patients suffering from chronic low back pain. therapy.
There was, nevertheless, no evidence that the con- Active aquatic exercises also compared favour-
trol interventions were more or less effective in the ably to land exercise.20,25 Both the aquatic and
treatment of low back pain at the end of interven- land-based exercise programmes produced signifi-
tion. The one study with a long-term follow-up did cant improvements in function as measured with
find that the alternative intervention had more the Oswestry Disability Index and reduction in
substantial long-term effects. Only one study23 pain scores (visual analogue scale), suggesting
included subjects suffering from acute low back that the water environment is possibly as effective
pain but due to poor methodological quality and for patients with low back pain as land. The study
limited intervention duration no conclusion on the by Yozbatiran et al.25 produced much larger
 Aquatic exercise for low back pain 11

improvements (although there was no statistical individual programmes.27 Social interaction was
difference). Possible reasons are that the interven- highlighted as an important factor increasing
tion was provided at a higher frequency than the patients’ adherence to exercise programmes for
Sjogren et al.20 study (three times a week com- chronic osteoarthritis.28 The programmes
pared to two), the earlier treatment phase or the described in this review were performed in
younger sample. The starting point of the patient groups. Adherence to an intervention is partly
group might have provided a larger potential for dependent on patient satisfaction, which was
improvement. The meta-analysis for this compar- examined in only one study.22 This study indicated
ison was not included in this study because it did that 98% of women would recommend aquatic
not provide any further information and because exercise to other pregnant women and would
of the differences in initial scores, the small sample also participate in aquatic exercise during their
size (n ¼ 45) and difference in methods. The com- following pregnancy.
parable effect of land and aquatic exercise is In all studies the aquatic exercise programmes
important to note in any case. used were different and in most cases not well
Schrepfer and Fritz 23 compared deep water reported, creating a major problem when trying
walking to deep water hanging with subjects suf- to apply the results of the trials clinically. Often
fering from acute low back pain (less than 90 days the details of the intervention were completely
duration of symptoms) and found no benefit from absent. The durations of the treatments ranged
either intervention. This study only included from one 20-minute treatment session to 21 one-
measurement of pain before and after a single hour treatment sessions and only one study
treatment session and scored very low on metho- attempted to reproduce a comparable control
dological quality and high on risk of bias. intervention. Frequency of the aquatic exercise
Inclusion of this article was nevertheless war- varied considerably from once to three times a
ranted as it fit the inclusion criteria of this week and interestingly three times a week pro-
review and considering that a secondary aim was duced the largest improvements.25 The degree
to investigate the quality of all relevant studies and duration of symptoms experienced by partici-
published. In addition, these interventions are pants in each study varied considerably. There was
not reproducible on dry land and therefore further no clarification whether symptoms were periodic
investigation into these methods is necessary. or constant or when the previous episode
Exclusion of this study would not have raised occurred. In some cases intervertebral disc invol-
this research question. vement was an exclusion criterion and in others it
Aquatic exercise is commonly used with pre- was not. This made comparisons between studies
and postnatal women and the evidence presented difficult, and combined with poor reporting
in this review indicates that it is both an effective prevented extraction of cohorts. It is therefore
and safe modality for the management of preg- unclear which patient groups would benefit
nancy-related low back pain. These findings sup- most from therapeutic aquatic exercise.
port those by Stuge et al.13 Pregnant women who Theoretically, patients with acute low back pain
undertook a one-hour active aquatic session once would find it easier to initiate an exercise pro-
a week had significantly less pregnancy-related gramme in water as it is easier to move, but results
back and pelvic pain (P ¼ 0.04) 26 and were 34% from these patients in this study were limited to
less often absent from work22 than pregnant one poor-quality study.23 Adherence to aquatic
women who received normal prenatal advices. therapy appears to be high and results were similar
During pregnancy, women receive information to other interventions. Therapeutic aquatic exer-
from various sources, family members, midwives cise could be used to motivate a patient whose
and friends and therefore the control of co- compliance to treatment is low or who has
interventions in these studies would have been become disillusioned with their current rehabilita-
difficult. tion programme. Therefore future research should
Compliance was high in the studies examined. focus on specific groups of patients to determine
Adherence to exercise has been shown to be higher when and how therapeutic aquatic exercise is most
for supervised exercise than for home-based effective in the treatment of low back pain.
12 B Waller et al.

