A new classification of spondylodiscitis: possibility of validation and multidisciplinary expert consensus

Abstract

Objective. To perform validation study of the E. Pola classification (2017) and to assess expert consensus on the diagnosis and treatment of infectious lesions of the spine in the national professional community. Material and Methods. A clinical and radiological database on 15 cases of infectious spondylitis, as well as the information about original article by Pola and a Russian translation of the classification and tactical tables from this article, were distributed to 408 orthopedic traumatologists, neurosurgeons and radiologists who have experience in treating patients with spinal pathology and whose data are available in the registers of the relevant professional associations of the Russian Federation. The coincidence/difference in the responses concerning the definition of lesion types and the choice of treatment tactics, as well as proposals for the use of classification were assessed. Results. Answers were obtained from 37 respondents from 11 regions of the Russian Federation. The general interobserver agreement index (Fleiss kappa) for all types of spondylodiscitis was 0.388 (95 % CI 0.374–0.402), including for lesion types: type A – 0.480 (95 % CI 0.460–0.499, type B – 0.300 (95 % CI 0.281–0.320), and type C – 0.399 (95 % CI 0.380–0.419). Agreement levels were higher among radiologists (type A – 0.486, type B – 0.484, and type C – 0.477), orthopedic traumatologists (type A – 0.474, type B – 0.380, and type C – 0.479), and specialists with clinical experience less than 10 years (type A – 0.550, type B – 0.318, and type C – 0.437). The pooled data for all 12 lesion subtypes showed general poor agreement (k = 0.247, CI 0.240–0.253), satisfactory level was found for B3.2 type (k = 0.561, CI 0.542–0.581), good agreement (k > 0.61) was achieved between orthopedic traumatologists for type B3.2 and between radiologists for B3.1 and B3.2 lesion types. Respondents refused to use basic treatment options for type A in 15.1 %, type B in 7.5 % and type C in 3.2 % of answers, while indicating the need for interventions through anterior approach in 24.7 %, 43.0 % and 46.2 %, respectively. Limitations of the classification use depending on the localization and etiology of spondylitis were noted. Authors recommended taking into account the presence of systemic inflammatory response syndrome, mandatory CT scanning, clarification of spinal instability criteria, and the addition of anterior surgical interventions to the treatment algorithm. Conclusion. The Pola classification of spondylodiscitis is currently considered the most successful for tactical algorithms and implementation in broad clinical practice for spondylodiscitis. However, at the stages of its clinical application, there is an unsatisfactory interobserver expert consensus on the types of lesions, and there are limitations related to the etiology, localization and severity of the disease. A modified classification taking into account the identified limitations and including anterior procedures in the tactical options is advisable.

Keywords

• spondylodiscitis
• vertebral osteomyelitis
• classification
• validity
• interobserver agreement
• consensus
• treatment tactics