Abstract

Low back pain is considered to fluctuate over time, and related prognostic factors may behave similarly, therefore classification of prognosis may be affected by timing of assessment. We aimed to investigate the implications of timing of assessment of prognosis in low back pain. In a prospective cohort of primary care low back pain consulters aged 30-59 years, 359 returned questionnaires at baseline, one-month and one-year. The stability of selected prognostic indicators between baseline and one-month later was assessed, and relative risks for their association with outcome (individually and in combination) were calculated. Reported absence of most outcome indicators was stable between baseline and one-month. However, among people reporting presence of indicators at baseline, only around half still reported them a month later. There were no important differences between the prognostic strength of indicators measured at baseline or one-month. People reporting presence of indicators at either point had significantly higher risk of poor outcome than people not reporting the indicator at all. Presence of the indicators at both time points was associated with even higher risk; people with persistent high pain intensity had 15 times the risk of a poor outcome (relative risk 15.1; 95% confidence interval 6.7-33.8) compared with people not reporting high pain at either point. Combining information on prognostic indicators from two time points provides better classification of low back pain patients' eventual outcome than a single measurement alone. This increased accuracy in predicting prognosis is relevant to both clinical and research practice.