Discrepancy between admission and discharge diagnoses as a predictor of hospital length of stay

Abstract

The addition of clinical data or present on admission (POA) codes to administrative databases improves the accuracy of predicting clinical outcomes, such as inpatient mortality. Other POA information may also explain variation in hospital outcomes, such as length of stay (LOS), but this potential has not been previously explored. To assess whether a discrepancy between the diagnosis coded at the time of admission and the diagnoses coded at discharge independently explains variation in LOS for general internal medicine patients. A retrospective data review of patients age 18 years and older admitted to general internal medicine units at a large, urban academic medical center between July 2005 and June 2006. A generalized linear regression model was constructed to adjust for patient factors known to be associated with LOS. Average LOS among patients with a discrepancy between the admitting and discharge diagnosis codes versus those patients without a discrepancy. The average LOS for patients without a discrepancy between the admitting and discharge diagnosis codes, adjusted for comorbid conditions, was 3.4 days compared to 4.2 days with a discrepancy (0.76-day increase; P < 0.01). Diagnosis discrepancy is associated with longer LOS. Diagnosis discrepancy on admission may be a marker of diagnosis uncertainty or poor patient assessment/documentation. Further research is needed to understand the underlying reasons for this discrepancy and its association with LOS, and, potentially, clinical outcomes.