Results
Study Sample
A total of 2,652 LBP patients above 18 years of age were registered in SpineData within the inclusion period of the study (56% females, mean age 50 years). From these 2,405 fulfilled the inclusion criteria and 1,752 (55% females, mean age 50 years) could be classified in the pre-defined subgroups (Figure 1).
(Enlarge Image)
Figure 1.
Flow chart from registration in clinical registry to 12-month follow-up. * Proportion of working population responding to the sick leave question.
Among those with LBP + NRI, 669 patients had more than one neurological sign. Of those with only one sign, 73 patients had only a positive straight leg raise, 45 had only reduced muscle strength, 135 had only altered sensation, and 18 had only impaired tendon reflexes.
Follow-up after 3 and 12 months was completed by 76% and 70% respectively, but response rates relating to sick leave were lower (Figure 1). Non-responders did not differ significantly from responders on LBP intensity, duration, activity limitation, depression or fear of movement at baseline. However, non-responders at 3-months follow-up were more often male (48% vs. 43%) and were on average 1.4 years younger as compared with the responders (p < .05). These differences between responders and non-responders did not differ significantly between subgroups. Non-responders at 12-month follow-up were on average 3.3 years younger (p < .05), and did not differ significantly from responders on other baseline factors. Also, non-response to the sick leave question did not differ across subgroups.
Baseline Characteristics
Patient self-reported characteristics are summarised in Table 1. Statistically significant differences across subgroups were observed for all measured baseline factors except fear of movement. Generally, those with Local LBP were the least severely affected and those with LBP + NRI had the most severe profile. The differences observed in duration indicated that patients with LBP + NRI were referred to the Department earlier than other patients, but even in that subgroup, many patients reported very long-lasting pain.
Associations Between Subgroups and Activity Limitation
Activity limitation from baseline to 12-month follow-up within the four subgroups is illustrated in Figure 2. Statistically significant associations were present between subgroups on change in activity limitation in both the unadjusted and adjusted analyses (Table 2). However, the residual variance was only slightly reduced by subgroups (R = .04). Patients with LBP + NRI improved more than other subgroups in pairwise comparisons, and the estimated effect of being in the LBP + NRI subgroup on the course of activity limitation was largely unaltered after adjusting for duration and other covariates.
(Enlarge Image)
Figure 2.
Mean RMDQ scores in four subgroups at baseline, 3 months, and 12 months. Activity limitation in four subgroups at the initial visit to the Department and over the clinical course.
Looking at absolute RMDQ scores, the subgroup with Local LBP had the least activity limitation at all time points and the LBP + NRI the most (Figure 2). Pairwise comparisons of absolute RMDQ scores adjusted for duration (Model II) were all significant (p < 0.05) except that LBP + pain above knee and LBP + pain below knee did not differ significantly at any time point. In Model III, LBP + NRI differed significantly from Local LBP at 3-month follow-up. At 12-month follow-up, significant differences existed between Local LBP and LBP + pain above knee and between Local LBP and LBP + NRI. However a very small proportion of the variance in activity limitation was explained by subgroups (R = .02 for 3- and 12-month analyses).
Associations Between Subgroups and Global Perceived Effect
At the 3-month follow-up, 31% of the cohort reported to be 'much better' or 'better'. This proportion varied across the subgroups from 23% in the LBP + pain below knee subgroup to 36% in the LBP + NRI subgroup. There were statistically significant associations between the subgroups and global perceived effect in Model I but the prognostic capacity in terms of AUC was low, and the association was not significant when duration was taken into account in Model II (Table 3). Therefore Model III was considered irrelevant. The LBP + NRI subgroup had higher odds of being 'much better' or 'better' as compared with the Local LBP and the LBP + pain below knee subgroups in pairwise comparisons (Table 3).
Associations Between Subgroups and Sick Leave
At baseline 1,003 (57%) of the participating patients were in the working population (Local LBP only 64%, LBP + above knee 62%, LBP + below knee 54%, and LBP + NRI 56%. P = 0.03). At the 3-months follow-up 29% (95% CI: 25–33%) of these were currently on sick leave, with the distribution in the subgroups ranging from 19% in the Local LBP subgroup to 35% in the LBP + NRI subgroup (p = 0.02). A larger proportion of patients in the LBP + NRI subgroup were on sick leave at 3 months and subgroups were significantly associated with sick leave in model I and model II but not in model III (Table 4).