Validation of the Calgary Spine Severity Score

Background context: The Calgary Spine Severity Score (CSSS) is a published triage score reported in the Spine Journal in 2010. It separates spine referrals into four time categories of urgency. It stratifies patients according to clinical, radiologic, and pathologic findings. The CSSS however still requires external validation at another institution and in an unselected sample of patients.

Purpose: The aim was to validate the CSSS.

Study design/setting: This was a validation study.

Patient sample: The sample included a total of 316 consecutive patients undergoing spinal surgery between April 2014 and September 2014 at a tertiary care hospital in Canada.

Outcome measures: The outcome was validity of the CSSS via its predicted time to operating room (OR) to predict actual time to OR.

Methods: We applied the CSSS to an unselected sample of consecutive patients from a tertiary care hospital between April 2014 and September 2014. Demographic and clinical data were collected. The CSSS was determined. We compared the time with OR predicted by the CSSS in one of four categories (routine>6 months=CSSS 3-5, priority<6 months=CSSS 6-8, urgent<1 month=CSSS 9-11, and emergent<1 week=CSSS 12-15) with the actual time to OR. We used Kaplan-Meier survival analysis to assess the CSSS predictive ability. Cox proportional hazard models were built and compared via analysis of variance to determine whether the models differed in their ability to fit the data.

Results: Three hundred sixteen patients were eligible. Two hundred eighty-nine had sufficient data. One hundred eighteen were a mismatch with the actual time to OR yielding an accuracy of 63%. The CSSS overestimated the urgency in 68 cases and underestimated it in 50 cases. Notably, seven cauda equina syndrome cases were classified as priority (<6 months) instead of emergent. The concordance was 0.70 and the R-square 0.33. We proposed several adjustments to the CSSS to increase its accuracy. The modified CSSS had an accuracy of 96%, overestimating nine cases and underestimating one case. The concordance was 0.77, and the R-square 0.70.

Conclusions: The modified CSSS is an easy-to-use triage score, which represents a substantial improvement as compared with the original CSSS. It now requires further external validation.