UTILIZATION OF MORTALITY PROBABILITY-MODE LS (MPM-II) TO EVALUATE THEEFFECTIVENESS OF PRIMARY ATTENTION TO PATIENTS CRITICALLY ILL

BACKGROUND: The performance of the Mortality Probability Models (MPM I I) has been assessed in Intensive Care Units (ICUs) in Catalonia and t he Balearic Islands. The MPM II system has been customized to that geo graphic area and quality performance has been evaluated in each ICU. M ETHODS: 1,270 adult critical patients, consecutively admitted in 16 IC Us from Catalonia and 1 from the Balearic Islands have been included, Probability of dying in the hospital has been calculated at admission in the ICU and at 24 hours using the models MPM II0 and MPM II24. Good ness-of-fit of the MPM II system in the overall group of 17 ICUs has b een analyzed. Logistic regression has been used to customize the MPM I I system to all the ICUs together. A Quality Performance Index (QPI) f or each ICU has been obtained by dividing the number of the observed d eaths by the number of deaths expected by the MPM II system. RESULTS: The overall QPI was 1.15 when using the MPM II0 and 1.17 when using th e MPM II24. The QPI in the 17 ICUs ranged from 0.58 to 2.05. Three ICU s showed excess of mortality and 2 ICUs had less deaths than expected. The process of customization of MPM II to the 17 ICUs as a group impr oved the estimation of expected mortality. CONCLUSIONS: The use of sev erity indixes allows to compare the outcome of patients in the ICU and provides an indicator of quality of care. The excess of mortality obs erved in some ICU should produce a watchful follow-up of outcome. Risk factors for excess of mortality should be studied.