You are on page 1of 9




ABSTRACT ACK pain is a major health problem be-
Background and Methods There are few data on cause of its high prevalence1,2 and costs in
the relative effectiveness and costs of treatments for terms of health care expenditures3-5 and lost
low back pain. We randomly assigned 321 adults productivity.6 Although there are many non-
with low back pain that persisted for seven days af- surgical treatments, there is little evidence that any
ter a primary care visit to the McKenzie method of are particularly effective. Systematic reviews have con-
physical therapy, chiropractic manipulation, or a min- cluded that chiropractic spinal manipulation appears
imal intervention (provision of an educational book- to be effective in some subgroups of patients with
let). Patients with sciatica were excluded. Physical
back pain7-9 and this is one of the few treatments rec-
therapy or chiropractic manipulation was provided
for one month (the number of visits was determined ommended in clinical-practice guidelines on the care
by the practitioner but was limited to a maximum of of adults with low back pain in the United States.10
nine); patients were followed for a total of two years. The effectiveness of physical therapy for back pain
The bothersomeness of symptoms was measured has not been well studied, and the results of com-
on an 11-point scale, and the level of dysfunction parisons of physical therapy with chiropractic ma-
was measured on the 24-point Roland Disability Scale. nipulation have conflicted.11-14 A popular form of
Results After adjustment for base-line differenc- physical therapy,15 the McKenzie method, has not
es, the chiropractic group had less severe symp- been rigorously evaluated16 or compared with spinal
toms than the booklet group at four weeks (P=0.02), manipulation. We compared the effectiveness and
and there was a trend toward less severe symptoms cost of the McKenzie method of physical therapy,
in the physical-therapy group (P=0.06). However,
these differences were small and not significant af-
chiropractic manipulation, and provision of an edu-
ter transformations of the data to adjust for their cational booklet for the treatment of low back pain.
non-normal distribution. Differences in the extent of
dysfunction among the groups were small and ap-
proached significance only at one year, with greater Study Sites
dysfunction in the booklet group than in the other This study was conducted at Group Health Cooperative of
two groups (P=0.05). For all outcomes, there were Puget Sound, a large staff-model health maintenance organiza-
no significant differences between the physical-ther- tion (HMO) in Washington. Subjects were recruited from two
apy and chiropractic groups and no significant dif- Seattle-area primary care clinics between November 1993 and
ferences among the groups in the numbers of days September 1995. Physical therapy was provided in a facility adja-
of reduced activity or missed work or in recurrences cent to these clinics, and spinal manipulation was performed by
of back pain. About 75 percent of the subjects in the four chiropractors in solo practice. The study protocol was ap-
proved by the institutional review boards of Group Health Co-
therapy groups rated their care as very good or ex-
operative and the University of Washington, and all subjects gave
cellent, as compared with about 30 percent of the written informed consent.
subjects in the booklet group (P<0.001). Over a two-
year period, the mean costs of care were $437 for the Subjects
physical-therapy group, $429 for the chiropractic
Patients 20 to 64 years of age who saw their primary care phy-
group, and $153 for the booklet group. sician for low back pain and who still had pain seven days later
Conclusions For patients with low back pain, the were eligible for the study. Physicians referred potential subjects
McKenzie method of physical therapy and chiroprac- to a research assistant, who obtained consent and collected infor-
tic manipulation had similar effects and costs, and mation on the subjects’ history of back pain, sociodemographic
patients receiving these treatments had only margin- characteristics, general health, and expectations of improvement.
ally better outcomes than those receiving the mini- General health was evaluated with the general health perceptions
mal intervention of an educational booklet. Whether and mental health subscales of the Medical Outcomes Study
the limited benefits of these treatments are worth
the additional costs is open to question. (N Engl J
Med 1998;339:1021-9.) From the Group Health Center for Health Studies (D.C.C., J.S., W.B.),
©1998, Massachusetts Medical Society. the Departments of Health Services (D.C.C., R.A.D.), Family Medicine
(D.C.C.), Medicine (R.A.D.), and Biostatistics (W.B.) and the Center for
Cost and Outcomes Research (R.A.D.), University of Washington, Seattle;
and the Department of Physical Therapy, University of Alberta, Edmon-
ton, Canada (M.B.). Address reprint requests to Dr. Cherkin at the Group
Health Center for Health Studies, 1730 Minor Ave., Suite 1600, Seattle,
WA 98101.

Vol ume 33 9 Numb e r 15 · 1021

Downloaded from on September 7, 2008 . For personal use only. No other uses without permission.
Copyright © 1998 Massachusetts Medical Society. All rights reserved.

