Of 1917 eligible consulting patients, 1104 agreed to participate in the study. Of these, 974 subsequently returned a completed baseline questionnaire (adjusted response rate: 51%). The median age of participants was 47 years (interquartile range: 38–56 years), 577 (59%) were female, and 608 (62%) were known to be in paid employment. Pain severity (visual analogue scale) and disability (Roland and Morris Disability Questionnaire) were moderate: median (interquartile range): 30 mm (12–50 mm) and 8 (4–13), respectively. Nine-hundred and twenty-two participants returned a 3-month follow-up questionnaire (follow-up response rate: 95%), of whom 363 (39%) reported persistent disabling low back pain (hereafter referred to, more simply, as ‘low back pain’). The flow of patients through the study can be seen in .
Flow of patients through the study.
The relationship between demographic characteristics and low back pain at follow-up can be seen in . Neither age (quartiles) nor sex was associated with low back pain at follow-up. However, persons with a lower income were 50% more likely to report a poor outcome than other individuals (RR = 1.5; 95% CI = 1.3 to 1.9). Similarly, persons not currently in paid employment experienced an increased risk of low back pain at follow-up.
Predictors of persistent disabling low back pain at follow-up: demographic characteristics and coping styles.
A high score on the active coping scale was associated with neither an increase nor a decrease in the risk of low back pain at follow-up (). In contrast, persons who adopted high levels of passive coping behaviour experienced a threefold increase in the risk of poor outcome (RR = 3.0; 95% CI = 2.3 to 4.0).
After adjusting for age, sex and socioeconomic status, a number of episode-specific factors were identified that were significantly predictive of pain outcome (). There was a significant trend to suggest that patients who, at baseline, reported higher levels of pain intensity experienced a increase in the risk of low back pain at follow-up: persons who rated their pain at 80–100 mm were five times more likely to report low back pain at follow-up than those who rated their pain at 0–20 mm (RR = 5.1; 95% CI = 3.8 to 7.0). Similarly, patients with high levels of disability were at increased risk of poor outcome: (RR = 4.8; 95% CI = 3.3 to 6.8). Subjects reporting low back pain every day, and those whose low back pain episode had lasted over 3 months prior to GP consultation were also at increased risk.
Predictors of persistent disabling low back pain at follow-up: episode-specific factors and pain history.
Persons with a prior history of non-consulting low back pain were no more likely to report low back pain at follow-up than those with no history (). However, those who had previously consulted their GP with low back pain (although not within the 6 months prior to the baseline consultation) experienced a 70% increase in the risk of poor outcome (RR = 1.7; 95% CI = 1.3 to 2.3). Also, those with a history of other chronic pain symptoms were at significant increased risk (RR = 2.0; 95% CI = 1.7 to 2.4).
A multivariable prediction model was constructed into which age, sex, socioeconomic status and passive coping score were forced. Because of the absence of any association between active coping and low back pain at follow-up, this variable was excluded from further analysis. Six episode-specific factors or variables relating to pain history entered the final model. Even after adjusting for these variables, passive coping significantly contributed to the final model (Wald χ2 = 7.87; P = 0.048); persons with high passive coping score experienced a significant increase in the risk of poor outcome (RR = 1.5; 95% CI = 1.1 to 2.0; , Model 1). The final model was highly discriminatory with respect to low back pain prognosis: among patients with none of the factors in the final model (at baseline), only 5% reported persistent disabling low back pain at 3 months. In contrast, among those with all seven factors, 87% reported a poor outcome.
Predictors of persistent disabling low back pain at follow-up: multivariable model.
Of the 1917 eligible consulting patients, only 974 returned a completed baseline questionnaire. Participants were more likely to be female (59%, versus 49% of non-participants; χ2 = 20.7; P<0.001) and were slightly older (median age = 47 years and 40 years, respectively; Mann–Whitney Z = 9.63; P<0.001). However, weighting the analysis back to the age/sex distribution of the original target population did not alter the multivariable model to any great extent (, Model 2) and did not change the study interpretation. Persons with high passive coping score still experienced a significantly elevated risk of low back pain at follow-up (RR = 1.4; 95% CI = 1.01 to 1.9).