Spine (Phila Pa 1976) 2004 (Jun 15); 29 (12): 1346–1351
Eric Bernstein, MD, MPH, Timothy S. Carey, MD, MPH, and Joanne Mills Garrett, PhD
Cecil G. Sheps Center for Health Services Research,
University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina 27599-7590, USA
STUDY DESIGN: Prospective cohort study.
OBJECTIVES: To determine the characteristics of patients who take muscle relaxants for back pain after seeking care and to determine the relationship of muscle relaxant use with recovery from the episode of low back pain.
SUMMARY OF BACKGROUND DATA: Low back pain is a common condition with a generally favorable short-term prognosis. Physicians in the acute setting commonly prescribe muscle relaxants. The indications for use and outcomes are not clear.
METHODS: We performed a secondary data analysis of a cohort of 1633 patients who sought care from a variety of practitioners (primary care, physician of chiropractic, orthopedic surgeon, Health Maintenance Organization) for low back pain. Patients were enrolled in the physician's office and interviewed at baseline, 2, 4, 8, 12, and 24 weeks. Pain, functional status, medication use, health care utilization, and satisfaction with care were assessed.
RESULTS: Muscle relaxants were used by 49% of patients; among those who sought care from doctors, 64% used muscle relaxants. Muscle relaxant users were more impaired at baseline. Over time, among patients with greater functional status impairment (Roland disability score > 12) at baseline, muscle relaxant users had somewhat slower recovery from the episode of back pain. This finding persisted after controlling for baseline functional status, age, worker's compensation status, and use of nonsteroidal inflammatory agents.
CONCLUSIONS: Use of muscle relaxants was very common among patients with acute low back pain. Muscle relaxant use was not associated with more rapid functional recovery.
Key words: back pain, cohort study, muscle
From the FULL TEXT Article:
Acute low back pain (LBP) is a common and disabling
condition. For people under the age of 45, back pain is
the most common cause of activity limitation.  Approximately
1% of the population is chronically disabled due
to back problems, and another 1% is temporarily disabled.  Although most individuals with acute back pain
manage the problem without availing themselves of the
care of health professionals, over a third of those afflicted
with acute LBP do seek care from a health care provider,
most commonly amedical doctor (MD) (allopathic) provider.  This accounts for 15 million physician visits per
year and an estimated cost of $192 million in 1990. 
Low back problems are the second most common cause
for office visits to primary care physicians, and back pain
is the most common reason for office visits to orthopedic
surgeons, neurosurgeons, and occupational medicine
physicians.  It ranks third among indications for surgery.  The content of the care provided by these allopathic
physicians consists of evaluation, reassurance, advice
regarding activities, physical methods, medications,
The most common medications prescribed by MDs
are nonsteroidal anti-inflammatory drugs (NSAIDs),
muscle relaxants, and narcotic analgesics. A recent observational
study in a managed care setting showed that
of patients treated for back pain 69% were prescribed
NSAIDs, 35% muscle relaxants, and 12% narcotics. 
The use of muscle relaxants has been examined in multiple
randomized controlled trials, but their utilization
and effectiveness in community practice has been relatively
The role of muscle spasm in the pathogenesis of back
pain and of muscle relaxants in its treatment have been a
subject of dispute. Multiple agents are currently used for
muscle spasm, including cyclobenzaprine, tizanidine,
dantrolene, carisoprodol, baclofen, and benzodiazepines.
Studies examining the use of these medications
have generally been in efficacy settings: randomized trials
comparing the study medication as a single agent compared
with placebo. In general, these agents have been
considered modestly efficacious, although there is considerable
variability in individual studies designed to
demonstrate their effectiveness. This may be due, at least
in part, to the wide variation in the tests, treatments, and
systems of quantifying back pain. This has resulted in a
correspondingly diverse literature regarding back pain
and the use of systemic muscle relaxants.  Systematic
review of this literature is essential to understanding its
impact on current and future treatment practices.
The comprehensive 1994 Agency for Health Care Policy
and Research (AHCPR) review of treatments for LBP
examined the use of muscle relaxants and considered
them an acceptable option in the treatment of acute LBP.
These guidelines reported that muscle relaxants were
probably more effective than placebo, but that they have
not been shown to be more effective than NSAIDs and
that no additional benefit is gained by using muscle relaxants
in conjunction with NSAIDs.  This statement
was supported by limited research-based evidence.
