Efficacy of Back Belts for
Prevention of Back Injuries in Material Handling Workers
Lytt I.
Gardner,
James W.
Collins,
Janet J. Johnston,
and James T. Wassell
Division
of Safety Research, National Institute for Occupational Safety
and Health
1095 Willowdale Road
Morgantown, West Virginia 26505
ABSTRACT
Back
injuries typically make up 20 to 25% of all injuries reported to
workers compensation, and the majority of these injuries are
related to material handling. The purpose of the National
Institute for Occupational Safety and Health (NIOSH) study of the
efficacy of back belts is to determine whether a standard
industrial-style back support belt is effective in reducing the
incidence of first or recurrent back injuries or pain in manual
material handling workers. The study began collecting data in
October 1996 and will continue until January 1999. Approximately
8000 hourly workers employed in retail merchandise stores in the
eastern United States will participate. Approximately 2000
workers have been interviewed to date.
This
prospective study of a dynamic employed cohort follows workers in
two types of stores: stores where back belt wearing is required,
and stores where back belt wearing is not required. The study is
powered to detect a difference of 30% in injury rates between the
two types of stores. The primary outcome variables are
company-reported back injuries requiring medical care or days
away from work, plus interview reports of a back problem
requiring medical care. Telephone interviews will be conducted at
6-month intervals to obtain risk factor information on previous
employment, previous back-injury history, back support
belt-wearing behavior, the psychosocial work environment,
self-rated physical fitness level, work exposure-related
activities, off-work lifting activities, and body mass index. The
wearing of back supporting belts is determined at the store level
by the belt policy, but will also be characterized at the
individual level using responses to belt wearing questions. The
validity of self-reports of belt wearing is being addressed in a
sample of stores by directly observing belt wearing of
individuals during the same time frame of telephone interviewing.
A job
analysis will be conducted in a sample of 48 of the 160 stores
(24 with belts required and 24 without belts required) to
characterize the work exposures in the relevant jobs. This subset
sample will be used to control for confounding of belt effects on
injury due to differences in work exposure.
The study
will advance understanding of the efficacy of back supporting
belts by: (1) evaluating the prevention of initial and
recurrent back problems; (2) using interviews to give a more
complete count of injuries independent of worker compensation
issues; (3) collecting belt wearing data on individuals, in
addition to store policy data; (4) eliminating confounding of
belt effects with changes in work practices over time; and (5)
evaluating the biological plausibility of reductions in injuries
which occur to locations other than the back.
INTRODUCTION
This
presentation concerns an intervention whose effectiveness is
currently being evaluated at NIOSH ,the use of flexible elastic back belts to
prevent back injury in material handling workers. NIOSH published
a recommendation in 1994 that the effectiveness of back belts
remains unproven, and that it does not consider back belts to be
personal protective equipment based on a review of the existing
scientific literature (1). However, at that time we also
expressed the need for additional research to evaluate the
efficacy of back belts to prevent work-related back injury, and
this project grows out of that need for additional research. The
study is now ongoing and will be completed in about 21 months.
NIOSH and
the Division of Safety Research have been working since 1994 to
design, obtain approval, and acquire the funds to execute an
epidemiologic study to determine the effect of back belts use on
back pain or injury in a cohort of material handling workers. In
this study, we will determine the efficacy of an industrial style
belt to prevent back problems in a cohort of 8,000 to 10,000
material handling workers. These workers are all with a single
large chain of retail merchandise stores, and they will be
enrolled for approximately 18 months in 160 stores. Each worker
will be followed for a period of 6 to 24 months, depending on
when they were enrolled. Enrollment began in October 1996 and
will continue through April 1998. Follow-up will conclude around
January 1999.
METHODS
This study
involves a periodic telephone interview every 6 months, which
collects data on the occurrence of back injury and pain, history
of injuries, work-related risk factors, demographic factors,
personal fitness related factors, and work organization or
psychosocial factors.
