Lumbar Radiographs for Low Back Pain

Abstract & Commentary

Source: Suarez-Almazor ME, et al. Use of lumbar radiographs for the early diagnosis of low back pain. JAMA; 227:1782-1786.

In 1994, the agency for health care policy and Research (AHCPR) published practice guidelines for management of patients with acute low back pain (defined as back pain of less than 3 months’ duration).1 These guidelines included recommendations about when practitioners should obtain lumbar radiographs.

In an interesting study recently published, Suarez-Almazor and colleagues performed a retrospective chart review of patients with acute low back pain who presented to four family medicine clinics in Alberta, Canada. They sought two outcome measures: how often did practitioners obtain lumbar radiographs at the initial "low back pain" visit, and what was the rate of osteomyelitis, tumor, or fracture noted on subsequent visits. The authors then applied the AHCPR criteria for obtaining a lumbar series to their patient cohort to assess the impact of the AHCPR guidelines on clinical practice.

The chart audit included patient visits from the year 1992-1993, just prior to the publication of the AHCPR guidelines, at two community and two academic outpatient family medicine clinics. The authors included all adult patients with an initial visit for acute low back pain (< 3 months duration) and excluded patients with visceral causes of low back pain, pregnancy, and ankylosing spondylitis. In total, 963 patients met study criteria, and 127 patients (13%) underwent radiologic evaluation; half of this group received oblique views in addition to the A-P and lateral views. Of the 127 lumbar series, 50% showed degenerative changes, 35% were interpreted as normal studies, 3% revealed a fracture, and one patient had a diagnosis of possible metastatic disease. Given that these clinics represented the point of care and continuity center for most of the patients, medical records were reviewed for an additional two years after the initial back pain visit and included 87% of the initial study group. Combining the initial and follow-up visits, three patients were noted to have spinal metastases, and five patients were diagnosed with spinal fracture (0.8% of the initial group). In this subset of eight patients, four had lumbar radiographs at the initial visit, and four had radiography performed in follow-up. For the four patients without radiographs at the initial visit, one had known myeloma at the initial visit, one had bony metastases with unknown primary diagnosed two months later, and two patients had compression fractures noted previously, with no new fractures on follow-up. No patient was diagnosed with osteomyelitis.

The 1994 AHCPR guidelines laid out "red flag" criteria for clinicians to help identify spinal fracture, infection, or cancer. In short, red flags for fracture include: major trauma; minor trauma in a geriatric patient; prolonged steroid use; osteoporosis; and patients older than 70. The red flags for infection or tumor include: patients older than 50 or younger than 20; a history of cancer; constitutional symptoms (weight loss, fever, chills); intravenous drug use; recent bacterial infection; immunosuppression; pain increased when supine; and severe nocturnal pain. When these criteria were applied to the cohort of 963 patients, the use of lumbar radiographs would have increased from 13% to 44%. That is to say, 44% of the patients had one or more of the above criteria. Comparing physician-use criteria with AHCPR guidelines, the physicians could report a sensitivity of 50% and a specificity of 84%, while applying the AHCPR guidelines resulted in a sensitivity of 93% and specificity of 56%.


Well, this is certainly an example of deciding where the rubber meets the road. The clinicians obtained a lumbar series in 127 patients, yet missed four patients with fracture or tumor for a sensitivity of only 50%. If the AHCPR guidelines were applied as written, 418 of 955 patients would have received a lumbar radiograph at the initial visit, to uncover eight patients with tumor or fracture. The sensitivity then increases to 93%, but with a three-fold increase in x-ray use, there would be a sharp decline in specificity.

How do we interpret this information? Certainly our intuition and our reading tells us that the prevalence of serious disease in patients with acute low back pain is low (ranging from 1-2 patients per 1000 in some decision analysis models). I do not routinely obtain a lumbar series on all patients older than 50 or younger than 20 (who have no other red flags) to screen for tumor or infection. Excluding this group alone would have resulted in nearly a 50% reduction in screening radiographs, without sacrificing sensitivity. So, although we can appreciate the intent of the AHCPR guidelines, we can appreciate the impact in the ED as well. For the vast majority of patients, the lumbar series will be negative or reveal nonspecific degenerative changes, while cost and throughput times will rise dramatically. If available for your patients, careful urgent outpatient follow-up visits for selected groups may make more sense. However, be cognizant of the guidelines and be prudent in documenting your decision to defer radiography. To my mind, these guidelines call for a prospective ED-based study.

One important additional point was raised by the AHCPR expert panel regarding the use of oblique views. Routine oblique lumbar views are discouraged by the AHCPR and World Health Organization. The dose of radiation doubles (in fact, the radiation dose for a lumbar series with oblique views is 40 times that of a routine chest radiograph), and the yield from the additional views is low. Therefore, AP and lateral views alone are the initial recommendation.


1. Acute Low Back Problems in Adults: Clinical Practice Guideline, Quick Reference Guide for Clinicians Number 14. Rockville, MD: U.S. Department of Health and Human Services, Agency for Health Care Policy and Research; 1994. AHCPR publication 95 0613.