A systematic review of randomized clinical trials
(RCTs) published by van Tulder et al found “strong evidence”
showing muscle relaxants were more effective
than placebo and that different types of muscle relaxants
are equally effective for acute LBP.  This review did not
address the effectiveness of muscle relaxants compared
to, or in conjunction with, NSAIDs. The frequent use of
muscle relaxants and the relatively modest strength of
recommendation from the AHCPR guideline point to the
need for further evidence regarding the utility of these
medications in everyday clinical practice. Muscle relaxants
can be sedating, may increase fall risk, and impair
the ability to drive automobiles or operate machinery.
There is also some concern relating to possible risk of
dependency for benzodiazepine medications. Incremental
effectiveness when other medications and treatment
types are used remains unclear.
We used a previously examined large cohort study,
the North Carolina Back Pain Project, to examine:
Our study assesses the effectiveness of muscle relaxants
in treating LBP in a generalizable sample of patients
with acute back pain. The results of this survival analysis
also raises questions regarding possible delay of functional
recovery related to the use of muscle relaxants in
this population. This will provide useful information in
addressing the question of whether oral muscle relaxants,
alone or in conjunction with NSAIDs, should be
used to treat LBP.
Materials and Methods
The North Carolina Back Pain Project is a collaborative effort
between the Sheps Center for Health Services Research and the
Departments of Medicine, Family Medicine, Biostatistics, and
Business at the University of North Carolina. The data collected
from this project already have provided information regarding
outcomes of acute back pain and its relation to different
types of practitioners,  provider self-confidence,13 and
worker’s compensation.  Additionally, the data have been used
to better understand the utilization of diagnostic tests and
physical therapy in the diagnosis and treatment of acute LBP. 
We have re-examined these data in an effort to better understand
the relationship between muscle relaxant use and functional
recovery from acute LBP.
Methods for data collection in this study have been previously
described by Carey et al and are outlined briefly below. 
The overall intent of the study was to examine the outcomes of
processes of care for community-based individuals with acute
LBP. We wished to enroll patients at the onset of care seeking
and followed them with periodic telephone interviews for almost
2 years. We observed care as delivered and made no attempt
to influence practice patterns. Practitioners were randomly
selected from medical and chiropractic state-licensure
files. Practitioners were recruited if they cared for self-referred
acute LBP and practiced in an ambulatory care setting more than
half time. A total of 208 (74%) eligible practitioners agreed to
participate. These 208 providers were made up of 87 primary care
practitioners, 64 chiropractors, 29 orthopedic surgeons, and
28 physicians, nurse practitioners, or physicians’ assistants in a
group-model Health Maintenance Organization (HMO). In
North Carolina, DCs are not permitted to prescribe medication,
but nurse practitioners and physician’s assistants can.
Each of the participating providers recruited sequential selfreferred
patients presenting with acute LBP. English-speaking
patients coping with an episode of back pain of less than 10
weeks’ duration who had not previously sought care for the
current episode were eligible for the study. Exclusion criteria
were pregnancy, past back surgery, known diagnosis of nonskin
malignancy in the past 5 years, or lack of a telephone at
home. Patients were told that the study’s purpose was to determine
the duration of LBP and the types of treatments used.
They were compensated $20 for their time to participate in the
study. The compensation was funded through the federal research
grant. Fifty percent of patients seen with back pain by
the participating providers were eligible for the study. Chronic
pain and previous treatment for the current episode of pain
were the main reasons for ineligibility. Staff members in each
practice kept a list of patients recruited for the study providing
an accurate assessment of recruitment rates. Only 8% of patients
who met these requirements declined participation in the
study. Participants’ histories were recorded and physical examinations
performed by practitioners during the initial visit.
Staff of the Survey Research Unit (SRU) at the University of
North Carolina at Chapel Hill contacted patients after the index
visit with a median lag time of 7 days. Patients were contacted
2, 4, 8, 12, and 24 weeks after the baseline interview or
until they felt that they were “completely better” if that was
before 12 weeks. All patients were interviewed at 24 weeks,
independent of recovery. Data were collected in a systematic
fashion and recorded on computer-assisted telephone interview
units and transferred to the Sheps Center for Health Services
Research at the University of North Carolina for analysis. At
each interview, patients were asked about prescription and
nonprescription medications they had used since the previous
interview. We queried patients regarding both classes of agents
(“muscle relaxants”) and also regarding specific commonly
used agents (“Flexeril, Soma. . .”). When a medication was not
on the prompted list of most common muscle relaxants and
other medications, they were asked to give the specific name.
