OBJECTIVE
De novo infections of the spine are an increasing healthcare problem. The decision for nonsurgical or surgical treatment is often made case by case on the basis of physician experience, specialty, or practice affiliation rather than evidence-based medicine. To create a more systematic foundation for surgical assessments of de novo spinal infections, the authors applied a formal validation process toward developing a spinal infection scoring system using principles gained from other spine severity scoring systems like the Spine Instability Neoplastic Score, Thoracolumbar Injury Classification and Severity Score, and AO Spine classification of thoracolumbar injuries. They utilized an expert panel and literature reviews to develop a severity scale called the "Spinal Infection Treatment Evaluation Score" (SITE Score).
METHODS
The authors conducted an evidence-based process of combining literature reviews, extracting key elements from previous scoring systems, and obtaining iterative expert panel input while following a formal Delphi process. The resulting basic SITE scoring system was tested on selected de novo spinal infection cases and serially refined by an international multidisciplinary expert panel. Intra- and interobserver reliabilities were calculated using the intraclass correlation coefficient (ICC) and Fleiss’ and Cohen’s kappa, respectively. A receiver operating characteristic analysis was performed for cutoff value analysis. The predictive validity was assessed through cross-tabulation analysis.
RESULTS
The conceptual SITE scoring system combines the key variables of neurological symptoms, infection location, radiological variables for instability and impingement of neural elements, pain, and patient comorbidities. Ten patients formed the first cohort of de novo spinal infections, which was used to validate the conceptual scoring system. A second cohort of 30 patients with de novo spinal infections, including the 10 patients from the first cohort, was utilized to validate the SITE Score. Mean scores of 6.73 ± 1.5 and 6.90 ± 3.61 were found in the first and second cohorts, respectively. The ICCs for the total score were 0.989 (95% CI 0.975–0.997, p < 0.01) in the first round of scoring system validation, 0.992 (95% CI 0.981–0.998, p < 0.01) in the second round, and 0.961 (95% CI 0.929–0.980, p < 0.01) in the third round. The mean intraobserver reliability was 0.851 ± 0.089 in the third validation round. The SITE Score yielded a sensitivity of 97.77% ± 3.87% and a specificity of 95.53% ± 3.87% in the last validation round for the panel treatment decision.
CONCLUSIONS
The SITE scoring concept showed statistically meaningful reliability parameters. Hopefully, this effort will provide a foundation for a future evidence-based decision aid for treating de novo spinal infections. The SITE Score showed promising inter- and intraobserver reliability. It could serve as a helpful tool to guide physicians’ therapeutic decisions in managing de novo spinal infections and help in comparison studies to better understand disease severity and outcomes.