Context.— Low back pain is a frequent and costly health problem. Prevention of
low back pain is important both for the individual patient and from an economic
perspective.
Objective.— To assess the efficacy of lumbar supports and education in the prevention
of low back pain in industry.
Design.— A randomized controlled trial with a factorial design.
Setting.— The cargo department of an airline company in the Netherlands.
Participants.— A total of 312 workers were randomized, of whom 282 were available for
the 6-month follow-up.
Interventions.— Subjects were randomly assigned to 4 groups: (1) education (lifting
instructions) and lumbar support, (2) education, (3) lumbar support, and (4)
no intervention. Education consisted of 3 group sessions on lifting techniques
with a total duration of 5 hours. Lumbar supports were recommended to be used
during working hours for 6 months.
Main Outcome Measures.— Low back pain incidence and sick leave because of back pain during the
6-month intervention period.
Results.— Compliance with wearing the lumbar support at least half the time was
43%. In the 282 subjects for whom data were available, no statistically significant
differences in back pain incidence (48 [36%] of 134 with lumbar support vs
51 [34%] of 148 without, P=.81) or in sick leave
because of low back pain (mean, 0.4 days per month with lumbar support vs
0.4 days without, P=.52) were found among the intervention
groups. In a subgroup of subjects with low back pain at baseline, lumbar supports
reduced the number of days with low back pain per month (median, 1.2 vs 6.5
days per month; P=.03).
Conclusions.— Overall, lumbar supports or education did not lead to a reduction in
low back pain incidence or sick leave. The results of the subgroup analysis
need to be confirmed by future research. Based on our results, the use of
education or lumbar supports cannot be recommended in the prevention of low
back pain in industry.
LOW BACK PAIN occurs frequently and is one of the most costly health
problems affecting industry and society. Lifetime prevalences of 60% to 90%
have been reported1 and the total (direct and
indirect) costs for back pain were estimated to be $27.9 billion in 1990 in
the United States.2 Therefore, it is not surprising
that many measures are available that claim to reduce low back pain and its
recurrence. Aside from ergonomic adjustments in the workplace, the most commonly
used preventive strategies in industry are fitness exercises, education on
back mechanics and lifting techniques, and lumbar supports.3
Although these measures are widely used, their efficacy is still uncertain.
In particular, the efficacy of lumbar supports is under debate. To date,
at least 3 randomized trials are available that evaluate the effect of lumbar
supports on the prevention of back pain in industry. In 2 of these, no effect
of lumbar supports was reported.4,5
The third study found a small reduction in the number of days lost from work
in a group receiving both a lumbar support and education compared with a control
group.6 In addition, 2 nonrandomized controlled
trials reported a positive effect of lumbar supports in the reduction of back
pain incidence.7,8 Review articles
concluded there is insufficient evidence for or against the effectiveness
of lumbar supports in the prevention of low back pain and that further research
is needed.3,9-11
The same conclusion was reached in a report of the US National Institute for
Occupational Safety and Health (NIOSH) and, consequently, the use of lumbar
supports among uninjured workers was not recommended by NIOSH.12
We conducted a randomized controlled trial to determine the effectiveness
of lumbar supports and education in the prevention of back pain in industry
conducted in the Netherlands.
A factorial randomized design was used: group 1 received both a lumbar
support and education in the form of lifting instructions, group 2 received
only education, group 3 received only a lumbar support, and group 4 received
no intervention (control group). In a factorial design, in addition to assessing
the effect of interventions by comparing them with a control group, the combination
of 2 interventions is compared with each intervention alone and with a control
group, thus allowing investigation of the interaction between the 2 interventions.
Workers were recruited from the cargo department of a major Dutch airline
at Schiphol Airport. All workers whose jobs included manual material handling
were invited to participate. Typical tasks of these workers included the loading
and unloading of cargo pallets and containers and the sorting and transportation
of cargo, both manually and with a forklift truck. Workers who had a permanent
partial work disability were excluded from the study. The study was approved
by the Medical Ethical Committee of the Vrije Universiteit, Amsterdam, the
Netherlands. Workers received personal information about the procedures of
the trial and enrolled after giving consent. The duration of the intervention
period was 6 months.