had re. sciatica. A previous trial found that the use of this booklet as a supplement Randomization to standard care was not associated with improved outcomes. the study and confirmed their compliance with the treatment tion. legs. ice.28 These Physical therapy was provided by 13 therapists with a median questions were modified to refer specifically to back-related re- of 14 years of experience.27 and the questions are eas- ily administered by telephone. trend toward worse outcomes in the group that received the and 12 weeks) and long-term outcome (1 to 2 years) were as- booklet. Therapists were asked to avoid adjuncts such as heat. the subjects were ran. high-velocity thrust directed specifically at a phone follow-up. most back. ders. A similar scale has shown Physical Therapy substantial construct validity. a history of back surgery. severe neurologic signs. spondylolisthesis. after the outcomes had been recorded the patient lying on his or her side on a segmental table. Use prognosis. discretion of the therapist. McKenzie Institute faculty trained the strictions. booklet to minimize potential disappointment with not receiving tive treatment for their current back pain. month supply of medications. Treatments subjects rated how “bothersome” back pain. even though subjects considered the booklet useful. or an educational booklet with the use Outcome Measures of sealed. . Physical therapy was covered for all ing in chiropractic orthopedics. The assessment of short-term outcome focused on the ef- cated 40 percent of subjects to physical therapy. to perform exercises that centralize their symptoms and to avoid Disability was measured with questions from the National Health movements that peripheralize them. Two had advanced train. fair. outcomes focused on recurrences of low back pain and the use of back-related health care. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne Short-Form Health Survey (SF-36). automated data on utilization for cushion. other than their primary care physicians were also excluded. sessed. and numb- Because Washington laws prohibit physical therapists from per. appropriate use of imaging studies and specialists. the assessment of long-term booklet. For personal use only. coagulation disor. and sensitive. leg pain. jects’ health insurance. ultrasonography. buttock.26 was used to measure In the McKenzie approach. or poor) at one and four randomization. point scale was used in which a score of 0 indicated that the cal therapies such as ultrasonography. the level of func- chiropractic manipulation. and one in sports chiropractic. were involved in claims for Educational Booklet compensation or litigation because of the back injury. The scores on this scale are based broad categories (derangement. can range from 0 (none of the 23 daily activities limited by back ment syndrome. Interviewers were unaware of the subjects’ “manipulable lesion. or visited practitioners a physical treatment. over the ensuing month. The first visit was scheduled within four days after pain (excellent. were pregnant. and of pain medications was not a reason for on September 7. The subjects typically had office-visit co- The chiropractors had 6 to 14 years of experience. at the dis- cluded if they had mild or no pain seven days after the visit to the cretion of the chiropractor. or with reduced activity. patients are placed in one of three the patients’ ability to function. At base line and at all follow-up visits. a vertebral fracture or disloca.21 Patients are taught ment is reliable. All rights reserved. 40 percent to fects of the assigned treatment on symptoms. an important part of McKenzie therapy. and postural syn- on the answers to 23 yes–no questions about daily activities and drome) that determine therapy. osteoporosis. but similar outcomes in the other two groups.18 The score on each of the The initial visit was scheduled within four days after randomiza- subscales ranges from 0 (worst) to 100 (best). The booklet discussed causes of back pain. A minimal-intervention control group received an educational ceived physical therapy or chiropractic or osteopathic manipula. systemic or visceral tor consultant observed the chiropractors at the start and end of causes of the pain. tion. Subjects were ex. very good. or buttocks to the lower back. Copyright © 1998 Massachusetts Medical Society.19. The subjects were asked to rate their care for back back classes. This instru- the feet. 24. opaque envelopes.” treated by exercises that “centralize” pain from pain) to 23 (all 23 activities limited by back pain). Those who had received chiropractic or physical therapy tients according to their usual procedures and were allowed to were asked about their compliance with therapy and the estimat- make the same recommendations about exercise and activity re- ed length of their visits. subjects indicated which treatment they had been physical treatments were permitted. and leg pain is thought to result from a “derange. the clinical practices are symptoms were “not at all bothersome” and a score of 10 that distinct. and questionnaires on visits not covered transcutaneous electrical nerve stimulation. both short-term outcome (1.17. and the level of disability. No other at four weeks. 19 9 8 Downloaded from www. and up to eight additional visits were scheduled. Because a pilot study suggested a Because back pain often recurs. however. and one had payments of $5 to $10 and drug copayments of $5 per one- a master’s degree in exercise physiology. spent tient-generated forces and emphasizes self-care. The need for additional radiographs was treatments were covered (within contract limitations) by the sub- determined by the chiropractor. Chiropractors evaluated pa- assigned. we allo.nejm.24 A modified 24 Roland Disability Scale 25. 4. good. Subjects received identified by a review of encounter forms completed by physical McKenzie’s Treat Your Own Back book19 and a lumbar-support therapists and chiropractors. and by the HMO. HMO-covered services. valid.” This procedure is typically performed with study assignment. and 20 percent to the educational tion. A chiroprac- physician. the symptoms were “extremely bothersome. or a severe concurrent illness. strictions that they usually did. HMO enrollees on a physician’s referral and with a visit copay- 1022 · Octo b er 8 . protocol. An exercise sheet was used that emphasized stretching and strengthening but excluded extension Costs of Care exercises. Subjects who had received corticosteroid therapy.” The score for the most bothersome symptom was used. and up to eight more visits were scheduled. chiropractic manipulation. No other uses without permission. Assessment of Outcome Chiropractic Manipulation Data on outcomes at one and four weeks were collected by The most common method of chiropractic manipulation 22 was telephone. 2008 . Nonstudy less than three years old. The HMO gave the chiropractors any radiographs of the subjects that were The study paid for the costs of the study treatments. This method relies on pa- Interview Survey about the number of days spent in bed.23 domly assigned without stratification to receive physical therapy. An 11- forming spinal manipulation and chiropractors from using physi. over the ensuing month. dysfunction. activities for promoting recovery and preventing recurrences. After base-line data had been collected.20 In this formulation. therapists before the study. subsequent outcomes were assessed by mail with tele- used: a short-lever. at the weeks. and all but one therapist passed an The subjects’ use of health care for back-related problems was advanced McKenzie credentialing examination. ness or tingling had been during the preceding 24 hours. home from work or school.

