Effects of Practice Style in Managing Back Pain

Methods

Setting and Sample

The setting was Group Health Cooperative of Puget Sound, a Seattle-area staff model health maintenance organization with more than 350 000 enrollees. Patients making visits for back pain to participating Group Health Cooperative primary care physicians during 1989 and 1990 were eligible if they were between 18 and 75 years old and had been enrolled continuously for at least 1 year. We excluded patients if they planned to disenroll or did not have a telephone. Reasons for visits were ascertained from the clinic's appointment log or an encounter form completed by the provider. We excluded visits to emergency room and walk-in facilities. Thus the resulting sample was a cross-section of patients with back pain seen by primary care physicians during a 1-year period, including acute and chronic cases. We excluded a few patients with kyphosis, scoliosis, lordosis, instability, and alignment disorders (n = 27). We excluded no patient based on the location of back pain (we included cervical pain). We excluded interviewed patients if they were seen by a physician with fewer than 10 enrolled participants (n = 59), if they did not consent to medical records abstraction (n = 18), or if their medical record was not abstracted for other reasons (n = 47).

Data Collection

Three to six weeks after the index visit, patients were asked to participate in a 30-minute telephone interview. Twenty-minute follow-up interviews were done by telephone 1 year and 2 years later. Patients provided informed consent for all interviews and for abstraction of medical records.

Patient Care Processes

We assessed patient care processes by information obtained from patient interviews, from automated health care information systems, and by abstraction of medical records. We used the following variables in the analyses reported here.

Advice Regarding Surgery

In the initial interview, patients were asked if their physicians had discussed surgery. If so, they were asked about the nature of the physician's advice (surgery was harmful, not necessary, possible if pain continues, probably needed, or definitely needed). Because of small numbers of participants, we combined patients who were told that surgery was possibly, probably, or definitely needed.

Opioid Prescriptions

We obtained information on prescription of opioid medications by patient report and from automated pharmacy records. Among patients who filled an opioid prescription from a Group Health Cooperative pharmacy within 14 days of their index visit, 88.3% reported receiving a prescription for an opioid medication in the initial interview. We report interview data on the percentage of patients given opioids for back pain and the percentage who said they had been instructed to use opioids for 14 or more days or until their back pain improved (extended use instructions).

Sedative-Hypnotic Medications

We identified prescriptions of sedative-hypnotic agents and muscle relaxants. Among patients who filled a prescription for these medications at a Group Health Cooperative pharmacy within 14 days of the index visit, 80.8% reported receiving a prescription for a sedative-hypnotic medication. We report the percentage who said they received a prescription for a sedative-hypnotic medication for back pain and the percentage who reported receiving extended use instructions as defined for opioids.

Multiple Medications

Some patients reported receiving prescriptions for three or more classes of back pain medication. The classes included opioid, sedative-hypnotic, anti-inflammatory, and antidepressant medications.

Bed Rest

We determined the percentage of patients who said their physician prescribed bed rest and the percentage prescribed extended bed rest. Extended bed rest was defined as advice to rest for 7 or more days or advice recommending extended bed rest.

Satisfaction with Medical Care

Patients rated the quality of care they received from their physician on a scale of 0 to 10, where 0 was false and 10 was completely true. We used results for the following criteria to assess global satisfaction with care: The doctor was very thorough in asking questions to understand the nature of your back problem; was very thorough in the physical examination of your back; let you tell your story and asked thoughtful questions; showed interest in you as a person; and treated you as if your back pain was an important problem and not something imagined.

Satisfaction with Patient Education

We also report results for the following ratings assessing the quality of patient education: Your doctor checked to make sure you understood what you were told about your back pain and its treatment; after your visit with the doctor, you fully understood the causes of your back problem; after your visit with the doctor, you fully understood how to care for your back problem; and your visit to the doctor helped you feel that you could control your back pain. These ratings were made on the same scale of 0 to 10 as the items assessing satisfaction with medical care.

Medical Record Diagnoses

We used diagnoses recorded during the index episode of care (defined as the month before and after the index visit) in case-mix adjustments. We combined diagnoses of disc disorder and sciatica because they often overlapped. A second diagnostic group included patients with arthritis. The remaining diagnostic categories included muscle strains, injuries, and back pain not otherwise specified. We assessed the agreement between abstractors in identifying and coding medical record diagnoses for a sample of records (n = 29), which was high (86% agreement). We also obtained information regarding the location of back pain from the medical record.

