Patients with low back pain had distinct clinical course patterns that were typically neither complete recovery nor constant pain. A latent class analysis of longitudinal data

The clinical presentation and outcome of patients with nonspecific low back pain (LBP) are very heterogeneous and may be better understood by the recognition of reproducible subgroups. One approach to subgrouping is the identification of clinical course patterns (trajectories). However, it has been unclear how dependent these trajectories are on the analytical model used and the pain characteristics included. Purpose: To identify LBP trajectories using LBP intensity and frequency measured once a week over 1 year and compare the results obtained using different analytical approaches. Study design: A prospective observational cohort study. Patient sample: Patients presenting with nonspecific LBP to general practitioners and chiropractors. Outcome measures: Weekly self-report of LBP intensity (0-10) and the number of LBP days measured by short message service cell phone questions over a 1-year follow-up period. Methods: Latent class analysis was used to identify the trajectories of LBP and 12 different analytical models were compared. The study was a component of a broader study funded by an unrestricted grant from the Danish Chiropractors' Foundation (USD 370,000). Results: The study included 1,082 patients. The 12 models resulted in 5 to 12 subgroups, with a number of trajectories stable across models that differed on pain intensity, number of LBP days, and shape of trajectory. Conclusions: The clinical course of LBP is complex. Most primary care patients do not become pain-free within a year, but only a small proportion reports constant severe pain. Some distinct patterns exist which were identified independently of the way the outcome was modeled. These patterns would not be revealed by using the simple summary measures traditionally applied in LBP research or when describing a patient's pain history only in terms of duration. The appropriate number of subgroups will depend on the intended purpose of subgrouping