Outcome
data
There are
two outcomes which will be used to determine the results of the
study: primary outcome and secondary outcome.
Primary outcome:
These both exclude
slips, falls, struck-by, and assault incidents.
Secondary outcome:
Pain reports:
- from a
body part diagram with a 0 to 10 scale
- National
Center for Health Statistics 1988
occupational health survey questions on back
pain
Activity
limitation questions:
- McGill
University work interference questions
- Quebec
Back Pain Disability Scale
Store
types
The study
will enroll workers from two types of stores. Workers from about
80 stores will have voluntary belt wearing; in these stores,
workers are told that the use of back belts is an individual
choice and they are not required to wear back belts. Workers from
another approximately 80 stores will have required belt wearing;
in these stores workers are told they must wear back belts.
From the
first 8 stores with voluntary belt wearing, the rate of back belt
wearing ranged from about 15% to about 75%, with an average of
40%. In the required belt wearing stores, the average belt
wearing rate was 80% and ranged from about 50% to about 99%. The
study was designed to be able to detect a difference of 30% in
the rates of (primary outcome) back problems requiring medical
care between voluntary and required belt stores. In some
analyses, the belt variable will be based on the individual's report of belt wearing; in other
analyses the belt variable will be based on the store policy.
Regardless of which way the belt variable is handled, the unit of
analysis is always the individual, and the power calculation is
based on individual injuries and person time.
An interim
analysis of the data is scheduled for December of 1997. This
analysis is in the tradition of an interim analysis in a clinical
trial to deal with the ethical concerns of failing to note a very
big effect of the belts, which would require us to take action
before the end of the study. The final analysis will take place
following the last data collection, which should take place about
January 1999. We anticipate a final analysis report to be made
public in the early Spring 1999.
Status
of Data Collection
As of
May 30, 1997, the NIOSH Back Injury Study had:
Telephone interview
completion or response rate is averaging 90%.
Approximately 2000
workers have been interviewed to date.
Monthly company
payroll tapes have been received covering the interview
period to date. These tapes will give us hours of work for
each worker, based on date of initial employment and date of
last pay record.
Exposure data
Work exposure data :
- A job
analysis using a modified Postures, Activities,
Tools, and Handling (PATH) timed sampling method
is being used in 48 of the stores to characterize
exposures to physical work (3). This subset
sample will be used to calculate an objective
measure of work exposure to supplement the
subjective measure of work exposure contained in
the interview self-report.
- A second
subjective measure of work exposure is the Borg
scale of perceived exertion(4). This will be a
second subjective measure of exposure which will
give the worker's estimate of the
level of exertion experienced in the work.
Back belt data:
- Self-report
from telephone interviews
- Direct
observation in 48 stores
- Direct
observation in 8 stores where timing of the
observation coincides exactly with the telephone
interviewing.
The direct
observation data will be used to verify self-reported belt
exposure data. The question best addressed by the 8 stores is, do
belt wearing reports over-estimate true belt wearing rates, and
to what extent? If the belt wearing reports are over-estimating
true rates of belt wearing, then one of the available statistical
methods will have to be used to estimate the unbiased belt
effect.
DISCUSSION
Table 1
shows that the NIOSH study has elements of both a community
intervention experiment and a more traditional epidemiological
prospective cohort study. In analyses where the allocation to
treatment groups (voluntary and required belts) is the focus, we
will have strong evidence concerning the belt effect, although it
will be a conservative estimate because the compliance with belt
wearing is not 100% in the required group, nor is compliance to
no belts 100% in the voluntary group. This type analysis is known
as intention to treat, or alternatively, as the analysis of the
effectiveness of a community intervention.
In other
analyses the focus will be belt wearing by individuals, and the
store type fades into the background as a covariate. In this type
analysis, which treats the individuals as members of a
prospective cohort , we are concerned with controlling for other
causes of back problems which could confound the association with
back belts. Table 2 summarizes the important elements of the
NIOSH study design, and lists the risk factors which must be
considered in an analysis where belt wearing as reported in the
telephone interview is the exposure variable.