All medications were then later categorized by one of the authors
(T.C.). Additional medications not in the prepared list
were recorded verbatim and classified by one of the project
The primary outcome was return to a functional status
equivalent to that before the onset of pain. At each interview,
patients were asked if they had returned to functional status
equivalent to one before this episode of back pain. If they answered
yes to this question, the interviewer worked with the
patient to determine the date on which the patient attained the
status of returning to their previous functional status. Patients
also were asked if they were “completely better” at each interview.
Functional status was assessed initially and at subsequent
interviews using the Roland-Morris adaptation of the Sickness
Impact Profile, a 23-item validated scale specifically designed to
assess loss of function due to back problems.  This instrument
was found to be the best available questionnaire for detecting
change over time in patients with LBP.  Patients were also
asked at each interview if they had taken muscle relaxants,
NSAIDs, acetaminophen, or steroids with prompting for specific
Bivariate analyses were performed using
2 tests for comparing categorical variables to muscle relaxant
versus no muscle relaxant use. The Student t test was used for
the bivariate analyses of continuous variables. The log-rank
test was used to compare time to recovery for the 2 muscle
relaxant use groups. Previous analysis of this data set revealed
similar recovery times for primary care practitioners, chiropractors,
or orthopedic surgeons, so these provider types were
not considered separately.
We used Cox proportional hazard models to estimate time
to functional recovery for the comparison of muscle relaxant
groups, adjusted for any potential confounders. Hazard ratios
and 95% confidence intervals (CIs) were calculated using the
covariate adjusted β coefficients and standard errors.  In addition,
we graphed adjusted Kaplan-Meier curves based on our
models. STATA Release 5 statistical software package was
used for all data analysis. 
Patients were enrolled in the study from June 1992 to
March 1993. Time-to-recovery data were obtained for
all 1,633 of these patients. The average age of patients in
the study was 41.4 years (Table 1). Approximately onefourth
of patients (24%) suffered from sciatica (pain radiating
to the level of the knee or below). When asked to
rate their pain on a 1 to 10 scale, the mean response was
5.4. The average baseline Roland score was 11.2. Of the
95% of patients who returned to their baseline functional
status by self-assessment during the study time
period, the mean time to functional recovery was 16.2
days with a median recovery time of 8 days from the
index visit. Seventy-eight percent of all enrolled patients
used NSAIDs, and 49% used muscle relaxants at some
point during the study.
Characteristics of Patients who Take Muscle Relaxants
Patients taking muscle relaxants were slightly more likely
to be younger, female, on worker’s compensation, and to
have had previous LBP treatment when compared to patients
not on muscle relaxants (Table 1). Muscle relaxants
were more commonly used by patients who initially
sought care from MDs (63% at some point during treatment
and 58% after the initial visit). Doctors of chiropractic
cannot prescribe medication in North Carolina,
but some patients who sought care from DCs did take
prescription medications, presumably either prescribed
by an MD without a visit, left over from previous care, or
used a relative’s medication. Although the differences in
pain at baseline are statistically significant, these differences
are not considered clinically important. The patients
taking muscle relaxants had significantly higher
Roland scores. Of patients visiting a physician, 63%
took a muscle relaxant medication during their episode
of back pain, compared with 23% of those who selected
a physician of chiropractic as their initial provider.
Higher back-related disability as measured by the Roland
scale was associated with muscle relaxant use
among both patients of MDs and DCs. Baseline pain
scores were similar between patients who did and did not
use muscle relaxants, in both MD and DC strata.
Nonadjusted Kaplan-Meier estimates were used to
compare the rates of recovery of patients who took muscle
relaxants with those that did not. The log-rank test
revealed a statistically significant difference between
these 2 groups (P < 0.001). The 2 groups might be expected
to have a significant difference in recovery time
due to their baseline characteristics. To account for the
differences between groups, several Cox proportional
hazards models were used.
The first model was adjusted for Roland score at baseline,
age, white or nonwhite race, education, the presence
or absence of sciatica at baseline, subjective pain at baseline
on a 1 to 10 scale, income, presence of back pain for
greater than or less than 2 weeks before baseline, and
worker’s compensation status. Age, race, and education
were not confounders and were subsequently excluded
from the model. The survival curves for this model are
shown in Figure 1. These analyses use muscle relaxant
use after the initial visit as the indicator of use. The hazard
ratio (HR) associated with muscle relaxant use was
0.88 (95% CI 0.80–0.99). This ratio, illustrated in Figure
1, indicates that patients taking muscle relaxants,
after controlling for baseline status, return to selfassessed
ability to perform their daily activities more
slowly than patients who do not take muscle relaxants.