Pain and Pain-Related Disability

We assessed pain status by measures of pain intensity, chronicity, and disability.

Pain Intensity

We obtained ratings of pain intensity during the interview: usual (or average) back pain intensity when the patient was in pain in the previous 6 months, and worst pain intensity in the previous 6 months on scales of 0 to 10, where 0 was no pain and 10 was worst possible pain [7].

Back Pain Days

We obtained the number of days in the previous 6 months in which the patient had back pain. We defined chronic back pain as pain present for 90 or more days in the previous 6 months [7].

Activity Limitation Days

We recorded the number of days in the previous 6 months that the patient was kept from usual activities (work, school, housework) because of back pain [7].

Disability Score

We tabulated the mean of three 0 to 10 back pain disability ratings (interference with daily activities; change in ability to participate in recreational, social, and family activities; and reduced ability to work, including housework) [7].

Pain Impact Scale Score

This scale included a subset of items from the Roland back pain disability scale [8] supplemented by other criteria relevant to assessing pain-related activity limitation [7]. Patients were asked if they had been limited in specific activities because of back pain in the previous 2 weeks. The scale score was the proportion of 16 items that was positive, indicating the presence of a specific activity limitation. Typical items included whether the patient had stayed in bed more, avoided heavy jobs around the house, shown less affection, participated in fewer social activities with groups of persons, accomplished less at work, or sought entertainment less often in the past 2 weeks because of back pain.

Chronic Pain Grade

Using a simple, empirically derived and validated grading system [7, 9], we divided severity of back pain into five categories: grade 0, completely free of back pain in the previous 6 months; grade 1, low-disability, low-intensity back pain; grade 2, low-disability, high-intensity back pain; grade 3, high-disability, moderately limiting back pain; and grade 4, high-disability, severely limiting back pain. Chronic pain was graded into these categories based on the three pain intensity ratings described previously, disability score, and activity limitation days.

Costs of Back Pain Care

Using automated cost and utilization data and information on back pain visits abstracted from medical records, we estimated the total costs of back pain care for 1 year starting with the index visit. For each type of service, we estimated average costs by the mean cost for a unit of service of the study sample during the 1-year cost analysis period. The average cost estimates were primary care, walk-in, and emergency visits, $69; specialty visits, $250; physical therapy visits, $33; chiropractic visits, $26; plain film radiograph, $50; computed tomography scan, $228; magnetic resonance imaging, $675; and drug prescriptions filled for medications used in pain management, $9.80. We assigned the same average cost to emergency and primary care visits because most of the emergency visits were to walk-in facilities with costs of care that were similar to the cost of a primary care visit. We used actual costs available for each inpatient episode associated with a back pain diagnosis for inpatient care. Cost estimates reflect the direct and overhead costs to Group Health Cooperative for providing different forms of service rather than a schedule of charges.

Physician Practice Style

We assessed back pain management by study physicians for each patient by the number of the following process of care variables that were positive: 1) the patient was prescribed opioid medications for back pain; 2) the patient was told to take opioids for 14 or more days or until the back pain improved; 3) the patient was prescribed sedative-hypnotic medications for back pain; 4) the patient was told to take sedative-hypnotic medications for 14 or more days or until the back pain improved; 5) the patient was managed with multiple pharmaceutical agents, as defined earlier; 6) the patient was prescribed bed rest; 7) the patient was told to rest in bed for 7 or more days or was told that extended bed rest would be helpful; and 8) the patient was told that surgery was possibly, probably, or definitely useful.

We used these process of care variables to characterize practice style because they have been hypothesized to have adverse effects on patient activity levels; reinforce patient expectations that physicians can cure back pain; and discourage self-care on the part of the patient. We estimated the mean number of these process of care variables that were positive for the study patients of each physician (n = 44). Physicians whose mean score was in the top quartile were classified as “more frequent prescribers,” physicians in the bottom quartile were classified as “less frequent prescribers,” and those in the middle half were classified as “intermediate prescribers.” Subsequently we refer to patients of less frequent prescribers as the “low group,” patients of intermediate prescribers as the “moderate group,” and patients of more frequent prescribers as the “high group.” The analyses we report assess the relation of physician practice style group to patient outcomes, satisfaction, and costs of care.

