ow back pain is common in the population and in the healthcare provider's office. The lifetime prevalence of episodes of low-back pain may be as high as 80%, and about 11% of the population has an episode of back pain bad enough to impair their usual daily activities during the course of a year.1 Acute back pain has a generally good prognosisover 90% of patients functionally recover-but in an unfortunate minority of patients the acute process becomes chronic.2 Chronic back pain, once established, has a generally poor prognosis with many patients having ongoing symptoms for months or years. Back pain remains a common cause of work disability with enormous social costs to employers, insurers, and families. The identification of an appropriate treatment approach would substantially aid providers, insurers, and the public. The lack of consistent treatment outcome information has led to claims by various provider groups that one approach or the other is preferred over all other approaches. Over the past decade multiple cohort studies and trials have been conducted. While health policy can generally not be established on the basis of any single study, accumulating evidence in studies from both the US and Europe can now begin to guide providers and the public. The article by Kominski and colleagues in this issue of Medical Care3 compares treatment costs in a highly managed care setting across 4 treatment approaches: MD only; MD plus referral to a physical therapist (PT); chiropractic only; and chiropractic also utilizing physical modalities. The patient population was quite heterogeneoussome of the patients had very acute back pain; others had symptoms for years. The almost entirely capitated nature of the insurance in this population would likely tend to dampen differences in the number of visits across groups; the providers had no incentive to over-use visits. Indeed, the number of visits to physical therapists and doctors of chiropractic is substantially less than reported in other cohort studies, such as that conducted by our group in North Carolina just a few years prior to data collection in Kominski's study. Therefore, the differences between specialties noted here are likely relatively conservative compared with the fee-for-service environment. We are told relatively little about the costs of imaging these patients. Spinal radiographs, CT, and MRI scans may be a substantial portion of the cost differential across specialties. Indirect costs of transportation and time off work were also not assessed. European studies have found that inclusion of these costs may modestly alter conclusions. Korthals de Bos, in a study of the care of neck pain, found that at one year direct costs were less and indirect costs were greater for individuals who received care from a general practitioner (GP) relative to a PT, with total costs similar between GP and PT. A manual therapy strategy was overall less costly.4 A final, acknowledged, limitation is that the investigators did not collect medication costs. Medication costs would modestly increase the costs of those who saw MDs as opposed to Doctors of Chiropractic (DC). Our work