Comparison Of APACHEâ…¡ And APACHEâ…£ In Geriatric Respiratory Intensive Care Unit

Objective In our article, we aim to assess and compare the predicted abilities ofAcute Physiology and Chronic Health Evaluation (APACHE) Ⅱ and Acute Physiologyand Chronic Health Evaluation (APACHE) Ⅳsystem in geriatric respiratory intensivecare unit (GRICU) patients. In predicting respiratory diseases compare the performanceof different diagnostic classification of these two models.Methods All elderly patients with respiratory disease admitted into the GRICU ofShaanxi Provincial People,s Hospital from June2013to December2013were admittedin the study. Patients met the inclusion criteria line. We collected the worst clinicalresults of patients within1day after admitted GRICU, respectively calculated the scoresand mortality probabilities with both APACHEⅡ and APAHCEⅣ rating software. Thepredicted scores、mortality and ability were compared and analyzed. SPSS18.0statisticcalculated and analyzed clinical data of patients. Forecasting performance of twodifferent models of discrimination and calibration were assessed. Discrimination of theprognostic models was assessed by using the area under the receiver operatingcharacteristic curve and its95%condidence interval. We used Hosmer-Lemeshowgoodness-of-fit χ2statistic and by SMR and its95%confidence interval to estimatemodel calibration.Results The study enrolled129critical ill patients. Death group has38cases,91cases of group survival. Observed hospital mortality probability was29.45%in this study.Both predicted mortality were APACHEⅡ (35.26%) and APAHCEⅣ (30.53%). Thescores of APACHEⅡ or APAHCEⅣ systems and disease outcome have a positivecorrelation relationship. The resulting score with mortality and severity was positively correlated. Comparison each group of actual and predicted mortality showed outoverestimated or underestimated. The average scores of APACHEⅡ were (16.72±6.04)in survival group and (28.34±9.12) in dead group, the difference between scores isstatistically significant (P＜0.05). The average scores of APAHCEⅣ were (59.64±17.00)in survival group and (101.84±37.42) in dead group, also the difference between scores isstatistically significant(P＜0.05). The AUROCC and its95%CI of the APACHEⅡ andAPAHCEⅣ for prediction of hospital mortality were0.85(0.78～0.93) and0.87(0.79～0.94) respectively in all patients. Discrimination was generally good for two models. Thesensitivity (55.30%、65.40%，)、specificity (94.50%、92.30%)、 Kappa value(0.55、0.63)、and Youden’s index(0.48、0.67) of APACHEⅡ and APAHCEⅣ respectively. TheHosmer-Lemeshow statistics for models were6.36and14.30for APACHEⅡ andAPAHCEⅣ, each of the P value＞0.05; Standardized mortality ratio (SMR95%CI) fortow models were0.8495%CI(0.576～1.113) and0.970，95%CI(0.664～1.280), both ofparameter indicating good calibration.To assess the capability of both models predicted mortality for different diagnosticcategories of respiratory diseases. The AUC and its95%CI of two models for threedifferent diagnostic categories diseases in GRICU, respectively sever pneumonia(SP)were0.75(0.56～0.93)、0.78(0.58～0.97); acute exacerbation of chronic obstructivepulmonary disease(AECOPD) were0.76(0.52～0.99)、0.56(0.34～0.77); mechanicalventilation were0.89(0.81～0.97)、0.89(0.80～0.98). The respective SMR (95%CI) ofAPACHEⅡ for mechanical ventilation was0.99(0.59～1.38), for AECOPD0.66(0.13～1.20), for SP0.5795%(0.19～0.94). The respective SMR (95%CI) ofAPACHEⅣ for mechanical ventilation1.01(0.60～1.41), for AECOPD0.94(0.18～1.70),for SP0.7795%CI(0.26～1.28). The difference of actual/predict mortality was nostatistical significance. The DIS and CAL of two rating models for mechanicalventilation were better than SP, which were better than AECOPD.Conclusion The APACHEⅡ and APAHCEⅣ score systems both had high accuracyin predicting the severity and mortality of patients in geriatric respiratory intensive careunit, however, death rates of two models in low scores group were overestimated, high score group were underestimated. The APACHEⅡ and APAHCEⅣ had very gooddiscrimination and calibration, APAHCEⅣ did better performance than APAHCEⅡ. Ourdata demonstrate that the performance of APACHEⅡ and APAHCEⅣ for mechanicalventilation were better than SP, which were better than AECOPD. The calibration ofAPAHCEⅣ for AECOPD was poor. In high risk groups, predicted death rates byAPACHEⅡ and APAHCEⅣ were similar to observed.