RESUMEN
BACKGROUND: Surgeons utilize a combination of preoperative tests and intraoperative findings to diagnose periprosthetic joint infection (PJI); however, there is currently no reliable diagnostic marker that can be used in isolation. The purpose of our study is to evaluate the utility of frozen section histology in diagnosis of PJI. METHODS: Retrospective analysis of 614 patients undergoing revision total joint arthroplasty with frozen section histology from a single institution was performed. Discriminatory value of frozen section histology was assessed using univariate analysis and evaluation of area under the curve (AUC) of a receiver operating characteristic curve comparing frozen section histology results to the 2018 International Consensus Meeting (ICM) PJI criteria modified to exclude the histology component. RESULTS: The sensitivity of the frozen section histology was 53.6% and the specificity was 95.2%. There was 99.2% concordance between the permanent section and frozen section results. The receiver operating characteristic curve for frozen section yielded an AUC of 0.744 (95% confidence interval 0.627-0.860) and the modified ICM score yielded an AUC of 0.912 (95% confidence interval 0.836-0.988) when compared to the full score. The addition of frozen section histology changed the decision to infected in 20% of "inconclusive" cases but less than 1% of total cases. CONCLUSION: In comparison to the modified ICM criteria, intraoperative frozen section histology has poor sensitivity, strong specificity, and acceptable overall discrimination for diagnosing PJI. This test appears to be of particular value for patients deemed "inconclusive" for infection using the remaining ICM criteria.
