Impact of national low back pain guidelines on clinical practice.

Objectives: The purpose of this study was to assess the impact of the 1994 Agency for Health Research and Quality (AHRQ) clinical practice guidelines on the management of acute low back pain.

Methods: From the National Ambulatory Medical Care Service database, the authors abstracted data on patients being seen in primary care settings, presenting with low back pain as their primary reason for visit, and aged between 20 and 55 years. Patients with an inflammatory or secondary diagnosis to explain their back pain were excluded. Using the sampling weights assigned by the National Ambulatory Medical Care Service, we assessed the medications prescribed, referrals for physiotherapy, and radiography usage for 3 years before (1991 to 1993) and after (1995 to 1997) release of the back pain guidelines.

Results: During these 6 years, more than 10 million ambulatory office visits were available for analysis, 5.2 million visits between 1991 to 1993 and 5.0 million visits between 1995 to 1997. The most common diagnosis was lumbago, present in 21% of these visits. Acetaminophen use increased 20-fold from 0.1 to 2%, nonsteroidal use increased from 40 to 43%, muscle relaxant use decreased from 29 to 20%, radiography ordering increased slightly from 15.4 to 19.4%, and physical therapy referrals declined from 27 to 22%. There was no evidence of a trend toward increased compliance with the AHRQ guidelines over time.

Conclusions: The AHRQ clinical practice guidelines for the management of acute low back pain had a modest impact on physician behavior, increasing the use of acetaminophen and nonsteroidals and decreasing the use of muscle relaxants and physical therapy referrals.

Key Words: low back pain, practice guideline

**********

Clinical practice guidelines are systematically developed statements that assist in decision-making about appropriate care for specific clinical conditions. (1) The goal of clinical practice guidelines is to provide busy clinicians with algorithms to quickly provide "best practice" alternatives to their patients, based on careful evaluation of the evidence. (2) Ideally, clinical practice guidelines focus on common problems with significant morbidity or mortality. Back pain fits this ideal. It is very common, accounting for 200,000 (1.8%) office visits in the United States per year, (3) and since only one fourth of patients with physical symptoms seek medical attention, (4) those presenting for care represent only the tip of the iceberg. It is estimated that up to 50% of working adults have back pain every year, (5) and 70% of all adults have it at some time in their life. (6) Although most patients recover to full activity within 1 month, (7) others incur serious disability and consume substantial health care resources; the 20% of claimants having back pain for more than 4 months account for 60% of the back pain health care costs. (8) Because back pain is common and expensive, the Agency for Health Care Research and Quality (AHRQ) selected it as a focus for one of their guidelines, released in 1994. The purpose of this analysis is to determine the effect of these national clinical practice guidelines on national physician practices.

Materials and Methods

National Ambulatory Medical Care Survey

Data for this study came from the 1991 to 1999 National Ambulatory Medical Care Surveys (NAMCS). (9-17) Conducted by the National Center for Health Statistics, Hyattsville, MD, the survey uses a multistage probability sampling strategy of American Medical Association and American Osteopathic Association master lists of all US ambulatory physicians. Each year, a sample of physicians from these lists is selected randomly, stratified by geographic area and specialty. Among physicians who agree to participate, patient visits during a randomly selected week are systematically sampled. For each selected visit, the physician completes a form detailing the specific clinical services provided, patient demographics, reasons for the visit, and diagnoses. The first listed diagnosis represents the primary reason for the patient's visit. Physicians are instructed to record all medications recommended to the patients, using brand or generic names. Among selected physicians, annual participation rates ranged from 74% in 1989 to 63% in 1999.

Data Extraction 

For this analysis, we included patients being seen in a family practice, internal medicine, or general practice clinic, who were between the ages of 20 and 55 years, with low back pain as the primary reason for the visit (codes 1910.0, 1910.1, 1910.3), and without an inflammatory or secondary diagnosis to explain their back pain. Data from 1991 to 1993 (before release of the AHRQ guidelines in 1994) were compared with data from 1995 to 1997. Data extracted and analyzed included medication recommended (up to 5), whether or not physiotherapy was prescribed, and whether or not a radiograph was obtained. Because of concern that some of the impact of national clinical practice guidelines may take several years to occur, additional data were pulled from the 1998 and 1999 NAMCS database, and were graphed for trend. In addition, beginning in 1997, NAMCS added a question to their survey regarding whether the primary problem for which the patient was seeking attention was acute, chronic routine, chronic flare-up, or a follow-up for either injury or surgery. We did a subgroup analysis of patients presenting between 1997 to 1999 reporting their low back pain to be acute in onset.