The NIOSH
study is prospective with respect to assessing the effects of
back belts. In this study, individuals wearing and not wearing
back belts are fairly closely matched in calendar time; in that
sense the belt effects are not confounded by calendar time. This
design feature is important because the alternative is a
historical design where belt wearing and non-wearing data occur
in different calendar time periods. The latter design has the
inherent weakness in that not all pertinent information about the
handling of freight can be reconstructed from historical records,
presenting difficulties in interpretation of changes in injury
rates which occur over time.
The
biological plausibility of reductions in injury rates is also a
concern in a back belt study. If back belts are associated with a
reduction in injuries, we would expect the reduction to be for
back injuries. If non-back injuries in belt wearers also drop by
a similar amount, then we have to analyze very carefully these
other body parts, to determine if there is or is not a plausible
relationship to belt wearing.
History of
back injury is also an important factor in understanding the
significance of the effects of back belt wearing. If the effects
of belts are mainly seen in individuals with a previous history
of a back injury, then the effects would concern mainly
prevention or delay of recurrences. This type analysis would help
distinguish therapeutic effects from primary prevention effects.
Finally, no
single study will provide all the evidence needed to ultimately
resolve the question of the benefits and risks of back belt
wearing in material handling workers. The NIOSH study has
elements of its data which are comparable to elements of studies
being conducted at the McGill University in Canada and the Vrije
University, the Netherlands. These comparable data, which concern
questions on belt wearing and pain and activity limitation, will
permit more fruitful pre-planned meta-analyses and pooled
analyses.
Table 1. Study Designs in
Back Belt Studies
TYPE
|
Experimental
|
Quasi-Expt=l
|
Observational
|
Observational
|
Sub-Type
|
individual intervention
|
community intervention
|
prospective cohort
|
historical cohort
|
Treatment Allocation
|
randomized treatment
groups
|
voluntarily formed
treatment groups
|
N/A
|
N/A
|
Strength of Inference Cbelt effect
|
++++
|
+++
|
++
|
+
|
Table 2. NIOSH Study
Design Summary
Study Design
|
prospective cohort with
belt wearing not confounded by time
|
Belt Wearing Status
|
observed in 8 stores
|
| reported on interview |
| observed in 48 stores |
Back Injuries/Pain
|
workers compensation
reports
|
| interview reports of work-related
injury repeated at 6 month intervals |
| interview report of pain and
activity limits |
Non-back Injuries
|
workers compensation
reports
|
| assess reporting bias and
biological plausibility |
Work Exposure Measures
|
self-report of tasks
|
| on site - 48 stores |
| job titles |
History of Injury
|
by interview |
record linkage in
non-interviewed
|
Risk factors/confounders
|
length of employment
|
| age, sex, race |
| psychosocial factors |
| body mass index |
| physical activity history |
| off-job lifting history |
| hours on 2nd jobs |
REFERENCES
1. NIOSH
Back Belt Working Group. Workplace Use of Back Belts. Review and
Recommendations. U.S. Department of Health and Human Services,
Public Health Service, Centers for Disease Control and
Prevention, National Institute for Occupational Safety and
Health, July, 1994. Publication number DHHS (NIOSH) 94-122.
2. Kopec,
J.A., J.M. Esdaile, M. Abrahamowicz, L. Abenhaim, S.
Wood-Dauphinee, D.L. Lamping and J.I. Williams: The Quebec back
pain disability scale. Measurement properties. Spine 20:
341-352 (1995).
3.
Buchholz, B, V. Paquet, L. Punnett, D. Lee and S. Moir: PATH: A
work sampling-based approach to ergonomic job analysis for
construction and other non-repetitive work. Applied Ergonomics
27: 177-187, 1996.
4. Borg, G:
Perceived exertion as an indicator of somatic stress. Scandinavian
Journal of Rehabilitation Medicine 2: 92-98, 1970.
