Morbidity during hospitalization: Can we predict it?

Abstract

Physicians use the concept of stability to estimate the likelihood that a patient will deteriorate during a hospitalization. To determine whether physicians can accurately predict a patient's risk of morbidity, 603 patients admitted to the medical service during a one month period were rated prospectively as to how stable they were.

Overall, 15% of patients had deterioration of already compromised systems, while 17% had new complications, such as sepsis. Eight percent of patients had both. Twelve percent of stable patients experienced morbidity; 39% of the somewhat unstable and 61% of the most unstable. When all of the demographic and clinical variables were taken into account including the reason for admission and comorbid diseases, the residents' estimates of the patient's stability was the most significant predictor of morbidity (p < 0.001).

The judgment that a patient was stable had an 87% negative predictive accuracy, while the judgment unstable had a 46% positive predictive accuracy.