Abstract/Text: Back pain is very common. Rarely, it may be the first manifestation of cancer. Although many advocate selective use of laboratory and x-ray tests for back pain patients, the early detection of cancer may be an important reason to obtain such tests. To develop a diagnostic approach that would identify malignancies while remaining parsimonious, the authors evaluated 1,975 walk-in patients with a chief complaint of back pain. Thirteen patients (0.66%) proved to have underlying cancer. Findings significantly associated with underlying cancer (p less than 0.05) were: age greater than or equal to 50 years, previous history of cancer, duration of pain greater than 1 month, failure to improve with conservative therapy, elevated erythrocyte sedimentation rate (ESR), and anemia. Combining historical features and ESR results led to an algorithm that would have limited x-ray utilization to just 22% of subjects while recommending an x-ray for every cancer patient. It would further suggest which patients with negative x-ray findings require further work-up.

Abstract/Text: PURPOSE: To review evidence on the diagnostic accuracy of clinical information and imaging for patients with low back pain in primary care settings.
DATA SOURCE: MEDLINE search (January 1966 to September 2001) for articles and reviews relevant to the accuracy of the clinical and radiographic examination of patients with low back pain.
STUDY SELECTION: The authors reviewed abstracts and selected articles for review on the basis of a combined judgment. Data on the clinical examination were based primarily on recent systematic reviews; data on imaging tests were based primarily on original articles.
DATA EXTRACTION: Diagnostic results were extracted by one or the other author. Quality of methods was evaluated informally. Major potential biases were identified, but neither quantitative data extraction nor scoring was done.
DATA SYNTHESIS: Formal meta-analysis was not used because the diagnostic hardware and software, gold standards, and patient selection methods were heterogeneous and the number of studies was small. Sensitivity for cancer was highest for magnetic resonance imaging (0.83 to 0.93) and radionuclide scanning (0.74 to 0.98); specificity was highest for magnetic resonance imaging (0.9 to 0.97) and radiography (0.95 to 0.99). Magnetic resonance imaging was the most sensitive (0.96) and specific (0.92) test for infection. The sensitivity and specificity of magnetic resonance imaging for herniated discs were slightly higher than those for computed tomography but very similar for the diagnosis of spinal stenosis.
CONCLUSIONS: The data suggest a diagnostic strategy similar to the 1994 Agency for Health Care Policy and Research guidelines. For adults younger than 50 years of age with no signs or symptoms of systemic disease, symptomatic therapy without imaging is appropriate. For patients 50 years of age and older or those whose findings suggest systemic disease, plain radiography and simple laboratory tests can almost completely rule out underlying systemic diseases. Advanced imaging should be reserved for patients who are considering surgery or those in whom systemic disease is strongly suspected.