The Work S'port back support (The Saunders Group Inc, Chaska, Minn)
was chosen over 2 other lumbar supports in a pilot study. In the pilot study
the comfort of the different supports was tested in a group of 20 workers
who were not included in the randomized trial. The lumbar support used in
the study has adjustable elastic side pulls with Velcro fasteners and flexible
stays, and is kept in place with an anchor belt. The lumbar support has no
shoulder straps and is available in 4 sizes. The workers were given written
and verbal instructions on the use of the lumbar support and were instructed
to wear the lumbar support at all times during work hours and to tighten the
side pulls for stronger support when performing strenuous tasks.
Education was given in the form of lifting instructions. The lifting
instructions were given by 2 experienced paramedical therapists. Instructions
were designed to make workers aware of their movements and postures during
work. Lifting instructions were given in 3 sessions for groups of 10 to 15
workers; the first session of 2 hours took place at the start of the intervention
period, and the other 2 sessions of 1.5 hours each were given at 6 weeks and
at 12 weeks. In the first session, information was given about the anatomy
of the spine and back muscles and about lifting techniques. Lifting techniques
were practiced in simple situations. In the second session, the instructions
on lifting techniques were repeated, and relaxation exercises for neck, shoulder,
and back muscles were explained and practiced. In the third session, which
took place at the workplace when workers were actually loading and unloading
cargo, individual advice was given on working methods.
The work at the cargo department was organized in work modules (n=6),
with each work module consisting of 6 work groups containing 6 to 20 workers
each. Workers in all work modules performed manual lifting tasks and used
a forklift truck. The work modules differed slightly in the proportion of
time spent using a forklift truck and performing lifting tasks. A total of
380 workers in 36 work groups were eligible for the study. We assumed that
compliance would be higher if all workers in a group, rather than a subset
of the group, were asked to wear a lumbar support. Furthermore, it was practical
to use the existing work-group organization for the education sessions. Therefore,
the 36 work groups, not individual subjects, were randomized over the 4 treatment
groups. Because of the slight variation in tasks among some work modules,
randomization was stratified for work modules.13
For each work module, a separate randomization list was prepared. Random permuted
blocks of 4 were used within each stratum. The blocks were assigned to the
strata using a random numbers table.
Randomization was performed by an investigator who was blinded with
respect to the characteristics of the work modules and the workers within
the work groups. Assignment to the intervention groups took place after completion
of the baseline measurements.
At baseline subjects completed a questionnaire on demographic data,
history of back pain, work perception,14 and
health status.15,16 During the
intervention period, subjects received a monthly questionnaire on the occurrence
of low back pain and sick leave. Subjects were asked if they had experienced
low back pain in the past month and, if they answered positively, how many
days they experienced low back pain. They were also asked if they had lost
time from work in the past month (and the number of days lost) and if they
had lost time from work because of back pain (and the number of days lost).
All subjects worked full time. The same questionnaire on back pain and sick
leave was completed at 9 months and 12 months after randomization.
At baseline and at 6 months, the end of the intervention period, trunk-muscle
strength was measured to assess whether wearing a lumbar support affected
trunk muscle strength. Subjects with current or past back pain who felt they
might injure or reinjure their backs by performing the tests were excluded
from the trunk-muscle tests. The endurance strength of the abdominal muscles
was determined with subjects lying supine with knees at 90° and feet flat
on the floor or table without support.17 Subjects
were asked to curl up with hands straight toward knees and fingertips of both
hands reaching midpatella. This posture was held for a maximum of 240 seconds.
If the posture was lost, the test was stopped and the number of seconds was
noted. The dynamic strength of the abdominal muscles was measured with subjects
lying supine with knees at 90° and feet flat on the floor without support.18 Subjects were asked to perform 3 series of 5 sit-ups
with increasing difficulty. The number of sit-ups performed by each subject
was noted. Endurance strength of the back muscles was determined with subjects
prone on the examination table with buttocks and legs fixed and trunk unsupported.19 Subjects were asked to hold their upper body and
head horizontal for a maximum of 240 seconds. If the posture was lost and
could not be corrected, or if a subject showed signs of exhaustion, pain,
or cramping, the test was stopped and the number of seconds was noted.
Compliance with wearing the lumbar support was measured every month.
In the monthly questionnaire subjects were asked if they had worn the lumbar
support in the previous month. Subjects were considered compliant if they
reported wearing the lumbar support in more than half of the questionnaires.
Based on the difference in days lost from work and corresponding variance
reported by Walsh and Schwartz,6 it was estimated
that with a significance level of .05 and a power of 80%, a difference of
0.9 lost work day per 6 months could be detected with 50 subjects per intervention
group. Our goal was to include 75 subjects per intervention group in order
to prevent inadequate power because of low compliance or withdrawals.