Most subjects had a $1 booklet (Table 2). and Sidak’s method for pairwise comparisons was used to adjust Chiropractic for multiple comparisons. Sixty-four cent) agreed to participate. percent) or at home (58 percent).2 vs. All subjects underwent manipula- were severe concurrent conditions (27 percent).29 A common set of base-line var. Data were analyzed according to the intention to treat. 76. included ice packs (20 percent). subjects in the physical-therapy group had detect a 2. logistic regres- sion was used.29 spent with the provider was virtually identical in the Nonparametric tests were used to confirm the results of the two groups (about 145 minutes). previous physical therapy. CHIROPRAC TIC MANIPUL ATION. regular use of the lumbar roll (71 percent) and rec- ceptions subscale were above national norms (77. white. SF-36 mental health score. For personal use only.” At four weeks. and 323 remained eligi. Ninety-two percent of the subjects were given a diagnosis of “derangement. Between 89 and 96 percent of the subjects responded to each of the Physical Therapy follow-up questionnaires. we used a visits by 50 percent (6.4). 2008 . practitioners other than their primary care physi. The most common reasons for exclusion at the initial visit. 122 to chiro. and 133 to physical therapy. base-line characteristics among the treatment groups (Table 1): subjects in the chiropractic group were Statistical Analysis less likely to have used chiropractic services previ- The study was designed to have at least 80 percent power to ously. and employed (Table 1). 66 sage (49 percent). with adjustment for base-line values whenever verse effects of treatment were reported in any of the available. on September 7. The Roland Disability and bothersomeness-of-symptoms scores were Study Treatments analyzed as continuous variables by analysis of covariance after adjustment for base-line values. or visits to iac manipulation. Eighty- the study.5).5-point difference in the scores on the bother- someness scale for the comparison between physical therapy and ucational-booklet group reported fewer days with chiropractic care. age squared. All 66 subjects in the booklet group were mailed practic care (some had had both). The mean cost of chiroprac- Vol ume 33 9 Numb e r 15 · 1023 Downloaded from www. 714 (19 percent) met the initial inclusion two percent of the subjects underwent manipulation criteria. and about one third had previ- ously received physical therapy and one third chiro.6. OR AN EDUCATIONAL BOOKLET FOR BACK PAIN ment. the total amount of time met the assumptions of the parametric test. of the preceding seven days. plying with recommended exercises. Other treatments randomization (one because of a urinary tract infec. nipulation of the hip. PHYSICAL THERAPY. Two subjects were excluded after more than one level of the spine. P value of 0. Ninety-six percent of the subjects in the chiro- iables predictive of these scores was included in all analyses of co. restricted activity and lower expectations that their sidered minimally important.30 All P values were two-tailed. many visits exceeded the mean number of physical-therapy subjects had substantial symptoms or dysfunction).e. . and 59 percent tic services. Copyright © 1998 Massachusetts Medical Society. which did not require a physician’s referral but did had had back pain for less than three weeks.001). 12 percent cervical manipulation. For dichotomous outcomes. All rights reserved. The cost of care was determined from the HMO’s cost-accounting system and reflects the costs to the HMO. had back pain for less than six weeks. practic group and 97 percent of the subjects in the variance (age. Most subjects had previously received treat- Cost ment for back pain.5-point difference in the scores on the Roland Disabil. There were only a few significant differences in Thus.. 27 percent thoracic manipulation. Because the primary outcomes were not normally distributed (i. volvement in claims for compensation or litigation In addition. These differences are consistent with those con. and 6 percent ma- cians (19 percent). more bothersome symptoms. tion of the lumbar or lumbosacral region (or both). The mean number of chiropractic subjects’ expectations of the likelihood of improvement). out-of-pocket expenses for patients are not included. 78 per- Base-Line Characteristics cent of the subjects reported that they had per- The typical subject was approximately 40 years formed the recommended exercises on at least four old. No important ad- parametric analyses. and exercises in the office (41 were assigned to receive the booklet. P<0. 54 percent underwent sacral or sacroil- because of the back injury (20 percent). pelvis.05 was considered to indicate statistical significance.nejm. Most subjects reported The mean scores on the SF-36 general health per. A total of 493 patients (69 per. physical-therapy group visited their assigned provid- eral health perceptions score.9 vs. require a copayment.29 For all comparisons of the three treatment groups a groups. or ischium. in. and subjects in the ed- ity Scale and a 1. brief localized mas- tion and one because of pancreatic cancer). well educated. The most common Recruitment and Follow-up of Patients diagnoses were sprain or strain (about 50 percent of Of 3800 patients who were potentially eligible for subjects) and facet syndrome (30 percent). and er at least once. Therapists vs. percent of the subjects underwent manipulation of ble one week later. Thus.24 condition would improve in a month’s time. No other uses without permission. Chiropractors obtained radiographs for 63 per- RESULTS cent of the subjects (58 of 92) who had not under- gone radiography at the HMO. About one third of the subjects had coverage for chiroprac. Accord- square-root transformation before analysis so that the data better ing to the subjects’ reports.31 but the mean scores on the mental considered 55 percent of their patients to be com- health subscale were below national norms (71.1 ommended sitting posture (83 percent). 4. practic. SF-36 gen.