Statistical Analysis

Comparison was between the period before and after release of the guidelines. All analyses were done with the use of Stata (Version 7.0, College Station, TX). The national estimate of the number of visits was done by using patient visit weights. Each visit is assigned a weight, by NAMCS, that is derived from the probability of being sampled, accounting for regional and specialty sampling bias as well as nonresponse. All analyses were done with the use of these analytic weights. Categorical data were analyzed by means of [chi square] and continuous variables with analysis of variance, comparing the periods before and after release of the guidelines. We also evaluated the study population for demographic and practice trends over time by using logistic regression. Because of the large number of records, very small differences could be detected as statistically significant. Therefore, a priori and arbitrarily, we defined a clinically significant change in physician behavior as more than a 5% change in behavior. Exploration of the effect of age, specific diagnosis, and sex on our results was done by means of analysis of variance.

Results

A total of 10,255,661 encounters met the criteria and were included in this analysis. The mean age of patients was 38 years, 57% were male, and 61% of the visits were injury-related (Table 1). There were no clinically significant differences between patients seen from 1991 to 1993 and those from 1994 to 1997 (age: 37.4 versus 38.5; sex: 54% male versus 56%). Forty different diagnoses were given, the most common being lumbago, present in 21%. Other common diagnoses included lumbar sprain (19%), sacroiliac sprain (14%), sprain not otherwise specified (13%), backache/back sprain not otherwise specified (6%), disk displacement not otherwise specified (5%), history of injury (3%), and sciatica (1%). The results are shown in Table 2.

Because of the very large sample size, all differences were statistically significant. There was no change in the analyses after adjusting for patient age, sex, and specific diagnosis, and there was no evidence of a trend toward increasing compliance with the AHRQ guidelines over time (Figs. 1 and 2).

Discussion

The AHRQ clinical guidelines on the management of acute low back pain recommended that acetaminophen be used as first-line drug therapy. Nonsteroidal anti-inflammatory drugs were reported to be effective, though their use was recommended as second-line therapy because of the higher incidence of side effects. Use of muscle relaxants and narcotics, routinely obtaining plain films, and referring patients for physical therapy were not recommended. These guidelines appear to have had a modest impact on the clinical management of back pain in the United States. There was a 20-fold increase in the use of acetaminophen, though its recommendation remained low, at only 2%. There was a 4% increase in the percentage of patients given a nonsteroidal anti-inflammatory drug or narcotics, and no change in the proportion given a muscle relaxant. There was a 2% increase in radiography usage, and an absolute 5% reduction in physical therapy referrals. There also appeared to be no evidence that these guidelines were slowly being further integrated into practice, with no evidence of favorable trends in any parameter during the 5 years after guideline release.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

The AHRQ consensus panel was convened in 1992 with more than 200 people from 19 different specialty groups, (18) with a total budget of $600,000. Guideline development took 2 years, with the final guideline published in 1994. The finished guideline received full support from the AHRQ in publishing and disseminating the guidelines to clinician practices and electronically at the AHRQ web site. By 1997, this guideline had received more visitors through the internet than any other AHCPR guideline. (19) Part of the dissemination process included encouraging the more than 200 guideline development participants to publish in their own respective specialty journals and to lead workshops at their respective national meetings.

Although the AHRQ has shifted their focus away from developing clinical practice guidelines, guideline promulgation continues to be a focus in our current health care system. The National Guidelines Clearinghouse, sponsored by the Agency for Health Research and Quality, the American Medical Association, and the American Associations of Health Plans, lists 918 practice guidelines that can be electronically searched and accessed from the web site. Other online services such as MDCONSULT list more than 600 peer-reviewed practice guidelines, retrievable in full text. This exponential growth in the interest in guidelines can be seen in MEDLINE; a literature search for practice or clinical guidelines yielded 10,000 citations from 1966 to 1990, increasing to 44,000 in the last decade. Thirty thousand of these articles have been published in the past 5 years, 15,000 since 1999. Along with the growth in practice guidelines, interest in formal evidence-based medicine and health care cost control have also emerged as powerful motivating forces in current medical care and policy. Theoretically, the development of guidelines to standardize care based on the levels of available evidence can improve the overall outcomes and reduce the overall cost of health care. Although the effectiveness of practice guidelines has generated much interest and literature, studies suggest that these theoretical advantages have not been realized. (20,21) There are many possible barriers to the effectiveness of guidelines. A recent review of more than 293 articles assessing barriers to physician adherence to guidelines found common obstacles included awareness of the guideline, familiarity and agreement with the guideline, physician belief in their ability to comply with the guideline, lack of belief in the guideline producing expected outcomes, inability to overcome inertia in practices, and the presence of external barriers to guideline compliance. (22) Another barrier is the lack of local input. Guidelines appear to be more effective when derived by the clinicians using them rather than imported from external organizations. (21)

