Objectives: The STarT Back Screening Tool (SBST) is validated to subgroup primary care patients with back pain into risk groups relevant to initial decision-making. However, it remains unclear how the tool's allocation of individuals compares with subjective clinical decision-making. We evaluated agreement between clinicians and the SBST's allocation to risk subgroups, and explored reasons for differences observed.
Methods: Twelve primary care back pain patients underwent a video-recorded clinical assessment. The SBST was completed on the same day. Clinical experts (3 general practitioners, 3 physiotherapists, and 3 pain management specialists) individually reviewed the patient videos (4 each), blind to SBST allocation. Their task was to subgroup patients into low, medium, or high-risk groups.
Results: Interrater agreement between clinicians was "fair" (κ=0.28), with consistent allocation between experts in 4 of 12 patients. There was observed agreement with the SBST in 17 of 36 cases (47%) and Cohen's weighted κ was 0.22, indicating fair agreement. Two reasons for differences emerged. Clinicians tailor their decisions according to patient expectations and demands for treatment and clinicians use knowledge of difficult life circumstances that may be unrelated back pain.
Discussion: Clinicians make inconsistent risk estimations for primary care patients with back pain when using intuition alone, with little agreement with a formal subgrouping tool. Unlike clinicians, the SBST could not make a sophisticated synthesis of patient preferences, expectations, and previous treatment history. Although acknowledging the limitations of back pain subgrouping tools, more research is needed to test whether their use improves consistency in primary care decision-making.