Differences in outcomes at the 6-month follow-up were analyzed for the
4 intervention groups. Differences between groups were tested for statistical
significance by using χ2 tests for categorical data (ie, experience
of back pain and sick leave) and Student t tests
for continuous data (ie, age, scores on trunk muscle tests). Nonparametric
testing (the Mann-Whitney U test or the Kruskal-Wallis
test, corrected for ties) was used for data on the number of days with back
pain or sick leave, since the distribution of these data is skewed. Differences
were considered statistically significant at the .05 level. For categorical
data, differences between groups and 95% confidence intervals (CIs) were calculated.20
Possible effect modification in the group with both lumbar support and
education was studied in an analysis in which the 2 groups with a single intervention
were combined and compared with the group that received both interventions.
The control group was removed from this analysis. If effect modification were
present, the groups would differ significantly. If no effect modification
were present in the group with the combination of lumbar support and lifting
instructions, groups would be combined as follows to increase the number of
subjects in the compared groups. The 2 groups with lumbar supports would be
compared with the 2 groups without lumbar supports, and a similar comparison
would be made for education.
For the effect of lumbar supports, 2 types of analyses were performed:
an analysis of the intervention group in which all subjects for whom data
were available were included, regardless of compliance with wearing the lumbar
support; and 2 subgroup analyses assessing only subjects who were compliant
with lumbar support use and whether subjects had a history of low back pain
at baseline.
Participant Flow and Follow-up
Of 380 eligible workers, 20 (5%) refused to participate and 312 (82%)
completed the baseline measurements (Figure
1). Because of holidays and a large workload, workers were not always
available for the study. Consequently, baseline measurements were missing
for 48 workers. During the intervention period of 6 months, 30 workers (10%)
withdrew from the study. Twenty-one workers (7%) withdrew because they lost
interest in the study after the baseline measurement (mostly because workers
did not think that the study was useful to them). Nine workers (3%) transferred
to another workplace or left the company. Another 14 workers (4%) transferred
to other workplaces in the second 6 months of the follow-up period. Only 1
of the 23 workers who transferred to another workplace left the cargo department
because of low back pain. The distribution of the withdrawals among the intervention
groups is shown in Figure 1. A description
of the study population is given in Table
1. There were no substantial differences among the 4 intervention
groups regarding the most important prognostic factors, such as age, history
of back pain, and past sick leave because of back pain. Work modules were
evenly distributed among the intervention groups; for all modules, about half
of the workers were given education and half received a lumbar support.
At baseline, 243 subjects were asked about their expectations of preventive
measures. Of these, 146 (60%) expected that lumbar supports could be helpful
in the reduction of low back pain in the workplace, and 170 (69%) felt lifting
instructions could be useful. The attendance rate for the first educational
session was 80%, and all workers attended at least 2 of the 3 sessions. Subjects
were asked each month if, in their own opinion, they lifted according to the
techniques taught in the educational sessions. Of 142 subjects, 16 (11%) answered
that they always lifted as taught, 104 (73%) answered that they lifted as
taught some of the time, and 15 (11%) answered that they never lifted as taught.
Compliance with wearing the lumbar support was low; only 58 (43%) of the 134
subjects in the lumbar support groups reported wearing the support in more
than half of the questionnaires. In random checks by the principal investigator
at the workplace, compliance was approximately the same as the subjects reported
(40%-50%). When subjects were asked how satisfied they were with the lumbar
support, 39 (49%) of 79 subjects reported the support restricted their freedom
of movement, 39 (48%) of 81 reported they could not sit comfortably with the
support, and 36 (45%) of 80 thought the support was too warm. On the other
hand, 49 (62%) of 79 thought the support provided support for the back and
37 (46%) of 80 reported the support was easy to use.
No data were available for 30 subjects (Figure 1, subjects withdrawn during intervention). Of the remaining
282 subjects, 99 (35%) experienced an episode of low back pain during the
intervention period and 29 (10%) reported having taken sick leave because
of low back pain
(Table 2). No
statistically significant differences were found among the 4 intervention
groups. No effect modification could be detected when comparing the group
with lumbar support and education with the other 2 intervention groups combined.
Therefore, the groups receiving lumbar supports were combined, as were those
receiving education. Table 2 shows
the main outcomes for the groups. No statistically significant differences
were present. Low back pain incidence was about 35% in all groups, and subjects
reporting sick leave due to low back pain varied from 8% to 13%.