Statistical Analysis

We analyzed the data using Statistical Analysis Systems (SAS) software [10]. We assessed whether patients in the low, moderate, and high groups differed in functional outcomes, satisfaction, and costs of care after adjusting for case-mix variables. We assessed differences in pain intensity and chronicity ratings after adjusting for patient age, sex, educational level, first treatment episode status, site of back pain, history of back surgery, and the provider's diagnosis (disc disorder or sciatica, arthritis). We assessed differences in disability, satisfaction, and costs of care after adjusting for the same set of case-mix variables plus average pain intensity and back pain days at the initial assessment. Using the General Linear Models procedure of SAS, we assessed the effect of physician practice style group (low, moderate, high) on study measures of average pain intensity, back pain days, activity limitation, global satisfaction ratings, patient education ratings, and costs of back pain care after controlling for the case-mix variables. For measures that were assessed at 1 month, 1 year, and 2 years, we report only results for participants with complete data at all three interviews by estimating a between-group F-test for each interview wave after adjusting for baseline characteristics. To reduce the heterogeneity of the sample, we repeated the analyses of practice style differences in functional outcomes and costs after excluding patients with disc disorder or sciatica. We also did stratified analyses for patients with chronic (90 or more days of back pain) or nonchronic back pain at baseline.

A statistical concern is that satisfaction, costs of care, and outcomes of patients managed by the same physician may be correlated, making true variances greater than the estimated variances based on the assumption of statistically independent observations. To evaluate this potential bias, we estimated intraclass correlations for the study variables in which significant between-group differences were observed and computed F-tests adjusted for the observed intraclass correlation [11]. We estimated intraclass correlations using the SAS VARCOMP procedure. After adjusting for patient level covariates, we divided the portion of the remaining variance attributable to individual physicians by the total remaining variance. Using the method of Donner and associates [11], we multiplied the mean square error in the analysis of variance by the factor 1 + r times the average sample size per physician, where r is the estimated intraclass correlation. We first report statistical analyses unadjusted for intraclass correlation, followed by analyses adjusted for intraclass correlation.

Results

Response Rates

We completed interviews with 1213 patients with back pain treated by their primary care physicians (with a 72% response rate), of whom 1071 met all eligibility criteria. We sought follow-up data for patients who did not die, became too ill to be interviewed, or moved out of the country. We completed 1- and 2-year follow-up interviews by telephone with 911 (85%) patients. Although response rates were high across the two follow-up waves, the nonrespondents differed in some respects from the nonrespondents: They were less educated, had more health problems, consulted a physician more often in the year before baseline, and were more likely to have severe pain dysfunction.

Sociodemographic Characteristics

The sample was divided almost equally between men and women (Table 1). The high educational attainment of the study patients is characteristic of Group Health Cooperative's enrollment. We found a modest, but statistically significant, difference in educational attainment across the three groups, with the patients of physicians who less frequently prescribed pain medications and bed rest reporting somewhat higher levels of education.

Clinical Characteristics

As determined from medical records data, disc disorder or sciatica and arthritis were diagnosed in a few study patients (Table 1). About one in five patients in each of the three groups had either cervical or thoracic pain (in the absence of lumbar or sacral pain).

Pain Chronicity and Intensity

We assessed several indicators of the chronicity of back pain. Less than one half of the patients reported 90 days of back pain in the previous 6 months (Table 1). Almost 80% reported that their first episode of care for back pain had occurred more than 3 months before the index visit. Fewer than 10% reported a history of back surgery. Table 1 also shows the distribution of pain intensity ratings in the three practice style groups.

Case-Mix Adjustment

For the 10 case-mix variables listed in Table 1, we found no significant differences across the low, moderate, and high groups, except for the modest difference in educational attainment noted previously. Although the three groups appeared similar, we entered these variables as case-mix adjusters in multivariate analyses assessing the relation of practice style group to process and outcome variables, as described previously in Methods.

Practice Style Profile

Table 2 profiles how study patients were managed by low-, moderate-, and high-group physicians. We used the eight practice style variables (listed in order of overall frequency of occurrence) to place physicians in the low, moderate, and high groups, so the differences in Table 2 simply convey the magnitude of practice style differences across the three groups.

