Externally validating the Mortality In Spinal Infection-20 score: A retrospective cohort study in New Zealand

There have been several prognostic tools proposed for determining spinal epidural abscess patient outcomes. One such method is the "mortality in spinal infection" scoring system. In this study, we aim to externally validate this score within the New Zealand population.

We identified all patients who were admitted to a tertiary referral centre with a diagnosis of spinal epidural abscess between February 2009 and January 2022. The mortality in spinal infection score was calculated for all patients. We developed a receiver operating characteristic curve and a calibration plot to externally validate the score.

A total of 140 patients with spinal epidural abscesses met the criteria for inclusion in our study. Within the one-year follow-up period, 18 patients died while 122 survived. In total, 105 patients underwent surgical intervention, while 35 patients were managed nonoperatively. In total, 5/15 (33%) of patients with scores ≥11 died compared with 13/125 (10%) of those with scores <11 (odds ratio 4.3, 95% confidence interval [CI] 1.3-14.6, p=0.03). This threshold had high specificity (0.92) and negative predictive value (0.90) but lower sensitivity (0.28) and positive predictive value (0.33). The area under the curve of the receiver operating characteristics plot for mortality in spinal infection scoring and death (0.67, 95% CI 0.57-0.79) did not meet the prespecified criterion of an area under the curve >0.80 for acceptable prediction.

In this cohort, the mortality in spinal infection-20 score demonstrated limited ability to discriminate between those patients who died vs those who survived at one year post follow-up. Prospective multicentre data collection may improve the development of predictive tools.