A subgroup analysis was conducted in which subjects who wore the lumbar
support were compared with subjects in the groups without lumbar support.
No significant differences were observed (Table 3).
Table 4 shows the results
of the analysis of subgroups based on the history of back pain. In the subgroup
of subjects who had never had low back pain (n=130), a higher number of days
of sick leave because of low back pain was observed in the group with lumbar
support (P=.05), although the median number of days
of sick leave per month was 0 days in both groups. In the subgroup of subjects
with low back pain at baseline (n=42), a reduction in the number of days with
low back pain per month was found in the group with lumbar support compared
with the group without lumbar support (median of 1.2 days per month vs 6.5
days per month; P=.03). No significant differences
were found between groups with and without education in the subgroup analysis
(data not shown).
To detect possible adverse effects of lumbar supports, we measured trunk-muscle
strength before and after the intervention period. Furthermore, after the
intervention period, subjects were followed up for an additional 6 months
to evaluate the occurrence of low back pain after subjects discontinued wearing
the lumbar supports. In the group compliant with lumbar support use during
the intervention period, 25% of the subjects reported low back pain during
the 6 months following completion of the intervention vs 20% in the group
without lumbar support (risk difference=5%; 95% CI, −13% to 23%; P=.6). Compliant subjects had a mean of 1.8 days of back
pain vs 1.1 days in the group that did not receive the lumbar support (median
of 0 days in both groups; P=.5 by Mann-Whitney U test corrected for ties). Results of the trunk-muscle
strength tests are shown in Table 5.
Thirty-eight subjects felt they would injure or reinjure themselves by performing
the trunk-muscle tests and were excluded from the tests (10 in the combination
group, 5 in the education only group, 12 in the lumbar support only group,
and 11 in the control group). Most subjects scored lower in the posttest,
but the difference between compliant subjects and subjects without lumbar
support was not significant.
Effect of Lumbar Supports and Education
No effect of education alone in the prevention of low back pain could
be demonstrated in this study. This result confirms previous controlled trials,4,6,21-24
all of which found no effect of providing lifting instructions or instructions
on body mechanics. The only controlled study that reported a positive effect
of education was a study in a population of bus drivers who received no lifting
instructions but received a back school program consisting of information
on back care, physical fitness, nutrition, relaxation, and coping with stress.25
Lumbar supports were not effective in the prevention of low back pain
and sick leave in this study. Two other randomized studies have reported the
same findings.4,5 Some nonrandomized
studies reported positive findings,7,8,13
but these studies are more susceptible to various forms of bias. A recently
published study by Kraus et al26 that included
36000 subjects reported a reduction in acute low back injury rate. However,
because of the nonrandomized, noncontrolled design of the trial, the findings
may be explained by confounding factors, such as changes in exposure or workers'
compensation laws.
In our study, the combination of lumbar support and education also was
not effective. Two other studies that combined these interventions showed
conflicting results. Walsh and Schwartz6 reported
a reduction in the number of days with sick leave, while Reddell et al4 found no effect. The cause of the contradictory results
is unknown, but factors such as work characteristics, differences in lumbar
support (the study by Walsh and Schwartz6 used
a rigid plastic support), and compliance rates could be important.
The study by Reddell et al4 had a compliance
rate nearly identical to our study (42% and 43%, respectively). The only other
study reporting compliance rate is the study by Anderson et al,7
in which the compliance was 80% according to the supervisors at the workplace.
Although our analysis of only subjects who were compliant failed to show a
positive effect of lumbar supports, it is possible that because of self-selection
of compliant subjects, an effect of lumbar supports may have been missed.
However, in the Netherlands an employer would be unlikely to require use of
lumbar supports or impose sanctions in the case of noncompliance. Therefore,
the results presented here represent what can be expected if an employer provides
but does not require lumbar supports for manual material–handling workers.
Subgroup analysis is difficult because it is potentially misleading,
even if the analysis is prospectively planned.27,28
Yusuf et al27 state that they "regard observed
qualitative interactions (treatment is beneficial in one subgroup and harmful
in another) with considerable skepticism, for they are often shown to be spurious
when the same comparison is made in similar trials." Yusef et al27
recommend that results from subgroup analyses should not be accepted until
they are confirmed by the results of other studies.