Overall, patients with back pain were often prescribed opioids (about 30%). Sedative-hypnotics and bed rest were also commonly prescribed. Physicians in the high group prescribed these interventions to their patients more than twice as often as did low-group physicians, as would be expected because frequency of prescribing was the basis for grouping patients. We found that advice that surgical intervention might be needed was given, multiple pharmaceutical agents were used, and extended bed rest was prescribed for approximately 10% of participants overall. We observed large differences in the frequency of use of multiple pharmaceutical agents and extended bed rest recommendations across the three practice style groups. Extended use of sedative-hypnotics and opioids was prescribed to about 5% of patients overall, with a more than twofold difference in prescribing rates between the high and low groups. Although differences in surgical advice were used in grouping physicians by practice style, patients in the low, moderate, and high groups showed only a small difference in the percentage advised that surgery was possibly indicated. Thus, practice style differences across low-, moderate-, and high-group physicians were confined to differences in prescribing pain medications and bed rest.

Pain Outcomes

At long-term follow-up, patients treated by physicians differing in practice style reported similar pain intensity and days of back pain (Table 3). Differences among the three groups in mean pain intensity and mean days of back pain were not significant at 1 month, 1 year, and 2 years. In each group, patients showed improvement in usual pain intensity ratings from the initial (1 month) assessment to the 1-year assessment. There was only slight additional improvement in pain intensity ratings from 1 year to 2 years. The median number of back pain days decreased substantially from 1 month to 1 year, with a modest additional decrease from 1 year to 2 years. Table 3 also shows the number of back pain days reported by patients at the 75th percentile of the distribution of days of back pain. At the 1-month assessment, 150 or more back pain days were reported by one in four patients with the most persistent back pain in the previous 6 months. At least one in four patients reported chronic back pain (90 or more days of back pain) at each of the follow-up interviews.

Patient Satisfaction Ratings

The patients of low-, moderate-, and high-group physicians reported similar ratings of satisfaction with medical care (Table 4). However, the high-group physicians received less favorable ratings on items assessing the quality of patient education. As shown in Table 4, differences in patient education ratings were consistently significant. For example, physicians who more often prescribed bed rest and pain medications received a mean agreement rating of 5.2 (based on a 10-point scale) on the item “your visit to the doctor helped you feel that you could control your back pain,” whereas the patients of infrequent prescribers gave a mean rating of 6.1 on that item.

Functional Outcomes

We found a statistically significant (P = 0.02) difference in the percentage of patients graded as having moderate to severe pain dysfunction (chronic pain grades 3 to 4) at the 1-month assessment (Table 5). A monotonic increase occurred in the percentage of patients at a dysfunctional chronic pain grade from the low to moderate to high groups. At 1 month, the percentage graded as having moderate to severe activity limitation because of back pain was 30% among patients of infrequent prescribers of pain medications and bed rest, 37% in the moderate group, and 46% among patients of frequent prescribers. Practice style group differences in chronic pain grade did not persist at 1- or 2-year follow-up. A significant minority of patients with back pain receiving primary care had fair or poor outcomes. The percentage with a fair or poor pain status (grades 2 to 4) was 64% at 1 month, 33% at 1 year, and 30% at 2 years.

We assessed the effect of physician practice style group on pain-related activity limitation, as measured by disability score and pain impact scale score, after adjusting for case-mix variables. As with the chronic pain grade, low-group patients reported the least pain-related activity limitation at 1 month, moderate-group patients were intermediate, and high-group patients reported the greatest activity limitation (Table 5). These differences were very significant after adjustment for case-mix variables for pain impact scale score and were of borderline significance for disability score. At the 1- and 2-year follow-up interviews, practice style group differences in pain impact scale score and in disability score were not significant.

Costs of Back Pain Care

Table 5 shows estimates of the mean costs of back pain care during 1 year of follow-up for low-, moderate-, and high-group patients. We found significant differences for total costs of back pain care and for costs of outpatient back care alone in the mean costs of care by practice style group. Patients treated by low-group physicians had substantially lower costs than did patients of high-group physicians. Compared with those of the low group, the costs for outpatient care for back pain were 53% higher and the total costs for back pain care were 79% higher among the patients of high-group physicians. The costs of care for patients of moderate-group physicians were intermediate between the low and high practice style groups. The adjusted difference in costs, after controlling for case-mix variables, was $277 (95% CI, $85.50 to $471.32). We examined the components of health care costs that explained cost differences across the low, moderate, and high groups. The total cost differences across the three groups were due primarily to differences in the use of inpatient care and specialty care. The difference in costs due to differences among the low, moderate, and high groups in use of primary care visits, imaging studies, and medications was negligible. For example, the per-patient cost for specialty visits was $66 for the low group, $104 for the moderate group, and $170 for the high group, whereas the average cost for inpatient care was $61 for the low group, $110 for the moderate group, and $170 for the high group. In contrast, the per-patient cost for primary care visits was $156 for the low group, $129 for the moderate group, and $162 for the high group; the average cost for imaging studies was $43 for the low group, $44 for the moderate group, and $54 for the high group.

