Background & purpose: A number of studies have identified factors associated with a poor prognosis in LBP patients. Many of these prognostic indicators have a natural course that is transient or recurrent in nature, and identification of these factors therefore depends on the timing of measurement. This work aims to examine the time-course of selected prognostic indicators for LBP in a group of primary care patients, and to assess the most appropriate points to measure those indicators.

Methods: Information on a group of selected prognostic indicators (including disability, pain intensity, leg pain, catastrophising, bothersomeness and work absence) was obtained from a consecutive cohort of 935 primary care LBP consulters. Data was collected using postal questionnaires within two weeks of consultation and each month for the subsequent six-months. Poor outcomes were defined as Chronic Pain Grade IV or work absence due to LBP at six-months.

Results: At baseline, 30% to 70% of the sample reported each of the indicators, this fell by half one month later. The baseline measures with the highest odds of a poor prognosis were high disability, high pain intensity and catastrophising. Most indicators had stronger associations with outcome when measured at one month than at baseline. People reporting the indicators at both baseline and one month had increased odds of a poor prognosis compared to people reporting the indicator once, or not reporting it at all.

Conclusion: Indicators for poor LBP prognosis are more strongly associated with outcome when present at more than one time point. This may be because multiple measurements better reflect the patients’ true status by adjusting for regression to the mean. Standard clinical practice of asking patients to return at a later time for re-assessment should be applied to research when the prognosis of LBP patients is to be evaluated.