SPINE Volume 30, Number 13, pp 1541–1548 ©2005, Lippincott Williams & Wilkins, Inc. Three-Year Incidence of Low Back Pain in an Initially Asymptomatic Cohort Clinical and Imaging Risk Factors Jeffrey G. Jarvik, MD, MPH,*†**†† William Hollingworth, PhD,* ** Patrick J. Heagerty, PhD,§†† David R. Haynor, MD, PhD,* Edward J. Boyko, MD, MPH,‡¶†† and Richard A. Deyo, MD, MPH†‡** Study Design. Prospective cohort study of randomly selected Veterans Affairs out-patients without baseline low back pain (LBP). Objective. To determine predictors of new LBP as well as the 3-year incidence of magnetic resonance imaging (MRI) findings. Summary of Background Data. Few prospective stud- ies have examined clinical and anatomic risk factors for the development of LBP, or the incidence of new imaging findings and their relationship to back pain onset. Methods. We randomly selected 148 Veterans Affairs out-patients (aged 35 to 70) without LBP in the past 4 months. We compared baseline and 3-year lumbar spine MRI. Using data collected every 4 months, we developed a prediction model of back pain-free survival. Results. After 3 years, 131 subjects were contacted, and 123 had repeat MRI. The 3-year incidence of pain was 67% (88 of 131). Depression had the largest hazard ratio (2.3, 95% CI = 1.2– 4.4) of any baseline predictor of inci- dent back pain. Among baseline imaging findings, central spinal stenosis and nerve root contact had the highest, though nonsignificant, hazard ratios. We did not find an association between new LBP and type 1 endplate changes, disc degeneration, annular tears, or facet degen- eration. The incidence of new MRI findings was low, with the most common new finding being disc signal loss in 11 (9%) subjects. All five subjects with new disc extrusions and all four subjects with new nerve root impingement had new pain. Conclusion. Depression is an important predictor of new LBP, with MRI findings likely less important. New imaging findings have a low incidence; disc extrusions and nerve root contact may be the most important of these findings. Key words: cohort study, epidemiology, low back pain, prognosis, risk factors, magnetic resonance imag- ing. Spine 2005;30:1541–1548 Researchers have questioned the clinical importance of many spine imaging findings for nearly 60 years. 1–13 Findings such as disc height loss and disc bulges are com- mon in individuals without low back pain (LBP). As im- aging techniques advance, our ability to accurately de- pict anatomy improves, yet paradoxically, confusion regarding the clinical importance of some anatomic find- ings increases. 6,8,9,12–14 We previously reported a cohort of subjects with little or no back pain and no sciatica at the time of magnetic resonance imaging (MRI). Some reported prior back pain, allowing us to identify imaging findings associated with earlier symptoms. 15 Most imaging findings fell into one of five categories: (1) common findings with little relationship to either aging or previous back pain (e.g., annular tears, disc protrusions); (2) common findings that were associated with increased age, but not with prior symptoms (e.g., disc bulges, facet joint degenera- tion, endplate changes, mild spondylolisthesis); (3) com- mon findings related both to aging and previous LBP (e.g., decreased disc signal on T2-weighted images, de- creased disc height); (4) rare findings unrelated to age, but strongly associated with previous back pain (disc extrusions and nerve root contact); and (5) moderate or severe stenosis, which was related to previous back pain, mild current symptoms and aging. We sought to determine the 3-year follow-up inci- dence of LBP in this cohort and identify its risk factors. We also aimed to describe incident imaging findings and their relationship to onset of new symptoms. Materials and Methods Cohort Assembly. We previously described details of the co- hort assembly. 15 Briefly, we used electronic records to ran- domly sample patients from four clinics at the Veterans Affairs Puget Sound Health Care System, Seattle Division, stratifying by age, with half of the subjects aged 35 to 52 years and the remainder aged 53 to 70. After excluding patients with ICD-9 diagnostic and proce- dure codes related to LBP or lumbar surgery, we contacted the remaining patients in random order. We excluded subjects with any history of back surgery, chymopapain injections, discogra- phy, acute lumbar trauma, fibromyalgia, peripheral neuropa- From the Departments of *Radiology, †Health Services, ‡Medicine, §Biostatistics, ¶Epidemiology and Neurological Surgery; the **Center for Cost and Outcomes Research, University of Washington, Seattle, Washington; and ††Veterans Affairs Puget Sound Health Care System and Seattle Epidemiologic Research and Information Center (ERIC), Seattle, Washington. Supported by the Department of Veterans Affairs-ERIC Grant and in part by Grants HS-08194, HS-094990 from the Agency for Healthcare Research and Quality and P60-AR48093-01 from National Institute for Arthritis and Musculoskeletal and Skin Diseases. Acknowledgment date: May 12, 2004. First revision date: July 23, 2004. Acceptance date: August 16, 2004. The manuscript submitted does not contain information about medical device(s)/drug(s). Federal funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript. Address correspondence to Jeffrey G. Jarvik, MD, MPH, Department of Radiology, University of Washington, Box 357115, 1959 NE Pacific St., Seattle, WA 98195; E-mail: jarvikj@u.washington.edu 1541