Cost and Outcome Analyses after Stratification by Diagnosis and Chronicity

To control further for heterogeneity of patients across the three groups, we examined the relation of practice style group to functional outcomes and costs for care in analyses excluding persons with disc disorder or sciatica. We also analyzed functional outcomes and costs for care stratified by chronic and nonchronic pain status. Exclusion of cases with disc disorder or sciatica did not alter the pattern of results, although the magnitude of cost differences across the three groups was reduced. When we stratified analyses by pain chronicity (90 or more days of back pain in 6 months compared with less chronic back pain), practice style group showed a similar relation with functional outcomes and costs for care. Again, the magnitude of cost differences by practice style group was smaller in the nonchronic group, but cost differences in both strata were significant after log transformation of ambulatory care costs.

Analyses Adjusted for Intraclass Correlation

For the disability variables, the intraclass correlations were so small (0.003 for pain impact scale score and 0.0008 for disability score) that they had no effect on significance tests. The intraclass correlations of outpatient costs (0.01) and total costs (0.006) were larger, but adjusted F-tests remained significant. The intraclass correlations for ratings of the quality of patient education were the largest: 0.0136 for “the doctor helped the patient feel in control of the pain,” 0.0165 for “the patient understood the cause of the pain after the visit,” 0.0326 for “the patient understood back care after the visit,” and 0.0588 for “the doctor checked to make sure the patient understood what the doctor said about back pain and its treatment.” All four F-tests for patient education variables, adjusted for intraclass correlation, were of borderline significance after adjusting for intraclass correlation (0.05 < P < 0.10). Overall, the basic pattern of results was not changed fundamentally by analyses accounting for within-physician intraclass correlation.

Discussion

The long-term outcomes, patient satisfaction, and costs of care among patients with back pain managed by physicians with a practice style more consistent with self-care were compared with those of patients managed by physicians who more often prescribed pain medications and bed rest. The practice style more consistent with self-care was associated with 1) higher ratings of satisfaction with patient education; 2) no differences in ratings of satisfaction with medical care; 3) no differences in long-term pain or functional outcomes; and 4) lower costs for back care during a 1-year period. These results suggest that a practice style more consistent with self-care of back pain achieved similar long-term outcomes at lower cost and with greater patient satisfaction.

Limitations

Limitations of this study included its observational design, the lack of an assessment of patients at the index visit for back pain, and the heterogeneity of diagnosis and previous history of back pain among participants. It is unclear whether the difference in functional status observed at 1 month was due to the effects of physician practice style or whether it was a preexisting case-mix difference. Although we used an observational design, the three groups of patients compared were similar for a profile of 10 demographic and clinical variables. Controlling for these 10 case-mix variables did not alter the basic pattern of practice style differences in satisfaction with patient education and costs for care. Analyses stratified by diagnosis and by chronicity yielded similar results, as did analyses accounting for intraclass correlation. An important limitation is that the study physicians were all practicing within the same health maintenance organization, which may have substantially reduced the extent of practice variation among physicians.

We found variation in the number of patients with back pain per physician enrolled in the study. One physician had 90 study patients (8.8% of the total), but removing this physician from the analysis did not change the study results. For the remaining physicians, the number of patients per physician ranged from 10 to 47.

Patients lost to follow-up tended to be more dysfunctional at baseline than did the patients interviewed at the 1- and 2-year follow-ups. Selective attrition may have reduced the ability of this study to record a long-term effect of practice style on functional status, but the percentage of patients lost to follow-up during 2 years was relatively small (15%).

Quality of Information about Self-Care May Be Important

Differences in practice style between physicians who frequently or infrequently prescribed pain medications and bed rest may reflect more than simple differences in prescribing patterns. The physicians who infrequently prescribed bed rest and pain medications were viewed more favorably in their abilities to educate patients about back pain and its care and to instill confidence in self-care abilities. Physicians able to educate patients about self-care may find they can meet patient needs while prescribing pain medications and bed rest less often. Physicians with less skill or interest in patient education may use short-term palliative interventions that reduce independent efforts by the patient to find ways to manage back pain on their own. Because back pain is a recurrent condition, long-term costs of back care may be influenced by whether a patient views back pain as a condition amenable to self-care or a condition that requires medical care during each “acute” episode.

