Background Stroke is the second leading cause of death worldwide and the third in Europe. Disease’s outcomes can be analyzed to profile hospital performance after appropriate adjustment. The aim of the study is to predict the mortality for hospitalized ischemic stroke using several severity measurement systems. Methods We used data from the hospital discharge records (HD) of a retrospective cohort of stroke patients (2005-2006) at Hospitals of Tuscan Region in Italy. The outcomes considered in the study were in-hospital mortality and mortality at 30 or 90 days after admission. Three Risk Adjustment tools were adopted to predict the outcomes: All-Patient Refined Diagnosis Related Groups (APR-DRG) system, that basis on information of HD considered, Charlson Index (CI) and Elixhauser Index (EI), which take into account the admitting diagnosis of the past three years. Logistic regression models were applied for the analysis of the performance of the three models. C statistic was used to define discriminative ability. Results The number of HD studied was 13.939. The in-hospital mortality was 10,2%, while 30 and 90 days mortality were respectively 12,1 and 17,4%. Female gender was found to be a significantly risk factor for in-hospital mortality in all the three models (OR 1,18 in CI, OR 1,17 in EI, OR 1,14 in APR-DRG; p < 0,05), such as increasing age was associated with higher risk of in-hospital, 30 and 90-day mortality. Both the CI and the APR-DRG risk of death were predictors of all outcomes considered. Even if the ability to predict mortality proved to be satisfactory in all three risk adjustemt algorithms, the performance of the APR-DRG model appeared to be better for all outcomes considered (C statistic 0.76 vs range 0,69-0,72 for CI and range 0,71-0,74 for EI). Conclusions The findings showed that administrative data (HD) are a reasonable resource to measure the patients’ severity and then to predict mortality. Although the APR-DRG model showed a slightly better performance than the others, it has an economic cost, while CI and EI are free.

Collini, F., Messina, G., Forni, S., Prisco, G., Ierardi, F., Quercioli, C., et al. (2013). Risk adjusted mortality predictive models in ischemic stroke. EUROPEAN JOURNAL OF PUBLIC HEALTH, 23, 285-285 [10.1093/eurpub/ckt124.119].

Risk adjusted mortality predictive models in ischemic stroke

MESSINA, GABRIELE;PRISCO, GABRIELLA;QUERCIOLI, CECILIA;NANTE, NICOLA
2013-01-01

Abstract

Background Stroke is the second leading cause of death worldwide and the third in Europe. Disease’s outcomes can be analyzed to profile hospital performance after appropriate adjustment. The aim of the study is to predict the mortality for hospitalized ischemic stroke using several severity measurement systems. Methods We used data from the hospital discharge records (HD) of a retrospective cohort of stroke patients (2005-2006) at Hospitals of Tuscan Region in Italy. The outcomes considered in the study were in-hospital mortality and mortality at 30 or 90 days after admission. Three Risk Adjustment tools were adopted to predict the outcomes: All-Patient Refined Diagnosis Related Groups (APR-DRG) system, that basis on information of HD considered, Charlson Index (CI) and Elixhauser Index (EI), which take into account the admitting diagnosis of the past three years. Logistic regression models were applied for the analysis of the performance of the three models. C statistic was used to define discriminative ability. Results The number of HD studied was 13.939. The in-hospital mortality was 10,2%, while 30 and 90 days mortality were respectively 12,1 and 17,4%. Female gender was found to be a significantly risk factor for in-hospital mortality in all the three models (OR 1,18 in CI, OR 1,17 in EI, OR 1,14 in APR-DRG; p < 0,05), such as increasing age was associated with higher risk of in-hospital, 30 and 90-day mortality. Both the CI and the APR-DRG risk of death were predictors of all outcomes considered. Even if the ability to predict mortality proved to be satisfactory in all three risk adjustemt algorithms, the performance of the APR-DRG model appeared to be better for all outcomes considered (C statistic 0.76 vs range 0,69-0,72 for CI and range 0,71-0,74 for EI). Conclusions The findings showed that administrative data (HD) are a reasonable resource to measure the patients’ severity and then to predict mortality. Although the APR-DRG model showed a slightly better performance than the others, it has an economic cost, while CI and EI are free.
2013
Collini, F., Messina, G., Forni, S., Prisco, G., Ierardi, F., Quercioli, C., et al. (2013). Risk adjusted mortality predictive models in ischemic stroke. EUROPEAN JOURNAL OF PUBLIC HEALTH, 23, 285-285 [10.1093/eurpub/ckt124.119].
File in questo prodotto:
File Dimensione Formato  
238+copertina.pdf

non disponibili

Tipologia: Abstract
Licenza: NON PUBBLICO - Accesso privato/ristretto
Dimensione 933.4 kB
Formato Adobe PDF
933.4 kB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11365/46457
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo