Povilas Masionis1, Valentinas Uvarovas2, Giedrius Mazarevičius2, Kiril Popov2, Šarūnas Venckus2, Karolis Baužys2, Narūnas Porvaneckas2. 1. Clinic of Rheumatology, Traumatology Orthopaedic and Reconstructive Surgery, Centre of Orthopedics and Traumatology, Medical Faculty, Vilnius University, Republican Vilnius University Hospital, Šiltnamių g. 29, LT-04130 Vilnius, Lithuania. Electronic address: povilasmasionis@yahoo.com. 2. Clinic of Rheumatology, Traumatology Orthopaedic and Reconstructive Surgery, Centre of Orthopedics and Traumatology, Medical Faculty, Vilnius University, Republican Vilnius University Hospital, Šiltnamių g. 29, LT-04130 Vilnius, Lithuania.
Abstract
INTRODUCTION: Successful treatment starts by accurate classification of pathology, but there is no conclusive, reliable and universally accepted method for classification of intracapsular femoral neck fractures. As a perfect classification should have high intra- and interobserver agreement, this study aims to access reliability of three classification systems: Garden, AO and simple II stage classification. MATERIALS AND METHODS: Four orthopaedic trauma surgeons (two of them professors) and two senior orthopaedic trauma residents were invited to evaluated 136 blinded anterior-posterior and lateral X-rays of patients with femoral neck fractures. Observers classified fractures according to IV stage Garden, AO and simple II stage classifications. The exercise was repeated after one month on same but randomised X-rays. Cohen kappa was used to measure inter- and intraobserver agreement. Fleiss kappa was used to access multi-rater agreement. RESULTS: AO classification showed an overall agreement of 0.22 (fair agreement). Garden classification had overall reliability slightly higher than AO, but matching same fair agreement group (0.33). II stage classification provided the highest estimates: from 0.35 (fair agreement) to 0.83 (almost perfect agreement) and multi-rater agreement of 0.50 (moderate agreement). There was seen no difference in intra- and interobserver agreement between observer groups (professors, trauma surgeons and trauma residents) DISCUSSION: All three classification systems showed equal adoption among differently experienced observer groups. Despite this finding, IV stage Garden and AO classifications should be avoided in clinical use because of poor reproducibility. Only simple II stage classification showed sufficient intra- and interobserver reliability. LEVEL OF EVIDENCE: IV, Retrospective study.
INTRODUCTION: Successful treatment starts by accurate classification of pathology, but there is no conclusive, reliable and universally accepted method for classification of intracapsular femoral neck fractures. As a perfect classification should have high intra- and interobserver agreement, this study aims to access reliability of three classification systems: Garden, AO and simple II stage classification. MATERIALS AND METHODS: Four orthopaedic trauma surgeons (two of them professors) and two senior orthopaedic trauma residents were invited to evaluated 136 blinded anterior-posterior and lateral X-rays of patients with femoral neck fractures. Observers classified fractures according to IV stage Garden, AO and simple II stage classifications. The exercise was repeated after one month on same but randomised X-rays. Cohen kappa was used to measure inter- and intraobserver agreement. Fleiss kappa was used to access multi-rater agreement. RESULTS: AO classification showed an overall agreement of 0.22 (fair agreement). Garden classification had overall reliability slightly higher than AO, but matching same fair agreement group (0.33). II stage classification provided the highest estimates: from 0.35 (fair agreement) to 0.83 (almost perfect agreement) and multi-rater agreement of 0.50 (moderate agreement). There was seen no difference in intra- and interobserver agreement between observer groups (professors, trauma surgeons and trauma residents) DISCUSSION: All three classification systems showed equal adoption among differently experienced observer groups. Despite this finding, IV stage Garden and AO classifications should be avoided in clinical use because of poor reproducibility. Only simple II stage classification showed sufficient intra- and interobserver reliability. LEVEL OF EVIDENCE: IV, Retrospective study.