This could be related directly to these medications or to
factors that may be indirectly related to muscle relaxant
use that were not measured (greater time spent in bed
after injury, increased likelihood of falls, etc.).
In addition, we fit amodel to control for NSAID use in
this population. This was done to determine if NSAID
use had any effect on the difference in time to functional
recovery for patients that used or did not use muscle
relaxants. A dichotomous variable representing any
NSAID use was added to the model. This gave a hazard
ratio for muscle relaxant use of 0.89 (95% CI 0.80–
0.99). Controlling for NSAIDs did not affect the difference
in time to functional recovery between the 2 muscle
relaxant groups. The HR associated with NSAID use
was not significant (HR = 0.94, 95% CI 0.83–1.07).
A third set of models was fit to examine the effects of
muscle relaxants in patients with moderate impairment
separately from those patients with severe functional impairment
from back pain. Two identical Cox proportional
hazard models were used to evaluate functional
recovery in patients with Roland scores greater than or
equal to 12, indicating more severe functional impairment,
and in patients with scores less than 12. The mean
Roland score at baseline in this study was 11, so patients
with scores of 12 or greater represent the upper 50%. In
patients with Roland scores greater than or equal to 12,
those taking muscle relaxants took longer to recover after
controlling for baseline Roland score, subjective pain,
sciatica, income, duration of episode before the initial
visit (greater or less than 2 weeks), NSAID use, and
worker’s compensation (Figure 2). The hazard ratio in
this group was 0.81 (95% CI 0.69–0.94), which indicates
that these patients took 19% longer to reach functional
recovery than those patients that did not take muscle
relaxants. The effect in patients with Roland scores
less than 12 was not statistically significant (HR 1.02,
95% CI 0.88–1.19)). These individuals recovered at the
same rate regardless of medication used.
We also examined the Roland functional status scores
adjusting for the same baseline variables for patients
who did and did not take muscle relaxants. Similar
scores were found at 2 and 4 weeks after baseline assessment
with statistically significantly higher scores in the
group taking muscle relaxants at 8-week and 12-week
interviews. The difference in the Roland scores were on
the order of 1 to 1.5 points, which is generally considered
not to be of clinical significance. Therefore, the Roland
scores also indicated little difference between the outcomes
of patients who do and do not take muscle
Our study is most similar to a study conducted by Cherkin
et al at Group Health Cooperative Sound and published
in 1998.  These data were derived from a randomized
trial with similar entry criteria (no previous surgery,
20–69 years old, first care seeking for this episode of
LBP) to the North Carolina Back Pain Project. All patients
were evaluated by a primary care MD. About onethird
of patients received amuscle relaxant, substantially
less than the proportion of MD patients receiving this
class of medications in our cohort (58% at the initial
visit), also conducted in the early 1990s. The Seattle
study found that receipt of muscle relaxant medications
was associated with lower (better) symptom scores when
compared to patients who did not take muscle relaxant
medications. Our cohort study found that use of muscle
relaxants was much more common in North Carolina
compared with theHMOsetting in Seattle, and we could
not find evidence of benefit. The differences in utilization
may be in part due to the practice setting, staff model
HMOcare is common in the Pacific Northwest, and may
in part be due to geographic variation in use of medications.
North Carolina does have relatively high rates of
surgical intervention for back procedures, and the high
use of medication may reflect an overall more aggressive
approach in this state. 
A review of existing literature suggests there is some
benefit from the use of muscle relaxants in the treatment
of acute LBP as monotherapy and possibly in conjunction
with NSAIDs. A recent meta-analysis of the cyclobenzaprine
literature found consistent evidence of a
moderate effect on pain, range of motion, and activities
of daily living.  There is a paucity of large, wellperformed
RCTs in the literature. The existing studies
comprise a wide chronological and methodologic spectrum.
Some of the muscle relaxants used in early trials
are rarely used today, and other newly available agents
have not been used in clinical trials relating to back pain.
The magnitude of this apparent benefit is unclear from
the available evidence.
