Predicting Persistent Disabling Low Back Pain in Veterans Affairs Primary Care Using the STarT Back Tool.
Background: The Subgrouping for Targeted Treatment (STarT Back) is a stratified care approach to low back pain (LBP) treatment. The predictive validity of STarT Back in Veterans Affairs (VA) primary care has not been demonstrated. Objective: To examine the validity of the STarT Back tool for predicting future persistent disabling LBP in VA primary care.
Design: Cohort study. Setting: VA primary care in Washington State. Participants: Veterans seeking care for LBP in VA primary care clinics. Interventions: Not applicable. Main outcome measures: The STarT Back tool was used to classify Veterans according to their baseline risk group (low vs medium vs high). The primary study outcome, persistent disabling LBP, was defined as a Roland-Morris Disability Questionnaire (RMDQ) score ≥ 7 at 6-month follow-up. Analyses examined discrimination and calibration of the baseline STarT Back risk groups for prediction of persistent disabling LBP at 6-month follow-up.
Results: Of the study sample, 9% were female and 80% reported longstanding LBP (>5 year duration). Among 538 participants, the baseline STarT Back risk groups were associated with future persistent disabling LBP at 6-month follow-up. Within each baseline STarT Back risk group, the proportions with future persistent disabling LBP at 6-month follow-up were 54% (low risk), 88% (medium risk), and 97% (high risk). The baseline STarT Back risk groups had useful discrimination (area under the curve [AUC] 0.79) for predicting future persistent disabling LBP, but the proportion of Veterans with persistent disabling LBP at 6-month follow-up was substantially higher than that observed in non-VA primary care settings.
Conclusions: The STarT Back risk groups had useful discrimination (AUC = 0.79) for future persistent disabling LBP, but calibration was poor, underestimating the risk of persistent disabling LBP. The STarT Back tool may require updating for use in VA primary care.