Are panels of clinical, laboratory, radiological, and microbiological variables of prognostic value in deep neck infections? An analysis of 301 consecutive cases

Background: Deep neck infections (DNIs) are often clinically challenging, and may be life-threatening. Objective: The present retrospective study aimed to identify panels of clinical, laboratory, radiological, and microbiological parameters that could identify patients with DNIs at higher risk of complications or long-term hospitalization. Materials and methods: The investigation concerned 301 consecutive patients with DNIs treated at our institution between 2000 and 2014. Results: The discriminatory power of a combination of two variables (unknown origin of the infection and the need for surgical treatment) in terms of deep neck infection complications occurrence featured an AUC (ROC) of 0.6701. The power of a panel of four variables (age, leukocyte count, need for surgical treatment, days elapsing from hospitalization to surgical procedure) to identify DNIs necessitating long-term hospitalization featured an AUC (ROC) of 0.7929. Conclusion: Using the scale proposed by Hosmer and Lemeshow, the four-variable panel showed an amply acceptable, nearly excellent discriminatory power for long-term hospitalization. Significance: Although this panel achieved promising results for prognostic purposes, other parameters potentially capable of predicting the outcome of DNIs and orienting treatment decisions need to be investigated.