In patients with severe acute LBP, as evidenced by a
Roland score greater than or equal to 12, muscle relaxant
use was associated with a statistically significant increase
in time to functional recovery. Patients with worse
symptoms at baseline were also more likely to receive the
muscle relaxant medication. Although we did control for
multiple prospectively assessed clinical and demographic
characteristics, uncontrolled differences between the 2
groups may have led to patients with a worse prognosis
being prescribed muscle relaxants and consequently having
a worse clinical course. However, we found no evidence
of benefit, which would indicate that a benefit, if
present, is likely modest. In patients with less severe back
pain, there was no demonstrable effect from muscle relaxant
use. This held true even after controlling for Roland
score, subjective pain, sciatica, income, duration of
episode (greater or less than 2 weeks), NSAID use, and
Systematic reviews have shown modest evidence for
the use of systemic muscle relaxants in the literature.
These reviews are, of course, based on existing RCTs,
which generally compare muscle relaxants with placebo,
not other active pharmacological agents. This large cohort
study showed no evidence of benefit and even a
delay in functional recovery for severely affected patients
who take muscle relaxants in the setting of acute back
pain. This cohort study should encourage development
of better studies examining the effects of muscle relaxants
and other drugs in the setting of acute LBP. Randomized
trials conducted in community practice and in
settings in which other commonly prescribed medicines
are coadministered (such as acetaminophen and
NSAIDs) may help practitioners to better understand the
most appropriate role of this class of medications. A
better understanding of their clinical effects in acute LBP
is needed, especially taking into account the scope of the
problem and the frequency of their administration.
Muscle relaxants are commonly used in the community-based
treatment of acute LBP.
In this observational study, the use of muscle relaxants was not
associated with more rapid functional recovery from back
pain, even after adjusting for baseline variables.
The authors recommend community-based, randomized trials of
the use of muscle relaxants to determine whether their use
in addition to standard treatment confers incremental benefit.
The authors thank Jane Darter and Anne Jackman for
their generous assistance with this project.
Verbrugge LM, Patrick DL.
Seven chronic conditions: their impact on US
adults’ activity levels and use of medical services.
Am J Public Health 1995;
The epidemiology of spinal disorders.
In: Frymoyer JW, ed. The Adult Spine. Principles and Practice.
New York, NY: Raven Press; 1991:
Browning R, Jackson JL, O’Malley PG.
Cyclobenzaprine and back pain: a meta-analysis.
Arch Intern Med 2001;161:1613–20.
Carey TS, Garrett J, Jackman A, et al.
The Outcomes and Costs of Care for Acute Low Back Pain Among Patients Seen
by Primary Care Practitioners, Chiropractors, and Orthopedic Surgeons.
The North Carolina Back Pain Project
New England J Medicine 1995 (Oct 5); 333 (14): 913–917
Cherkin DC, Wheeler, KJ, Barlow W, et al.
Medication use for low back pain in primary care.
Hadler NM, Carey TS, Garrett J.
The influence of indemnification by workers’ compensation insurance on recovery from acute backache. North Carolina Back Pain Project.
Kelsey JL, White AA.
Epidemiology and impact of low-back pain.
Survival Analysis: A Self-Learning Text.
New York, NY: Springer-Verlag; 1996.
Masterson JH, White AE.
Electromyographic validation of pain relief: a pilot study in orthopedic patients.
Am J Orthop Surg 1996;8:35–40.
Mielenz TJ, Carey TS, Dyrek DA, et al.
Physical therapy utilization by patients with acute low back pain.
Phys Ther 1997;77:1040–51.
Riddle DL, Stratford PW, Binkley JM.
Sensitivity to change of the Roland-Morris Back Pain Questionnaire: part 2.
Phys Ther 1998;78:1197–207.
Roland M, Morris R.
A study of the natural history of back pain. Part I: development of a reliable and sensitive measure of disability in low-back pain.
Smucker DR, Konrad TR, Curtis P, et al.
Practitioner self-confidence and patient outcomes in acute low back pain.
Arch Fam Med 1998;7:223–8.
Stata Statistical Software: Release 5.0.
College Station, TX: Stata Corporation; 1997.
van Tudler MW, Koes BW, Bouter LM.
Conservative treatment of acute and chronic nonspecific low back pain. A systematic review of randomized controlled trials of the most common interventions.
Joines JD, Hertz-Picciott I, Carey TS, et al.
A spatial analysis of country-level variation in hospitalization rates for low back problems in North Carolina.
Soc Sci Med 2003;56:2541–53.