the percentage of Eighteen percent of the subjects in the booklet subjects who used back-pain medication of any type group visited a health care provider for back pain decreased from 82 percent to 18 percent in the chi- during the study month. 4 percent had ever been hospitalized for a back problem.04).02).02) and had a lower expectation that back pain would be much better in one month’s time (P=0.1±5.3±16. with higher scores indicating greater reductions in daily activities.3±1.2±5.6 5. For personal use only.5 5. other than those assigned. and only 8 percent of the ropractic group.8±17. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne TABLE 1.3 7.5±16.1 6. 21 percent had a job requiring lifting more than 9 kg (20 lb) at a on September 7.0±2.7±3. BASE-LINE CHARACTERISTICS OF THE SUBJECTS.0±2.1±11.8±2.4 41.1±17.8±11.7±10.7±20.* PHYSICAL- CHIROPRACTIC THERAPY BOOKLET GROUP GROUP CHARACTERISTIC GROUP (N=66) (N=122) (N=133) ALL SUBJECTS NO. The treatment groups were similar with respect to the following variables: 79 percent overall had attended college.9 5. and four in the physical-therapy group).2±16.000 per year.8 5. LBP denotes low back pain. ercise was almost identical in the three groups at base line (about 57 percent) and one month (about Other Treatments 81 percent). ¶Scores can range from 0 to 10. 5.6±15. with higher scores indicating better function.7±9.54 per subject.0±5. 39 percent had a family income of more than $50. OF VALUE SUBJECTS† Age (yr) 40.2 313 24 hours¶ Roland Disability score¿ 11.7±2. 6 percent considered their health fair or poor. the mean bothersomeness score was higher in the physical- therapy group (P=0. No other uses without permission.1±3.nejm.7 77.6 313 Mental health score‡ 69. and the booklet group was less likely to have had »1 day of restricted activity due to back pain in the preceding week (P=0.2 6. §Scores can range from 0 to 10.9 77.5 313 »1 Day of bed rest due to LBP in preceding 22 35 24 28 313 week (%) »1 Day of work lost due to LBP in preceding 30 39 41 38 279 week (%) »1 Day of restricted activity due to LBP in 52 72 65 65 312 preceding week (%) Taking medication for LBP (%) 77 82 84 81 313 Taking narcotic analgesics for LBP (%) 8 13 15 13 313 Recent LBP care very good to excellent (%) 34 42 50 43 305 Expect LBP to be much better in a month’s time (%) 59 79 71 71 305 *Plus–minus values are means ±SD. those in the physical-therapy group visited providers apy averaged $238.8).2±2. During the month. .2 100 Prior chiropractic for LBP (%) 40 24 35 32 313 Helpfulness§ 6.7±3. ‡Scores can range from 0 to 100. 60 percent had read articles about back pain (mean helpfulness of readings on back pain on a scale of 0 to 10.6 12. ¿Scores can range from 0 to 23.1±2. 18 percent of subjects in the chiropractic group and 9 percent of which was for radiography. The cost of physical ther.4 6.7 321 Female sex (%) 42 53 47 48 321 Employed or self-employed (%) 85 91 89 89 321 Cigarette smoker (%) 11 17 19 17 313 General health perceptions score‡ 75. with higher scores indicating more bothersome symptoms. †The sample sizes for questions on the base-line questionnaire are slightly lower than those for questions on the initial eligibility questionnaire because of the loss of a diskette with base-line data for eight subjects (one in the booklet group. tic treatment was $226. 68 percent were married.6 71. with visits to pro. 19 9 8 Downloaded from www.4 12. from 84 percent to 27 percent in 1024 · Oc to b er 8 . The reported use of ex- viders accounting for 93 percent of the cost.4 70.04). 10 percent had pain that extended below the knee. three in the chiropractic group. and 53 percent had had back symptoms during most of the preceding 24 hours. The following differences among the groups were significant: prior chiropractic treatment was less likely in the chiropractic group (P=0.2±3.1 7. 2008 .5±2.5 12. All rights reserved.5 40.2 39.6±16. 89 percent were white.6 92 Current episode of LBP Duration <6 wk (%) 72 83 77 78 313 Bothersomeness of symptoms in preceding 5. with higher scores indicating greater helpfulness.7 72.5 77. 57 percent had actively exercised in the preceding week.7±5.4 313 History of LBP >2 Prior episodes of LBP (%) 54 58 56 56 321 First sought care for LBP in preceding year (%) 40 40 29 36 321 Prior physical therapy for LBP (%) 37 31 32 33 310 Helpfulness§ 5. Copyright © 1998 Massachusetts Medical Society.08 per subject.