Our findings are similar to the results of other studies (21,23) that show little change in clinician behavior. Although our study had a number of strengths--including a large sample size, sampling representative of ambulatory practices across the entire United States, and a long study period--there are important limitations. First, the AHRQ clinical practice guidelines are applicable only to patients without findings on the history or physical examination that are worrisome for more serious underlying pathology. Since the sample is a "snapshot," one cannot know the final diagnosis. However, for this analysis, the final diagnosis that emerges over time is less important than what the clinician believes them to have at that particular visit. Our interest was determining how physicians behaved when they believed the patient to have the kind of back pain that falls into the AHRQ guidelines. Our marker for this belief is based on the clinician's ICD-9 code assigned for that patient's back problem. We selected only diagnoses with benign connotations, such as back strain, and limited the analysis to patients between 20 and 55 years of age. It seems unlikely that a significant proportion of patients could have been miscoded into nonspecific ICD-9 codes such as lumbar strain if the clinician was worried about a more serious problem.

Second, we are assuming that any change seen in physician behavior can be attributed to the AHRQ guidelines. This is impossible to prove because the guidelines are based on the evidence upon which clinicians should be making their decisions. The same evidence available to the AHRQ in creating the guideline was available to clinicians treating patients. Was the modest effect seen the result of the impact of the AHRQ guideline, or was it due to clinician response to the literature? However, since the guidelines were promulgated specifically by the AHRQ with the intent of modifying clinician behavior, holding them to this standard is not unreasonable. This analysis thus may overestimate the importance of the AHRQ guidelines in effecting the modest changes seen.

Third, it is likely that acetaminophen use was significantly underreported because it is a nonprescription drug, although the directions for clinicians completing the NAMCS forms specifically ask that both prescription and nonprescription drugs recommended as treatment be recorded. It is hard to believe that acetaminophen was recommended by clinicians less than 2% of the time. The percentage of patients given no medications, according to the database, declined from 24% before the guideline to 14% after. If any significant portion of these were told to take acetaminophen, this data would not suggest that this proportion was increasing since 1994.

Third, before 1997, the timing of the back pain could not be clearly determined, that is, whether the patient was presenting with acute low back pain or with a back pain problem that had been present for many years. Although we selected diagnostic categories that represent acute syndromes, some categories could fit with chronic pain. In our analysis of patients reporting their back pain to be acute in onset, presenting between 1997 to 1999, there were no significant differences in our findings.

Finally, the NAMCS database does not indicate what type of radiograph was obtained in response to a particular visit. Although we limited our patients to those presenting only with low back pain, we cannot be certain that the plain films obtained were of the back. The rates we report are comparable to those found in clinical studies of back pain management and were steady during the 8 years of sampling for this study.

Conclusion

Our analysis suggests that the AHRQ guidelines had little or no impact on the clinical management of low back pain. Most studies that have examined the impact of clinical practice guidelines have also found modest or no effects. Whether clinical practice guidelines are worth the resources that continue to be dedicated to them remains a matter of speculation.

Happiness is good health and a bad memory.
--Ingrid Bergman

Table 1. Sample characteristics

                                     1991-1993  1995-1997

Demographics
  Age, mean (yr)                     38.3       39.3
  Male, %                            53         50
  White, %                           86         84
  Black, %                           12         12
  Hispanic, %                         2          3
Injury related, %                    45         38
Clinician diagnosis, %
  Lumbago                            21         22
  Lumbar sprain                      19         18
  Sacroiliac sprain                  14         15
  Sprain, NOS                        13         12
  Back pain NOS                       6          5
  Rheumatism                          4          5
  Osteoarthritis                      4          6
  Unspecified disorders of the back   3          2
  Spondylosis                         2          1
  Contusion of back                   1          1
Physician specialty
  Internal medicine                  34         40
  Family practice                    31         25
  General practice                   35         35

NOS, not otherwise specified.

