Compare the difference between prognostic scoring systems in ICU patients.

• Main Authors: Yi-Ting He, 何依婷
• Other Authors: Gau-Jun Tang
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/75552586243412459447

碩士 === 國立陽明大學 === 醫務管理研究所 === 103 === BACKGROUND：Patients in intensive care units are under severe condition and their survival rate vary with time. We often use APACHE II scoring system to predict patient’s mortality rate in ICU, but those score can’t be attainable by secondary databases, like administrative data. With administrative data we can get the comorbidity index to predict general in-patients’ prognosis and survival rate. There are huge different between general in-patients and patients in ICU, so that only using comorbidity index to predict prognosis and survival rate of patient in ICU maybe biased; hence, we should take physiological changes into consideration. OBJECTIVE：Comparison of difference between comorbidity measures and APACHE II scores for predicting of patient outcome in intensive care units. METHOD：Collecting medical records from one particular hospital with one year. Calculating the APACHE II score and comorbidity index from those medical records. Comparing the ICU patients’ mortality in different period of time by different model. Considering the APACHE II score as the gold standard, we use logistic regression to find out what extra information we need to combine with comorbidity index, which can make us get closer to APACHE II. Finally, we use ROC curve to compare the difference in prediction. RESULTS：In order to predict the outcome of patients in ICU, the one of single-use CCI, Elixhauser Comorbidity Index have poor ability to predict. Our reaserch find CCI, Elixhauser Comorbidity Index combining with administrative data approach to predict the outcome of patients in ICU, and the predictive ability is better when using the CCI combining with information available admission data or the Elixhauser index approach. The CCI with information of age, gender, using mechanical ventilation revealed c-statistics of 0.773 (95% CI 0.744-0.803) for in-hospital mortality, 0.782 (95% CI 0.755-0.809) for 30-day mortality, and 0.775 (95% CI 0.751-0.799) for 1-year mortality. The prediction ability is similar with APACHE II (c-statistics=0.798、0.805、0.766). CONCUSION：In order to predict the outcome of patients in ICU, CCI and Elixhauser Comorbidity Index have poor ability to predict. Its predictive ability can be improved while combining with administrative data. Using the CCI or the Elixhauser Comorbidity Index combining with administrative data can increase prediction ability to predict the outcome of patients in ICU.