A diagnostic classification for lumbar spine registry development

Background context Low back pain (LBP) is a symptom, not a diagnosis. The failure to differentiate underlying diagnoses in patients complaining of LBP is one of the primary reasons that studies examining treatments for LBP have yielded inconsistent results. To design a lumbar spine registry such that the accumulated data provide applicable guidance for clinical treatment, the incorporation of a functional diagnostic matrix is critical. Purpose To propose a clinically relevant diagnostic classification scheme, simple enough for use in clinical practice but granular enough to differentiate characteristics that impact clinical outcome. Study design Case-based development, feasibility, and reliability testing of a classification scheme. Patient sample Thirty case histories were compiled. Each case consisted of a brief clinical history with physical examination findings as well as pertinent radiographic images, including computed tomography scan and magnetic resonance image when available. Outcome measures Kappa values for inter- and intrarater reliability. Methods Thirty-six physicians were asked to provide a three-digit diagnostic code, specifying Symptoms, Structural Pathology, and Compressive Pathology for each case. The cases were then randomly rearranged and sent back to the physicians 2 weeks later for a second review. Inter- and intraobserver reliability was calculated using Randolph's free-marginal multirater kappa. Symptoms are classified differentiating between patients with back or leg pain dominance and those with equal back and leg pain; symptoms are also divided as acute versus chronic. Additional categories denote neurogenic claudication and cauda equina. Structural Pathology includes an option for age appropriate changes. Additional options include disc pathology with and without disc space collapse; spondylolisthesis or spondylolysis without olisthesis; and regional spinal deformity, either scoliosis or kyphosis. The remaining structural categories are primary facet pathology and nonunion after attempted fusion. Compressive Pathology includes a category indicating the absence of any study on which to judge compressive pathology and a category indicating an available study, but the lack of any clinically relevant compressive pathology. Additional options are for either central compression or lateral recess/foraminal/extraforaminal compressive pathology of any etiology. There is also a category for combined central and lateral compression. The final category is recurrent compression, central, lateral, or combined compression after surgical treatment at the involved level. It is important to emphasize that the intention is only to categorize Structural and Compressive Pathology that is relevant to the patient's symptom complex. When more than one relevant structural or compressive lesion exists, the provider should select the most clinically relevant finding. Results The interrater agreement was substantial for Symptoms (κ=0.70) and moderate for Structural Pathology (κ=0.58) and Compressive Pathology (κ=0.53). The intrarater agreement was substantial for Symptoms (κ=0.78), Structural Pathology (κ=0.70), or Compressive Pathology (κ=0.67). Conclusions This study demonstrates that improved diagnostic stratification of lumbar spine disorders is a feasible goal. The diagnostic coding matrix, based on clinically relevant descriptors, yielded substantial interrater consistency for symptoms, moderate interrater consistency for structural and compressive pathology, and substantial intrarater consistency for all elements.