The present study demonstrates a qualitative interaction. On one hand,
the use of a lumbar support seemed to increase sick leave due to low back
pain in subjects who had never had low back pain before the start of the intervention
(n=130). On the other hand, in a small group of subjects who had low back
pain at the start of the intervention (n=42), the use of lumbar supports reduced
the number of days per month with low back pain.
Among workers without a history of low back pain, the differences between
groups were small (0.6 vs 0.2 mean days per month of sick leave) and perhaps
not clinically relevant, and because we did not correct for multiple comparisons,
differences may be due to chance. Therefore, we draw no conclusions from this
analysis but believe studies should investigate the effect of lumbar supports
for this subgroup in more detail.
Walsh and Schwartz6 reported a larger
effect of the combination of lumbar supports and education in a subgroup of
workers who had low back pain in the 6 months prior to the study, while we
found no effects in a group of subjects who had low back pain in the year
prior to the study (n=69).
A review on the effectiveness of orthoses in the treatment of back pain
concluded that the therapeutic effect of lumbar supports for subjects with
back pain has not yet been demonstrated, although some promising findings
were reported in the literature.29 This is
consistent with our finding of a positive effect of lumbar supports in the
subgroup of workers with low back pain at baseline. Nevertheless, future randomized
trials on the effects of lumbar supports for patients with low back pain are
needed to determine their therapeutic value, if any.
No effects of lumbar supports on trunk-muscle strength were observed.
Other studies also reported no adverse effects of lumbar supports on abdominal
and back-muscle strength.6,30
These results indicate the use of lumbar supports will not cause atrophy of
trunk muscles.
Reddell et al4 reported an increase in
incidence of sick leave because of back injury in groups of workers who stopped
wearing a lumbar support during the intervention period. We could not reproduce
these findings, as no increase in the incidence of back pain was found after
compliant workers discontinued wearing lumbar supports.
The major limitation of this study is the lack of blinding; neither
the subjects nor the therapists who conducted the educational sessions were
blinded for the intervention. The direction and magnitude of the potential
bias is not clear. It could lead to a larger estimate of the effect of the
intervention,31 or perhaps to a smaller estimate,
depending on the expectations of subjects and investigators. However, at baseline
most workers believed both interventions would be beneficial, so any potential
bias from expectations would be in the positive direction.
The subjects in our study all had very similar work tasks of loading
and unloading cargo, including heavy-lifting tasks. Extrapolation of the results
is most relevant for situations in which workers perform similar lifting tasks.
Based on the results of our study, we do not recommend education (in
the form as investigated in our study) or the use of lumbar supports in the
prevention of low back pain. The therapeutic effectiveness of lumbar supports
for workers with low back pain in industry needs further investigation.
1.Andersson GBJ. The epidemiology of spinal disorders. In: Frymoyer JW, ed. The Adult Spine: Principles and Practice. New York, NY: Raven Press; 1991:107-146.
2.Frymoyer JW, Cats-Baril WL. An overview of the incidences and costs for low back pain.
Orthop Clin North Am.1991;22:263-271.Google Scholar 3.Lahad A, Malter AD, Berg AO, Deyo RA. The effectiveness of four interventions for the prevention of low back
pain.
JAMA.1994;272:1286-1291.Google Scholar 4.Reddell CR, Congleton JJ, Huchingson RD, Montgomery JF. An evaluation of a weightlifting belt and back injury prevention training
class for airline baggage handlers.
Appl Ergonomics.1992;23:319-329.Google Scholar 5.Alexander A, Woolley SM, Bisesi M, Schaub E. The effectiveness of back belts on occupational back injuries and worker
perception.
Professional Safety.1995;40:22-26.Google Scholar 6.Walsh NE, Schwartz RK. The influence of prophylactic orthoses on abdominal strength and low
back injury in the workplace.
Am J Phys Med Rehabil.1990;69:245-250.Google Scholar 7.Anderson CK, Morris TL, Vechio DC. The Effectiveness of Using a Lumbar Support Belt. Dallas, Tex: Advanced Ergonomics Inc; 1993.
8.Thompson L, Pati AB, Davidson H, Hirsh D. Attitudes and back belts in the workplace.
Work.1994;4:22-27.Google Scholar 9.Barron BA, Feuerstein M. Industrial back belts and low back pain: mechanisms and outcomes.
J Occup Rehabil.1994;4:125-139.Google Scholar 10.Genaidy AM, Simmons RJ, Christensen DM. Can back supports relieve the load on the lumbar spine for employees
engaged in industrial operations?