Table 2. Impact of the Agency for Health Care Research and Quality's
1994 clinical practice guidelines on physician management of low back
pain

                            1991-1993  1995-1997

Medication prescribed, %
  Acetaminophen              0.1        2
  Nonsteroidal anti-        39.6       43.3
  inflammatory
  Muscle relaxant           20.4       20.3
  Narcotic                   2.5        6.4
  No medication prescribed  24.0       14.0
Physiotherapy prescribed    27         22
Radiograph obtained         15.4       19.3

All comparisons, P < 0.0001.

Presented as an oral abstract at the 24th Annual Society of General Internal Medicine meeting in San Diego, CA, on May 4, 2001.

Accepted May 5, 2004.

Please see Ronald C. Hamdy's editorial on page 135 of this issue.

References

1. Institute of Medicine. Clinical Practice Guidelines: Directions for a New Program. Washington, DC, National Academy Press, 1990.

2. Hayward RS, Wilson MC, Tunis SR, et al. Users guides to the medical literature, VIII: how to use clinical practice guidelines JAMA 1995;274:570-574.

3. Ambulatory care visits to physician offices, hospital outpatient departments, and emergency departments: United States, 1997. Vital Health Stat 1999;13:1-36.

4. White KL, Williams TF, Greenberg BG. The ecology of medical care. N Engl J Med 1961;265:885-892.

5. Wipf JE, Deyo RA. Low back pain. Med Clin North Am. 1995;79:231-246.

6. Deyo RA, Rainville J, Kent DL. What can the history and physical examination tell us about low back pain? JAMA 1992;268:760-765.

7. Staiger TO, Paauw DS, Deyo RA, Jarvik JG. Imaging studies for acute low back pain: when and when not to order them. Postgrad Med 1999;105:161-172.

8. Williams DA, Feuerstein M, Durbin D, Pezzullo J. Health care and indemnity costs across the natural history of disability in occupational low back pain. Spine 1998;23:2329-2336.

9. National Center for Health Statistic. 1991 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1993. Public use data tape documentation.

10. National Center for Health Statistic. 1992 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1994. Public use data tape documentation.

11. National Center for Health Statistic. 1993 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1995. Public use data tape documentation.

12. National Center for Health Statistic. 1994 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1996. Public use data tape documentation.

13. National Center for Health Statistic. 1995 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1997. Public use data tape documentation.

14. National Center for Health Statistic. 1996 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1998. Public use data tape documentation.

15. National Center for Health Statistic. 1997 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 1999. Public use data tape documentation.

16. National Center for Health Statistic. 1998 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 2000. Public use data tape documentation.

17. National Center for Health Statistic. 1999 National Ambulatory Medical Survey. Hyattsville, MD: National Center for Health Statistics. US Public Health Services; 2001. Public use data tape documentation.

18. Cuzick J. A. Wilcoxon-type test for trend. Stat Med 1985;4:87-90.

19. Bigos SJ. Lower back pain, perils, pitfalls, and accomplishments of guidelines for treatment of back problems. Neurol Clin 1999;17:179-192.

20. Worrall G, Chaulk P, Freake D. The effects of clinical practice guidelines on patient outcomes in primary care: a systematic review. Can Med Assoc J 1997;156:1705-1712.

21. Grimshaw JM, Russell IT. Effect of clinical guidelines on medical practice: a systematic review of rigorous evaluations. Lancet 1993;342:1317-1322.

22. Cabana MD, Rand CS, Powe NR, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999;282:1458-1465.

23. Kosecoff J, Kanouse DE, Rogers WH, et al. Effects of the National Institutes of Health Consensus Development Program on physician practice. JAMA 1987;258:2708-2713.

RELATED ARTICLE: Key Points

* Clinical practice guidelines for management of acute low back pain were issued in 1994.

* In the 3 years following release of these guidelines, the use of acetaminophen and nonsteroidal anti-inflammatory drugs increased, while the use of muscle relaxants decreased, as did referrals for physical therapy.

* Agency for Health Research and Quality back pain guidelines had a modest impact on clinical practice.

Jeffrey L. Jackson, MD, MPH, and Robert Browning, MD

From Uniformed Services University of the Health Sciences, Bethesda, MD, and National Naval Medical Center, Bethesda, MD.

The views presented in this article are those of the authors and should not be construed to represent, in any way, those of the US Navy, the US Army, or the Department of Defense.

Reprint requests to Dr. Jeffrey L. Jackson, Medicine-EDP, 4301 Jones Bridge Road, Bethesda, MD 20814. Email: [e-mail]