MEAN COSTS OF THE TREATMENTS. There ous electrical nerve stimulation.28 Plain films (anteroposterior and lateral) 63‡ 74 67.24 40. ‡Chiropractors ordered radiographs for 58 of the 92 subjects (63 percent) who had not undergone radiography at the HMO. All rights reserved. 2).00 9. cent of the subjects in each group reported re- the differences among the treatment groups were no currences at one year.604. In the 11 months after for the base-line bothersomeness scores and the treatment.30. or injections.02). 7 percent.29).22.g. scores were associated with worse outcomes.02). We assumed that the chiropractors would also have ordered radiographs for 63 percent of the 25 subjects who had undergone radiography at the HMO. as compared with 29 percent Adjustment for the base-line Roland Disability score of the subjects in the chiropractic group and 20 per- and the other covariates had little effect (Table 3 and cent of the subjects in the physical-therapy group Fig. and 12 weeks (P=0.00 1.g. No other uses without permission. 2008 .02) but not at let group. braces.50 7.50 1.94) showed that the chiropractic group had less severe as were the percentages who reported the need for symptoms than the booklet group at four weeks bed rest (9 percent. There were no significant differences be. with greater dysfunction jects in the booklet group reported visits to any in the booklet group than in the other two groups.08 Physical therapy (n=133) Visits 97 589 49. poorer mental health reported using corsets.51 The bothersomeness of symptoms differed among and P=0. transcutane. 8 percent. .* PERCENT OF SUBJECTS MEAN INCURRING NO. and 11 percent..04) and after 4 weeks The number of days of back-related disability was (P=0.06).13) or with the use of the nonpara- 32 percent in the booklet group (P<0.062. the differences among groups duced activity were similar (36 percent in the book- remained significant at 4 weeks (P=0. †Four percent of subjects in the chiropractic group and 3 percent of subjects in the physical-therapy group did not visit their assigned provider at all.07 185.06) (Fig. provider for back pain. P=0. 24 percent of the sub- only at one year (P=0. 33 percent in the chiropractic group.16). OR AN EDUCATIONAL BOOKLET FOR BACK PAIN TABLE 2. and from 77 percent to mation (P=0.00 1.00 Mean total cost per subject 1. After adjustment similar among the groups.26 4.11 221.05). traction.38 McKenzie book 94 125 on September 7. and 70 percent reported re- longer significant at 4 weeks (P=0.nejm. Fewer than 2 percent of the subjects tics.14) or 12 weeks currences during the second year.99 Lumbar roll 92 122 10.977.28). P=0. Data on primary outcomes are shown in Table 3. 1). PHYSICAL THERAPY. respectively). and there was a trend toward less severe spectively. CHIROPRAC TIC MANIPUL ATION. respectively. P=0.19).00 Chiropractic (n=122) Visits 96 807 28. P=0. the percentages of subjects reporting re- prognostic covariates. respectively.220. P=0. (P=0.80 Mean total cost per subject 226. were no significant differences among the chiroprac- tors or the therapists with respect to their effect on Outcomes the subjects’ symptoms (e.007) and 12 weeks (P=0.54 *Costs are in 1995 dollars. Pairwise comparisons 35 percent in the physical-therapy group. Analysis of HMO records yielded similar tween groups at one year after square-root transfor. For personal use only. OF COST TOTAL COST PER GROUP COST† UNITS PER UNIT COSTS SUBJECT dollars Booklet (n=66) Booklet 100 66 1.. Approximately 50 per- non-normal distribution of the outcome measures. There were few interactions differences among the groups after adjustment for between the treatment and the subjects’ characteris- base-line use). during the second year. results (data not shown).81) and the percentages who re- symptoms in the physical-therapy group (P=0.17 Mean total cost per subject 238.99 29. but in the booklet group. Copyright © 1998 Massachusetts Medical Society. For exam- the groups were small and approached significance ple. the physical-therapy group. the groups at base line (P=0.05 for the metric test (P=0.71 and P=0. Vol ume 33 9 Numb e r 15 · 1025 Downloaded from www.01 22. at four weeks) or function (e. re- (P=0. and 13 After square-root transformation to adjust for the percent.444. significant differences among the groups in the per- Differences in the Roland Disability scores among centages who sought care for back pain. but there were no (P=0. ported missing work (17 percent.00 66.

2008 .5–2.2) 2.5) 3.1 (3.2–13. physical therapy group.1 (3.1–5. SF-36 general health perceptions and mental health scale scores.5) 4.0 (1.7 (2. Higher scores indicate more severe symptoms or dysfunction. chiropractic group vs. booklet group. The pairwise comparisons were adjusted to account for multiple comparisons. P=0.14 The British study found the At both one and four weeks.37.1) 0.5 hours) and were almost three times as high as in the book. and patient’s expectation of the likelihood of improvement in one month’s time).1) 12.7 (10.1–5. previous physical therapy.0) 12.7 (2.02 — mean (95% CI) Roland Disability score 4.3 (4. For personal use only.5) 0.9) 1. of subjects 65 119 129 Bothersomeness of symptoms 5.02 0. P=0. P=1.12 whereas low back pain was also lower in the booklet group a Swedish study found no significant differences in than in the other two groups.28 — mean (95% CI) *CI denotes confidence interval for unadjusted means.4–3. P=0. AND 