Laboratory values improve predictions of hospital mortality

Abstract

Objective. To compare the precision of risk adjustment in the measurement of mortality rates using: (i) data in hospitals' electronic discharge abstracts, including data elements that distinguish between comorbidities and complications; (ii) these data plus laboratory values; and (iii) these data plus laboratory values and other clinical data abstracted from medical records. Design. Retrospective cohort study. Setting. Twenty- two acute care hospitals in St Louis, Missouri, USA. Study participants. Patients hospitalized in 1995 with acute myocardial infarction, congestive heart failure, or pneumonia (n=5966). Main outcome measures. Each patient's probability of death calculated using: administrative data that designated all secondary diagnoses present on admission (administrative models); administrative data and laboratory values (laboratory models); and administrative data, laboratory values, and abstracted clinical information (clinical models). All data were abstracted from medical records. Results. Administrative models (average area under receiver operating characteristic curve = 0.834) did not predict death as well as did clinical models (average area under receiver operating characteristic curve = 0.875). Adding laboratory values to administrative data improved predictions of death (average area under receiver operating characteristic curve = 0.860). Adding laboratory data to administrative data improved its average correlation of patient-level predicted values with those of the clinical model from r=0.86 to r=0.95 and improved the average correlation of hospital-level predicted values with those of the clinical model from r=0.94 for the administrative model to r=0.98 for the laboratory model. Conclusions. In the conditions studied, predictions of inpatient mortality improved noticeably when laboratory values (sometimes available electronically) were combined with administrative: data that included only those secondary diagnoses present on admission (i.e. comorbidities). Additional clinical data contribute little more to predictive power.