Evaluation of Combination Models Predicting Longterm Hospitalization and Complication Occurrence in Deep Neck Infection.

Background: The purpose of this retrospective study is to identify independent factors and their combinations that could predict complications and long-term hospitalization in deep neck infection (DNI) patients. Methods: We analyzed data from 169 patients that were hospitalized from 2012 to 2022 due to deep neck infections. In addition to evaluating patient characteristics, we identifed independent signifcant variables and their combinations that could predict complication occurrence and/or long-term hospitalization using multivariate analysis. Results: According to the multivariate backward procedure, while only presence of comorbidity and neutrophil to lymphocyte rate (NLR) were found to be independent signifcant risk factors of complication (P=.011 and P=.020, respectively), site of origin (non-odontogenic), need for both medical and surgical treatment, C-reactive protein (CRP) and NLR were found to be signifcant risk factors predicting long-term hospitalization (P < 0.001, P < 0.001, P=0.037 and P=0.008, respectively). The discriminatory power of a combination of 3 variables (presence of comorbidity, NLR, and white blood leukocyte count) for the occurrence of DNI complications yielded an AUC (ROC) of 0.764. The discriminatory power of a combination of 4 variables (non-odontogenic origin, need for both surgical and medical treatment, CRP, and NLR) to identify DNIs requiring prolonged hospital stay yielded an AUC (ROC) of 0.900. Conclusion: The fnal models obtained by the combination of variables for both complications and long-term hospitalization are promising for prognostic purposes. [ABSTRACT FROM AUTHOR]