12 WEEKS.6–6.11.93. P=0. .1 (2.” as compared with about 30 percent of the toward chiropractic care.007 0.8) 0. disability.32 For patients with low back pain. physical-therapy group vs. Copyright © 1998 Massachusetts Medical Society. †The P values pertain to analyses that were adjusted for the base-line value of the outcome measure and prognostic variables (age.06 — mean (95% CI) Roland Disability score 4.00.45. of subjects 63 118 117 Bothersomeness of symptoms 3. physical-therapy group. practic and physical therapy.06. booklet group.9–4.9 (3.2–13. chiropractors. No other uses without permission. and phys- only 18 percent received care during this period. P=0. and at 12 weeks.50.8–6. subjects in the booklet group sought additional care 1026 · Oc to b er 8 . Although chiroprac. of subjects 60 118 129 Bothersomeness of symptoms 3.3 (3. The respective values for the Roland Disability score were as follows: at 4 weeks. physical-therapy group vs.9–2.5 (5. age squared. a group presumably favorably inclined lent.3–4.9–5.2 (11.8) 6. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne TABLE 3. treatment comparison group. chiropractic group vs. booklet group.08 and let group (Table 4).6–2. and chiropractic group vs. A Dutch study found about one quarter of the subjects in the booklet that manual therapy (manipulation and mobiliza- group failed to answer this question. Although most of the higher $238. P=0. booklet group. Total costs of care to the HMO for low back pain tic treatment involved more visits than physical ther- over a two-year period differed by less than 2 per.001).org on September 7.32.* PHYSICAL- BOOKLET CHIROPRACTIC THERAPY VARIABLE GROUP GROUP GROUP P VALUE UNADJUSTED ADJUSTED† Base line No. recurrences of back pain. the total time spent with either a chiropractor cent in the chiropractic and physical-therapy groups or a physical therapist was similar (about 2.06.0) 0. P=0. P=0.1 (2.04 — — mean (95% CI) Roland Disability score 11.4–13. costs in the chiropractic and physical-therapy groups Among other rigorous studies comparing chiro- were attributable to the treatments themselves (Ta.2) 0. physical-therapy group. possibly because tion) and physiotherapy (exercise. physical-therapy group. Because only 18 percent of the subsequent visits for back pain. and at 12 weeks. However. chiropractic group vs.4) 2.2–3.2–5. as was the total cost of the treatments ($226. about 75 percent of benefits of chiropractic treatment to be most evident the subjects in the physical-therapy and chiropractic among patients who had previously been treated by groups rated their care as “very good” to “excel. and chiropractic group vs. and educational booklet. outcomes or costs.25 0. Neither study included a no- subjects in the booklet group (P<0.nejm. Pairwise comparisons adjusted for these same variables yielded the following results for the bothersomeness of symptoms: at 4 weeks. All rights reserved. satisfaction ly better outcomes than those who received only an with care. 19 9 8 Downloaded from www. we found that We found that patients who received chiropractic physical therapy and chiropractic manipulation had manipulation or physical therapy had only marginal- similar effects on symptoms.54. P=0. a British study conclud- bles 2 and 4). chiropractic group vs.97.1 (11.66. chiropractic group vs. function. P=0.9) 0.02. P=0.3 (1. book- let group.45.28 — mean (95% CI) 12 Weeks No. apy.0) 3. respectively).15 0. booklet group.0) 2. the cost of other HMO services for ed that chiropractic was more beneficial.7) 5. massage. booklet group.2 (2. physical-therapy group vs. chiropractic group vs. booklet group.9) 4.4–4. ical-therapy methods) had similar effects and were superior to continued treatment by a general practi- DISCUSSION tioner. 4 WEEKS.83 — — mean (95% CI) 4 Weeks No. BOTHERSOMENESS OF SYMPTOMS AND ROLAND DISABILITY SCORES AT BASE LINE.0 (5.9 (1.4–3. physical-therapy group vs.

previous physical therapy. age squared. Roland Disability Scores. previous physical therapy.nejm. The numbers of subjects are the same as in Figure 1. 16 14 Roland Disability Score Booklet 12 Chiropractic Physical therapy 10 8 6 P=0. Bothersomeness of Symptoms.28 P=0. SF-36 general health perceptions and mental health scale scores.16 P=0.06 1 P=0.06 P=0.34 0 0 1 4 12 52 104 Weeks of Follow-up on September 7. Copyright © 1998 Massachusetts Medical Society.74 4 P=0.02 P=0. OR AN EDUCATIONAL BOOKLET FOR BACK PAIN 10 9 Bothersomeness Score 8 Booklet Chiropractic 7 Physical therapy 6 5 4 3 P=0. 2008 . adjusted for the base-line Roland Disability scores and prognostic variables (age. Higher scores indicate greater disability. No other uses without permission. The values are least-squares means. For personal use only.PHYSICAL THERAPY. CHIROPRAC TIC MANIPUL ATION. OF SUBJECTS BookletJ 56J 60J 63J 55J 55J ChiropracticJ 114J 118J 118J 109J 116J PhysicalJ 127 129 117 122 121 therapy Figure 1.28 2 P=0. and subjects’ expectations of the likelihood of improvement). adjusted for the base-line bothersomeness scores and prognostic variables (age. and subjects’ expectations of the likelihood of improvement).41 2 P=0. . Higher scores indicate more severe symptoms. SF-36 general health perceptions and mental health scale scores. Vol ume 33 9 Numb e r 15 · 1027 Downloaded from www. age squared. All rights reserved.11 0 0 1 4 12 52 104 Weeks of Follow-up Figure 2. The values are least-squares means.

