Comparison Of Different Predictive Rules For Assessing Severity Of Community Acquired Pneumonia

Objectives: The present study was undertaken to compare the performance of two foreign predictive rules (the pneumonia severity index, PSI and CURB-65) with the severity criteria of the Chinese guidelines for community-acquired pneumonia (CAP) for validation of the Chinese guidelines for community-acquired pneumonia in evaluating hospitalization criterias, intensive care unit(ICU)admission and hospital mortality.Methods: We retrospectively analyzed data from patients admitted with CAP to the respiratory and ICU department of the First Affiliated Hospital, Medical College of Shantou University, from September, 2009 to October, 2010. Patients were stratified according to PSI, CURB-65 severity score and the severity criteria of the Chinese guidelines. We analyzed the correlations between the severity criteria of the Chinese guidelines and PSI,CURB-65 severity score. Further more, patients were classified into three risk groups(low, moderate and high) in the light of each rule. We compared the mortalities of severe pneumonia according to different severity scoring systems between ICU and non-ICU treatment, between ICU and non-ICU to ICU treatment. constituent rates,hospital mortalities and intensive care unit (ICU) admission rates in each group of the predictive rules were analyzed. Sensitivity, specificity, positive predictive values, negative predictive values and the areas under the receiver operating characteristic curve (AUC) for predicting hospital mortality and ICU admission rate in each rule were compared too. Statistical differences in Length of Stay(LOS), cost, number of risk factors of the three subgroups stratified by the severity criteria of the Chinese guidelines were also analyzed.Results: (1) The correlation was significant between the severity criteria of the Chinese guidelines and PSI or betweeen the severity criteria of the Chinese guidelines and CURB-65 severity score (Pearson coefficient 0.639, 0.582; p<0.001). (2) In the light of PSI(classⅠ～Ⅲ) and CURB-65 (score 0～1), greater poportions of patients were classified as low risk (73.2%, 85.2%, respectively), while smaller proportions of patients were classified as moderate and high risk (16.2%, 28.6%; 10.6%, 6.3%, respectively). Both low and moderate risk patients identified by these two rules had mortalities (1.9%, 4.5%; 17.4%, 25.0%, respectively), also, both had ICU admission rates(6.7%, 10.7%; 43.5%, 66.7%, respectively). However, according to the severity criteria of the Chinese guidelines, smaller proportion of patients (25.7%) was identified as low risk, while greater proportion of patients were defined as moderate and high risk (45.1%, 29.2%, respectively). Nevertheless, no death or ICU admission occurred in the low risk group, hospital mortality and ICU admission rate were low in moderate risk group (1.6%, 2.3%, respectively). ICU admission rates and mortalities in the high risk were as follow: PSI 83.3%, 46.7%; CURB-65 94.4%, 50.0%; the severity criteria of the Chinese guidelines 67.5%, 28.9%. All severity score systems showed that in the high risk group, the mortality of patients who took ICU treatment was higher than non-ICU treatment, the mortality of patients who took ICU treatment right at admission was lower than patients who took ICU treatment after the failure of non-ICU treatment(p<0.05). The severity criteria of the Chinese guidelines had high sensitivity(92.3%, 94.9%, respectively) and low specificity (77.1%, 88.0%, respectively) for predicting hospital mortality and admission to ICU. On the contrary, The PSIⅤand CURB-65 score 3～5 all had high specificity(>90%, generally) but low sensitivity (28.8%～53.8%)for the two outcomes. All these rules had low positive predictive values (46.7%～84.4%) but high negative predictive values (84.2%～99.0%) for the two outcomes.The severity criteria of the Chinese guidelines had a greater discriminatory power to predict mortality and ICU admission (AUC: 0.847, 0.915, respectively). (3) The differences in Length Of Stay, cost, number of risk factors statistically significant between patients with severe pneumonia and patients who did not base on the hospitalization criterias specified by the severity criteria of the Chinese guidelines (p<0.05).Conclusion: PSI, CURB-65 scoring system were useful for evaluating severity of Chinese patients with CAP. The severity criteria of the Chinese guidelines differentiate patients with different risks, which had a greater discriminatory power to predict mortality and ICU admission than PSI or CURB-65. Therefor, the severity criteria of Chinese guidelines should be implemented. Further more, our research found that patients with severe pneumonia should take ICU treatment directly to reduce mortality during hospitalization.