Objective: We studied the utility of the Clinical Pulmonary Infection Score (CPIS) for the diagnosis of VAP in multidisciplinary medical-surgical-trauma and neurosurgical ICU compared with the CDC NHSN surveillance definition of VAP.
Methods: Over a period of 3 months, surveillance for VAP was carried out by applying both the CPIS and the NHSN definitions. A CPIS score of greater than 6 was considered to represent VAP. We used both the original (incorporating temperature, white count, tracheal secretions, arterial blood gas, chest radiograph and culture) and modified CPIS score (all the variables used in the original except culture). Agreement between the two methods was calculated using the kappa statistic.
Results: In the three month period of study, 18 cases of VAP were identified using CDC NHSN criteria. Applying the CPIS score to these cases, we found that CPIS score was concordant with NHSN definitions in 16 of the 18 cases (89%) for the original score and in 9/18 (50%) using the modified score (which does not incorporate microbiology). We also calculated the CPIS for those cases where VAP was initially suspected based on one of the CDC NHSN criteria but did not meet criteria for confirmed VAP. In these 18 cases, the original CPIS score was concordant with NHSN for 15 of the 18 (83%) and the modified CPIS score was concordant with NHSN for 18 of 18 cases (100%). The kappa statistic was 0.72 (9% CI 0.50- 0.95) for the comparison between original CPIS and NHSN denoting moderate to very good agreement and was 0.50 (95% CI 0.26-0.74) for the comparison between modified CPIS and NHSN, denoting fair to very good agreement between the two methods. Overall, the mean CPIS score for cases diagnosed as VAP by the NHSN criteria was 8.0 and was 3.9 for cases not confirmed to be VAP.
Conclusions: The original CPIS score has good correlation with CDC NHSN criteria for diagnosis of VAP. When microbiology is omitted, as in the modified CPIS score, the concordance drops considerably. Our results suggest that the CPIS score may represent a promising strategy for surveillance of VAP. Further research is essential to identify an optimal reference standard for VAP diagnosis.