Ergonomics.1995;38:996-1010.Google Scholar 11.van Poppel MNM, Koes BW, Smid T, Bouter LM. A systematic review of controlled clinical trials on the prevention
of back pain in industry.
Occup Environ Med.1997;54:841-847.Google Scholar 12.National Institute for Occupational Safety and Health Back Belt Working
Group. Workplace Use of Back Belts: Review and Recommendations. Washington, DC: US Dept of Health and Human Services, Public Health
Service, Centers for Disease Control and Prevention; 1994. DHHS (NIOSH) publication
94-122.
13.Pocock SJ. Clinical Trials: A Practical Approach. New York, NY: John Wiley & Sons Inc; 1983:83-87.
14.Dijkstra A, van der Grinten MP, Schlatmann MJT, de Winter CR. Funktioneren in de Arbeidssituatie: Uitgangspunten, Ontwerp en
Handleiding Voor Onderzoek Onder Werknemers Naar Gezondheid, Werk en Werkomstandigheden. Leiden, the Netherlands: NIPG-TNO; 1981.
15.Larson CO, Hays RD, Nelson EC. Do the pictures influence scores on the Dartmouth COOP Charts?
Qual Life Res.1992;1:247-249.Google Scholar 16.Scholten JHG, van Weel C. Functional Status Assessment in Family Practice: The Dartmouth
COOP Functional Health Assessment Charts/WONCA. Lelystad, the Netherlands: Meditekst; 1992.
17.Hyytiäinen K, Salminen JJ, Suvitie T, Wickström G, Pentti J. Reproducibility of nine tests to measure spinal mobility and trunk
muscle strength.
Scand J Rehabil Med.1991;23:3-10.Google Scholar 18.Oja P, Tuxworth B. Eurofit for Adults: Assessment of Health-Related Fitness. Tampere, Finland: Council of Europe; 1995.
19.Biering-Sørensen F. Physical measurements as risk indicators for low-back trouble over
a one-year period.
Spine.1984;9:106-119.Google Scholar 20.Altman DG. Practical Statistics for Medical Research. London, England: Chapman & Hall; 1991.
21.Donchin M, Woolf O, Kaplan L, Floman Y. Secondary prevention of low-back pain: a clinical trial.
Spine.1990;15:1317-1320.Google Scholar 22.Daltroy LH, Iversen MD, Larson MG.
et al. Teaching and social support: effects on knowledge, attitudes, and behaviours
to prevent low back injuries in industry.
Health Educ Q.1993;20:43-62.Google Scholar 23.Daltroy LH, Iversen MD, Larson MG.
et al. A controlled trial of an educational program to prevent low back injuries.
N Engl J Med.1997;337:322-328.Google Scholar 24.Feldstein A, Valanis B, Vollmer W, Stevens N, Overton C. The back injury prevention project pilot study: assessing the effectiveness
of Back Attack, an injury prevention program among nurses, aides, and orderlies.
J Occup Med.1993;35:114-120.Google Scholar 25.Versloot JM, Rozeman A, van Son AM, van Akkerveeken PF. The cost-effectiveness of a back school program in industry: a longitudinal
controlled field study.
Spine.1992;17:22-27.Google Scholar 26.Kraus JF, Brown KA, McArthur DL.
et al. Reduction of acute low back injuries by use of back supports.
Int J Occup Environ Health.1996;2:264-273.Google Scholar 27.Yusuf S, Wittes J, Probstfield J, Tyroler HA. Analysis and interpretation of treatment effects in subgroups of patients
in randomized clinical trials.
JAMA.1991;266:93-98.Google Scholar 28.Oxman AD, Guyatt GH. A consumer's guide to subgroup analyses.
Ann Intern Med.1992;116:78-84.Google Scholar 29.Koes BW, van den Hoogen HMM. Efficacy of bed rest and orthoses of low-back pain: a review of randomized
clinical trials.
Eur J Phys Med Rehabil.1994;4:86-93.Google Scholar 30.Holmström E, Moritz U. Effects of lumbar belts on trunk muscle strength and endurance: a follow-up
study of construction workers.
J Spinal Disord.1992;5:260-266.Google Scholar 31.Schulz KF, Grimes DA, Altman DG, Hayes RJ. Blinding and exclusions after allocation in randomised controlled trials;
survey of published parallel group trials in obstetrics and gynaecology.
BMJ.1996;312:742-744.Google Scholar