The overall quality of the articles was poor

Clinical messages
with a number of methodological faults, especially
concerning randomization and its reporting. All
studies included in this review claimed to be ran-  Therapeutic aquatic exercise appears to be
domized controlled trials. However on evaluation, an effective treatment intervention for
with the help of a standard checklist18, only three chronic and pregnancy-related low back
studies used appropriate randomization methods, pain.
two studies used quasi-randomization methods  No studies reported a negative effect on low
and the remaining two papers did not report the back pain due to therapeutic aquatic
method used. Intention-to-treat is another essen- exercise.
tial part of evaluating the clinical relevance of the  More high-quality trials are needed to clarify
results. Only one study included a follow-up the role of therapeutic aquatic exercise in the
assessment.24 None of the reports examined management of low back pain.
stated that an intention-to-treat analysis was per-
formed although one study reported a 31% drop-
out rate.26 Only one study reported a much higher Acknowledgements
drop-out rate in the aquatic therapy group. In this We want to thank Laima Laurinavičiúte_ of the
case however, the alternative group was on a wait- Lithuanian Academy of Physical Education and
ing list for aquatic therapy and thus had every Boglárka Lipták, University of Szeged, Hungary
reason not to abandon the study. Only one for their help during the literature collection. We
paper contained a flowchart showing the phases also want to thank Professor Esko Mälkiä of the
of the randomized trial, as suggested by the University of Jyväskylä, Finland for his help and
CONSORT29 group. It is therefore essential that support.
all researchers undertaking a randomized con-
trolled trial familiarize themselves with the
CONSORT checklist when planning their study. Competing interests
The use of this checklist has been shown to signif- None identified.
icantly improve the quality of reporting an rando-
mized controlled trial.29 It must be stressed that
even though all the studies included showed sev- Funding
eral methodological and reporting flaws, all but Parts of this work were made possible by the
one study reported a positive benefit for the financial assistance of the ERASMUS IP Grant
patients as a result of active aquatic therapy Agreement no. 2006-2151/001-001 S02-21CIEU;
while no study found a negative effect from an Pr. no. 27945-IC-1-2005-1-BE-ERASMUS-IPUC-6.
aquatic intervention. and the LIKES-Research Centre for Health and
The weaknesses of this systematic review may be Sports Science, Jyväskylä, Finland.
in the exclusion criteria used. Spa therapy and
balneotherapy were both excluded, but distin-
guishing the difference between ‘active’ and ‘non- References
active’ aquatic therapy is difficult. The inclusion of
studies only written in English and limits within 1 Hayden JA, van Tulder MW, Malmivaara A, Koes
the keywords could have eliminated some appro- BW. Exercise therapy for treatment of non-specific
priate studies. The quality of the articles available low back pain. Cochrane Database Syst Rev 2005;
and the small sample size of 288 when excluding Issue 3. Art. No.: CD000335. DOI: 10.1002/
expectant mothers, created the potential for fault 14651858.CD000335.pub2.
2 Maniadakis N, Gray A. The economic burden of
in the results. However, it is the opinion of the back pain in the UK. Pain 2000; 84: 95–103.
author that the results accurately represent the 3 Balagué F, Mannion FA, Pellisé, Cedraschi C.
quality of the current literature covering this Clinical update: low back pain. Lancet 2007; 369:
subject. 726–28.
 Aquatic exercise for low back pain 13