Thus. our consultants (Willard Man- ning. The extent to which during the month after randomization.5-point difference in the scores on the bother. most subjects in the booklet Jane Steetle. The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne services during follow-up. tors. The generalizability of this study is limited by the tions or operations for back problems in the two-year period. Although some outcomes for the chiro. unwise to refer all patients with low back pain for ence in the scores on the Roland Disability Scale or chiropractic or McKenzie therapy. toms. after adjustment. Stephanie Hauge. these results bution of scores. and the smaller number of visits to Total costs 153 429 437 McKenzie therapists (HMO employees) than to chi- ropractors (in private practice) may reflect differenc- *These costs include the costs to the HMO for office visits. The over- all costs of care for back pain were lowest in the PHYSICAL. The chiropractic manipulative technique was chosen because it was well under- stood and widely practiced. their main benefit for patients with low ropractic manipulation — in terms of either effec- back pain appears to be increased satisfaction with tiveness or cost — over physical therapy. Russ Bradley. There were no hospitaliza. Eileen Schrode.33 Meta Thayer. are difficult to interpret. there would a 1. Copyright © 1998 Massachusetts Medical Society. contact with providers. be some way of identifying the subgroups that are someness scale. the use of one month of therapy. The McKenzie thera- trol group. vided elsewhere is less clear.8 to 21. No other uses without permission. MEAN COSTS OF CARE FOR LOW BACK PAIN ability of the self-care–oriented McKenzie physical OVER A TWO-YEAR PERIOD. The num- Study treatments and supplies (mo 1) 1 226 239 ber of visits was left to the discretion of providers HMO services for low back pain* First year 87 114 134 (who were aware that both costs and benefits would Second year 65 89 64 be measured). radiology. 1028 · Octo b er 8 . The difference in the bothersome. it seems unlikely that the ef- small and. Cathie Hobbs. For personal use only. there are no clear advantages of chi- symptoms. 20. toms or function exceeded the predefined criteria Given the limited benefits and high costs. Kathy Hoyt. In any event. dollars The relative costs of chiropractic and McKenzie treatments could differ in other settings. although chiropractic ma. Sarah Parkhurst. pists received special training and were among a practic and physical-therapy groups were superior to small number worldwide who had passed a certifica- those for the booklet group. laney. and Kelli Pearson). Because only a small fraction of subjects in the booklet group received care during follow-up. Carol Canfield. Nancy treatments (contact with providers) and improve. the participating chiroprac- group had none. The mean use of a single health care system. laboratory tests. es in practice styles. and medications but exclude the costs of the study treat. predictive characteristics. worth their additional costs is open to question. Nancy Monroe. Ideally. the heads of the Group Health Northgate and Eastside physical- isfying and this may affect their perceptions of symp. Macy Massey. None of the observed differences in symp. Dakota Duncan. the differences were tion examination. Ben Givens. Leona Hokanson. and Carol Worley). Macy Massey. We are indebted to the many people who contributed to the success- The marginally better outcome of the physical ful completion of this study. on September 7. ment of back-related symptoms has been reported. Rick Bocko.5 months). their Supported by a grant (HS07915) from the Agency for Health Care Pol- responses to questions about satisfaction with care icy and Research. Casey. 2008 . a 2. All rights reserved. forms of chiropractic and physical therapy. and the exclusion of pa- tients with sciatica.nejm. Patients’ copayments are not included. A positive association between the number of the participating physical therapists (Ann Barry. so there was no evidence that the BOOKLET CHIROPRACTIC THERAPY GROUP GROUP GROUP higher initial costs of the physical treatments were TYPE OF COST (N=66) (N=122) (N=133) offset by later savings. this group McKenzie therapy in this study resembled that pro- was similar in some respects to a no-treatment con. Patients may find such contact sat. therapy departments (Nancy Casey. Renee Joer- gens. in particular our project team (Jenny treatments raises the possibility that effects were Achilles. though we were unable to identify any transformation to compensate for the skewed distri. . the use of specific number of months of continued enrollment in the HMO was similar in the three groups over the 24-month period (range. it seems for clinical importance — that is. Kay Hooks. and our Agency for Health Care Policy and Re- groups in recurrences of back pain or in the use of search project officer. were significant only for fectiveness of McKenzie therapists would be greater the bothersomeness of symptoms at four weeks and in other settings. Katie Saunders. most likely to benefit from one or both of these ness of symptoms was not significant after data therapies. Mary Cummings. Hillary Haug. and Joel Suelzle).15 ments. Robin McKenzie and the McKenzie We found no significant differences among the Institute faculty. and Kim Wheeler). Bonnie Kane. Dennis Bejin. therapy to reduce the utilization of services.5-point differ. ical-therapy and chiropractic groups had repeated Marcia Hunt. booklet group. and those of another recent study14 suggest that for nipulation and physical therapy may slightly reduce low back pain. Sheila Markman. Although nearly all subjects in the phys. This casts doubt on the TABLE 4. Kristin De- nonspecific. 19 9 8 Downloaded from www. Thus. the subjects’ satisfaction with care at one and four Whether the small benefits of these treatments are weeks.