Increasing evidence indicates that the exchange of information between doctors and patients is an essential component of patient care [12, 13]. Physicians who place less emphasis on providing information to patients with back pain and more emphasis on physician-directed palliative care may inadvertently discourage the initiative, information seeking, and experimentation by patients that is essential to acquiring confidence and skills in self-care. In previous research [14, 15] at Group Health Cooperative of Puget Sound, efforts to teach primary care physicians skills in patient education did not result in measurable effects on how physicians managed patients, on patient satisfaction, or on outcomes. How physicians can interact with patients in ways that foster attitudes and behaviors consistent with self-care is an issue that deserves increased research.

Implications in Light of Previous Research

Patients of physicians who infrequently prescribed medications and bed rest reported less pain-related activity limitation at the 1-month assessment. However, we observed no long-term effects of practice style on either pain or functional outcomes. Short-term differences in functional status may have resulted directly from differences in physician advice about rest, which could explain why we observed no long-term differences in functional status.

The results of this observational study must be interpreted in light of the four relevant experimental studies conducted to date [16-19]. One reported short-term (but not long-term) differences in functioning for patients prescribed 2 days rather than 7 days of bed rest [17]. Two reported a long-term effect on functioning of early advice to engage in a graded resumption of activities [16, 19], although one found this effect only among patients with no previous treatment history [19]. A fourth study found no effects on outcome but did find that early behavioral activation changed patient strategies of coping with back pain [18]. They reported that they were more likely to exercise even with back pain.

Implications of the Chronic-Recurrent Course of Back Pain

Most study participants had back pain that ran a recurrent course, whereas more than we expected had a chronic phase of back pain during the 2-year follow-up. Forty-four percent were in a chronic phase (90 or more days of back pain in 6 months) at either the 1- or the 2-year follow-up, whereas 29% of patients with recent-onset pain described their back pain as persistent at either the 1- or the 2-year follow-up [20]. Patients with back pain receiving primary care typically have recurrent pain, and evidence is increasing that patients are more likely to have chronic phases of back pain than was previously believed [9, 20, 21]. For patients with recurrent or chronic back pain, inadequate patient education and training in self-care skills may increase long-term costs. Available evidence from randomized, controlled trials does not support the effectiveness of medical treatments for long-term control of chronic and recurrent pain for most patients with back pain [6, 22]. Unfortunately, health care providers receive less training and have fewer incentives to provide information and teach self-care skills than to order diagnostic studies, medically prescribed palliative care, and surgical interventions of unknown efficacy and high cost.

Some physicians have negative attitudes toward patients with chronic back pain who expect a definitive diagnosis, a definitive remedy, or analgesic drugs strong enough to eliminate their pain. These patient expectations may be shaped by repeated contact with health care providers who provide a diagnosis and short-term palliative care without educating the patient about the long-term course, approaches to back pain self-care, and the limitations of medical care to manage recurrent or chronic back pain episodes. When doctors provide palliative care during an acute or recurrent flare-up of back pain, it is understandable that patients attribute subsequent improvement to their doctor's care, even when improvement would have occurred without medical intervention [23]. In the absence of information that prepares patients for self-care of recurring episodes of back pain, patients may tend to seek additional medical treatments when their pain recurs and may consider surgical intervention if their back pain becomes chronic.

Conclusions

The major question posed by our study was whether different styles of prescribing pain medications and bed rest to patients with back pain who are receiving primary care influenced long-term functional outcomes. Long-term differences in functional outcomes by practice style group were not evident. However, a practice style characterized by infrequent prescribing of pain medications and bed rest was associated with greater patient satisfaction with education about back care, similar short- and long-term pain relief, and lower costs of back care. Cost differences were due to differences in use of inpatient and specialty care more than to costs of primary care visits.

These results support the cost-effectiveness of a style of managing patients with back pain characterized by less frequent prescribing of pain medications and bed rest in combination with adequate attention to providing information about the condition and self-care. Because this research was observational, experimental studies are needed to confirm these results. Relevant experimental studies have yielded suggestive but inconsistent results. Research supports the need for development and rigorous evaluation of interventions for patients with back pain receiving primary care that emphasize patient education, prompt return to function, and greater reliance on self-care for long-term management of recurrent or chronic back pain.