Multivariate approach to investigating prognostic factors in deep neck infections

Deep neck infections (DNI) spread along fascial planes and involve neck spaces. Very few studies have investigated potentially prognostic factors using multivariate statistical models. Our aim was to analyze 282 consecutive cases of DNI using multivariate (logistic) statistical models to identify independent significant factors influencing prognosis in terms of complications and long-term hospitalization (>6 days). In our series, only involvement of more than one neck space was independently significant in prognosticating complications of DNI (odds ratio [OR] 2.46). The presence of comorbidities (OR 2.13), non-odontogenic sites of origin (OR 1.88), leukocyte counts above 11.0 cells × 10(9)/L at presentation (OR 3.57), and the need for both medical and surgical treatments (OR 4.66) was significantly and independently prognostic of long hospital stays. Multivariate analysis can distinguish between risk factors and their relative contribution to outcome. The few published studies using multivariate models to analyze DNI prognosis considered quite large cohorts, but no clinical variables persistently revealed an independent significant prognostic role. This evidence seems to underscore the complex interdependence of several clinical variables in contributing to DNI prognosis, and the heterogeneity of the diagnostic/therapeutic approaches adopted.