In Reply.—Chandrasekar points out an important limitation of attempts to devise diagnostic algorithms based on relatively small numbers of patients. Rare conditions such as spinal epidural abscess are unlikely to be observed except in an extremely large sample of primary care patients with back pain. To put the problem of epidural abscess into perspective, its prevalence is estimated to be about 37 per million patients with acute low back pain in primary care practice.1 From the perspective of an infectious disease specialist at an inpatient center that serves substantial numbers of intravenous drug users, it is likely to seem much more common.
The frequent absence of fever in patients with pyogenic vertebral osteomyelitis or spinal epidural abscess is well described, and I would agree that fever alone is a relatively insensitive indicator for these infections. Because of the increased risk of occult serious infection in intravenous drug (Continued on p