Series 10. Roland M. Clin J Pain 1986.2:49-53. OR AN EDUCATIONAL BOOKLET FOR BACK PAIN REFERENCES 17. No. Patrick DL. New Zealand: 4. Survey of pain in the United States: the Nuprin pain re. Spine 1996. As- 507. Dyer S. In: Frymoyer JW. servative treatment in low back pain: a comparative study. No other uses without permission. Cypress BK.21:345-55. Postacchini F. Cherkin DC. Md. Managing manual therapy. Battie MC. 30. National Center for Health Statistics. Spine 1983. 95-0642. II. Coxhead CE. North WRS. Bowyer OR. Vol ume 33 9 Numb e r 15 · 1029 Downloaded from www. Rockville. Frank AO. Morris R. McHorney CA. Atlas SJ. 20.311:349-51. pain: frequency. Facchini M. Med Care 1993. Efficacy of various forms of con. 33. Braen GR. 14.20:11-9. 16. 8.: Government Print- outpatient treatment. Comparisons of the costs and dics 1988.19:499.31:247-63. 1986. Ciol MA.21:2860-71. Hurwitz EL. no.C. 22. Features include a library of all issues since January 1993. Am J Public Health 1983. van Mameren H. After this one-time registration. (AHCPR publication pain. Brook RH.73:389-95. back pain. Hart LG.30:473-83. Koes BW. 1993. and treatment patterns from a U. Frank AO. Rectangular confidence regions for the means of multivariate 13. Shekelle PG. 32. BMJ 1991. 67. Spine 1995. Waikanae. na. van der Heijden GJMG. Multi- schild PG. The effectiveness of 15. I. Orthop Clin North Am 1991. Assendelft WJJ. Spine 1997.62:626-33. for nonspecific back and neck complaints: a randomized clinical trial. A study of the natural history of back pain. van der Heijden GJMG. BMJ 1990.) 28. Ware JE Jr. 1981. Lancet 1981. 1991:1581-605. N Engl J Med 1988. et al. a full-text search capacity. Troup JDG.S. 1992. Meade TW. Low back pain the National Health Interview Survey.117:590-8. manipulation for low back pain: an updated systematic review of random. and free software for downloading articles so they can be printed in a format that is virtually identical to that of the typeset pages.300:1431-7. 1984. blinded review. Clinical practice guideline no. Chapin A. 1985. Bouter LM. D. Spine 1995. Development 10. Deyo RA. Physician office visits for low back Hutt.nejm. Neuroorthope. Randomised comparison 29. Koes BW.22:2167-77. of low back pain. 21. Spine Phys Ther 1994. An overview of the incidences and costs Spinal Publications. normal distributions. Koes BW. The ef. manipulation for low-back pain.8:145-50.303:1298-303. Conceptual framework and item selection. van der Heijden GJMG.17:28-35. Spinal 20:1899-908. Bouter LM. tional perspective. Current estimates from 11. Cherkin DC. Vol. Spine 1983. Spinal manipulative therapy in the manage- nants. Deyo RA. Keller RB. Cost and effectiveness analysis of chiropractic and phys. Lower 5. Washington.: National Center ment of low back pain. Acute low-back pain problems of guidelines for trials of treatment in primary care. Med Care 1. (DHHS publication no. 18. Assendelft WJJ. Frymoyer JW. et al. Assendelft WJJ. J Am Stat Assoc 1967. Morbidity cost: national estimates and economic determi. FULL TEXT OF ALL JOURNAL ARTICLES ON THE WORLD WIDE WEB Access to the complete text of the Journal on the Internet is free to all subscribers.318:291-300. Arch Phys Med Rehabil 1988. Adams AH. port. Street JH. For personal use only.6:28-35. BMJ 1995. Browne W. Barlow W. The lumbar spine: mechanical diagnosis and therapy. Cats-Baril WL. physiotherapy. Wheeler KJ. J Manipulative Physiol Ther 1996. and treatment by the general practitioner low back pain: attitudes and treatment preferences of physical therapists. A study of the natural history of low-back pain. The McKenzie approach to evaluating and treating low tional survey. The adult spine: principles for Health Services Research. Psychometric and clinical tests of validity in meas- 3. NCHSR research summary series. Characteristics of physician visits for back symptoms: a na. 26. (PHS) 86-3393. Bouter LM. Health Survey (SF-36). 1992. Frymoyer JW. fectiveness of chiropractic for treatment of low back pain: an update and Spine 1996. Meade TW. 156. Sternbach RA. Vital and of mechanical origin: randomised comparison of chiropractic and hospital health statistics. 6th ed. Salkever DS. Copyright © 1998 Massachusetts Medical Society. McHorney CA. Idem.69:1044-53. sults from extended follow up.1:1065-8. Kosinski M. Sidak Z. PHYSICAL THERAPY. Ries PW. December 1994.) 23. Bouter LM. Md. subscribers should go to the Journal’s home page (www. Meade TW. Orthop Rev 1990. New Zealand: Spinal Publications. 2008 . (PHS) 86-1584. To use this Web site. 24. Deyo RA. I. 31. Med Care 1994. The MOS 36-item Short-Form Health Survey (SF-36). II. Dunn R. Sherbourne CD. phone interview: results from a national survey. 25. Pitfalls of pa- 7. New York: Raven Press. . and practice. sessing health-related quality of life in patients with sciatica.) 12. clinical evaluation. Raczek AE.32:551- iotherapy treatment for low back pain and neck pain: six-month follow-up. in adults. ized clinical trials. 2. Knip. Measuring the functional status of patients with low back Health Care Policy and Research. Singer DE. Donelson R. CHIROPRAC TIC MANIPUL ATION. 19. Back pain and sciatica. October 1985. Spinal pain. Inskip H. Development of a reliable and sensitive measure of disability in low-back 9. Fisher LD. ing Office. Deyo and register by entering their names and subscriber numbers as they appear on their mailing labels. Skargren EI. Ann Intern Med 1992. Phillips RB. attempt at statistical pooling. Koes BW.8:141-4. Idem. Browne W. Biostatistics: a methodology for the health sci- of chiropractic and hospital outpatient management for low back pain: re. van Belle G. Dyer S. New York: John Wiley. ences. 6. Rockville.nejm. tient education: limited success of a program for back pain in primary care. ed. Palieri P. Ware JE Jr. United States. quality of norms for the SF-36 health survey collected by mail versus tele- 14. The MOS 36-item Short-Form 2. subscribers can use their passwords to log on for electronic access to the entire Journal from any computer that is connected to the Internet. Townsend J. Haldeman S. All rights reserved. (DHHS publication no.74:219-26. uring physical and mental health constructs. Ware JE. Hunt M. McKenzie R. Cherkin on September 7.: Agency for 27.19:681-6. Treat your own back. Spinal manipulation and mobilisation for back and neck pain: a centre trial of physiotherapy in the management of sciatica symptoms. Chassin MR . Bigos SJ. a personal archive for saving articles and search results of interest.