Text

Objectives and questions: Despite the rising global incidence of primary spinal infections, optimal treatment algorithms remain uncertain. Prior classification attempts have been limited by regional bias or small consensus groups. The AO Spine Knowledge Forum therefore developed a comprehensive system integrating anatomic location, morphology, neurologic status, and systemic modifiers. This study aimed to internationally validate the AO Spine Primary Infection Classification by evaluating accuracy, reliability, and variation across region, experience, specialty, and practice setting.

Material and methods: An internal cohort comprised AO Spine Knowledge Forum surgeons, and an international cohort included AO Spine surgeons from diverse regions. Using expert-panel consensus as the gold standard, de-identified imaging and standardized vignettes from 19 primary spinal infection cases were independently classified in two rounds with the AO Spine Primary Infection Classification System. Accuracy was defined as exact agreement at the anatomic level (A/B/C) and across the 12 subcategories (A0–A3, B0–B3, C0–C3). Inter- and intra-rater reliability were assessed using Fleiss’ kappa.

Results: A total of 570 expert ratings were generated across two rounds, with 228 surgeons participating and 150 completing both. Respondents were mainly from Europe (41.2%), followed by Asia (22.8%), Africa/Middle East (15.4%), Central/South America (10.5%), and North America (10.1%). Most had 11–20 years of experience (36.7%), then 5–10 years (30.7%), >20 years (18.7%), and <5 years (14.0%). Practice settings included hospital-employed (48.7%), academic (36.0%), and private practice (15.4%), with most at level I trauma centers (64.9%); 56.6% were orthopaedic surgeons and 43.4% neurosurgeons. Round-1 accuracy vs the gold standard was highest in Central/South America (85.3%) and lowest in Asia (78.9%, p=0.01); in round 2 it was highest in North America (86.6%) and lowest in Central/South America (78.9%, p=0.046). Accuracy was highest in surgeons with 11–20 years’ experience (86.3%, 88.9%) and lowest in <5 years (77.0%, 75.5%; p<0.001). Private practice showed the highest accuracy (86.2–89.3%) without group differences (p=0.05), with no differences by trauma center level. Neurosurgeons showed slightly higher round-1 accuracy than orthopaedic surgeons (84.3% vs 81.3%, p=0.02), which disappeared in round 2 (85.9% vs 83.5%, p=0.05). Intrarater reliability did not differ by region, experience, practice, or specialty (p>0.05). Interrater reliability showed substantial agreement across regions (κ=0.63–0.72), experience (κ=0.60–0.74), practice settings (κ=0.66–0.74), and specialties (κ=0.66–0.71), consistent across rounds.

Discussion and conclusions: The AO Spine Primary Infection Classification showed high accuracy and strong reliability across international surgeons, independent of region, experience, or specialty, supporting its validity and global applicability. Further education and outcome studies are warranted.