The occurrence of pneumonia diagnosis among neurosurgical patients: The definition matters

Abstract

Background Diagnosis of pneumonia in the hospital setting is a challenge due to the absence of a generally accepted pneumonia definition. Methods We prospectively evaluated six different clinical criteria for diagnosing pneumonia-physician's diagnosis (A), was compared to common methods, i.e., general surveillance method (B), consensus clinical criteria (C), and other three criteria (D, E, and F) among 390 neurosurgical patients treated in either the intensive care unit or the neurosurgical ward in a university hospital. Results The frequencies of pneumonia cases were: group A, 66 (16.9%); group B, 41 (10.1%); and group C, 55 (14.1%). Only 28 pneumonia cases were identified by all three criteria. The kappa values were: between A and B, 0.42 (95% confidence interval, CI; 0.27-0.57); between Aand C, 0.49 (95% CI; 0.35-0.63); and between B and C, 0.68 (95% CI; 0.55-0.80). In group A, there were 34 false positive cases according to criterion B and 25 according to criterion C. Pneumonia cases were identified according to criterion D in 13 (3.3%), E in 35 (9.0%), and F in 51 (13.1%) cases, respectively. Conclusions There was great variability in the presence of pneumonia among neurosurgical patients, depending on the criteria used. Our results support the idea that a more exact method for pneumonia diagnosis should be implemented to obtain more reliable results in this important infection of hospitalized patients, which is also used for benchmarking purposes. Furthermore, it seems important to treat all clinically suspected pneumonia cases whether or not the surveillance criteria are fulfilled. © 2011 Springer Science+Business Media, LLC.