4 Pengel HM, Maher CG, Refshauge KM. scale for rating quality of randomised controlled
Systematic review of conservative interventions trials. Phys Ther 2003; 83: 713–21.
for subacute low back pain. Clin Rehabil 2002; 16: 18 SIGN 50 (Scottish Intercollegiate Guidelines
811–20. Network). Checklist2: Randomised controlled trials.
5 Daimond S, Borenstein D. Chronic low back pain 2004. Accessed 7 July 2007, from: http://
in a working-age adult. Best Pract Res Clin www.sign.ac.uk/guidelines/fulltext/50/
Rheumatol 2006; 20: 707–20. checklist2.html
6 Airaksinen O, Brox JI, Cedraschi C et al. European 19 SIGN 50 (Scottish Intercollegiate Guidelines
guidelines for the management of chronic non-speci- Network). Checklist 4: Case–control studies. 2004.
fic low back pain. Accessed 12 July 2007, from: Accessed 7 July 2007, from: http://www.sign.ac.uk/
http://www.backpaineurope.org/web/files/ guidelines/fulltext/50/checklist4.html
WG2_Guidelines.pdf 20 Sjogren T, Long N, Story I, Smith J. Group
7 Bekkering GE, van Tulder MW, hydrotherapy versus group land-based treatment
Hendriks HJM, Oostendorp RAB, Koes BW, for chronic low back pain. Physiother Res Int 1997;
Ostelo RWJG, Thomassen J. Dutch phy- 2: 207–17.
siotherapy guideline for low back pain. (KNGF 21 McIlveen B, Robertson VJ. A randomized con-
richtlijn lage rugpijn). Ned Tijdschr Fysiother trolled study of the outcome of hydrotherapy for
2001; 111(suppl. 3): 1–24. subjects with low back or back and leg pain.
8 Burton AK, Balagué F, Cardon G et al. Physiotherapy 1998; 84: 17–26.
European guidelines for the prevention of low back 22 Kihlstrand M, Stenman B, Nilsson S, Axelsson O.
pain 2004. Accessed 7 July 2007, from: http:// Water-gymnastics reduced the intensity of back/
www.backpaineurope.org/web/files/ low back pain in pregnant women. Acta Obstet
WG3_Guidelines.pdf Gynecol Scand 1999; 78: 180–85.
9 Choi BKL, Verbeek JH, Jiang Y, Tang JL. 23 Schrepfer R, Fritz J. A comparison of change in
Exercises for prevention of recurrences of low-back visual analogue pain rating of acute low back pain
pain. (Protocol). Cochrane Database Syst Rev 2007; patients following deep water walking or deep
Issue 2. Art. No.: CD006555. DOI: 10.1002/ water hanging. J Aquatic Phys Ther 2000; 8: 25–28.
14651858.CD006555. 24 Saggini R, Cancelli F, Di Bonaventura V,
10 van Tulder M, Becker A, Beckering T et al. Bellomo RG, Pezzatini A, Carniel R. Efficacy of
European Guidelines for the management of acute two micro-gravitational protocols to treat chronic
non-specific low back pain in primary care. low back pain associated with discal lesions: a
Accessed 12.07.2007 http://www.backpaineuro- randomized controlled trial. Eur Medicophys 2004;
pe.org/web/files/WG1_Guidelines.pdf 40: 311–16.
11 Kristiansson P, Svärdsudd K, von Schoultz B. 25 Yozbatiran N, Yildirim Y, Parlak B. Effects of
Back pain during pregnancy. Spine 1996; 21: fitness and aquafitness exercises on physical fitness
702–709. in patients with chronic low back pain. Pain Clin
12 Morgren IM, Pohjanen AI. Low back pain and 2004; 16: 35–42.
pelvic pain in pregnancy. Spine 2005; 30: 983–91. 26 Granath AB, Hellgren MSE, Gunnarsson R. Water
13 Stuge B, Hilde G, Vøllestad N. Physical therapy for aerobic reduces sick leave due to low back pain
pregnancy-related back and pelvic pain: a sys- during pregnancy. J Obstet Gynecol Neonatal Nurs
tematic review. Acta Obstet Gynecol Scand 2003; 2006; 35: 465–71.
82: 983–90. 27 Friedrich M, Gittler G, Arendasy M,
14 Cole MD, Becker BE (eds). Comprehensive aquatic Friedrich KM. Long-term effect of a combined
therapy, second edition. Butterworth-Heinemann, exercise and motivational program on the level of
2004. disability of patients with chronic low back pain.
15 Pittler MH, Karagülle MZ, Karagülle M, Ernst E. Spine 2005; 30: 995–1000.
Spa therapy and balneotherapy for treating low 28 Marks R, Allegrante JP. Chronic osteoarthritis and
back pain: meta-analysis of randomized trials. adherence to exercise: a review of the literature. J
Rheumatology 2006; 45: 880–84. Aging Phys Activity 2005; 13: 434–60.
16 PEDro. PEDro scale. 1999. Accessed 17 July 2007, 29 Moher D, Schulz KF, Altman DG. The
from: http://www.pedro.fhs.usyd.edu.au/ CONSORT statement: revised recommendations
scale_item.html for improving the quality of reports of parallel-
17 Maher CG, Sherrington C, Herbert RD, group randomization trials. Ann Intern Med 2001;
Moseley AM, Elkins M. Reliability of the PEDro 134: 657–62.
14 B Waller et al.

Appendix 1 – Keyword and search limits

used

Hydrotherapy Low back pain RCT

Aquatic therapy LBP CCT
Aquatics Back pain
Water therapy Spine pain
Water exercises
Swimming
Aquatic exercise
Limits:
Human
Adult (age 419)
Published in the previous 17 years (1990–2007)
English language
Example of search and number of hits (PUBMED):
Search (‘‘Hydrotherapy’’[Mesh] OR ‘‘Swimming’’[Mesh]) AND ‘‘Low Back Pain’’[Mesh]
Limits:
Publication Date from 1990 to 2007/07/01, Humans, Clinical Trial, Meta-Analysis, Randomized Controlled Trial, All
Adult: 19þ years Hits ¼ 21