Background: Diagnosis discrepancy, defined as an admitting ICD-9CM diagnostic code that does not match the discharge diagnosis code, is associated with longer length-of-stay (LOS). However, the association between diagnosis discrepancy, cost of care, and 30-day readmission rate has not been studied. In this study, we examined this association and also explored those clincal domains most likely to have a diagnosis discrepancy.
Methods: We used a retrospective, cross-sectional design. The study population was all patients admitted to general internal medicine units between July 2009 and June 2010. We excluded routine admissions for oncology patients, patients with surgical DRGs, and those with scheduled admissions. Patients were classified as discrepant if the 3 digit diagnosis code did not match. Outcome measures were: differences in LOS, total direct cost of care as calculated by the hospital’s cost accounting system, and 30-day readmission rate for discrepant and non-discrepant patients. We fit generalized linear regression models for direct costs and LOS and a binary logistic regression model for 30-day readmission, controlling for demographics, insurance status, admission source and number of comorbidities.
Results: Of the 2,390 patients fitting our inclusion criteria, 1,624 (68%) had a diagnosis discrepancy. Patients with discrepancy had longer LOS (4.2 days vs 3.9 days; p=0.058), cost of care ($4,627 vs $4,090; p=0.003), and 30-day readmission (18% vs 12%, p<0.001). The proportion of discrepant diagnoses was lowest for patients admitted with neoplastic admitting diagnoses (18%) and highest for respiratory admitting diagnoses (86%). When diagnoses were grouped by clinical category (i.e., digestive) we found that the clinical category changed often; from a low of 44% (digestive to other [i.e., digestive to respiratory]) to a high of 82% (endocrine to other [i.e., endocrine to infection]). The adjusted models indicated that diagnosis discrepancy was an independent predictor of all outcomes.
Conclusion: Diagnosis discrepancy is associated with increase LOS, cost of care and readmission at 30 days, and is independent of patient demographics, insurance status, and risk adjustment. Diagnosis discrepancy varies by clinical domain and specific diagnostic entities. Given the high proportion of changes in diagnosis and even clinical category between admitting and discharge diagnoses, our finding most likely reflects diagnostic complexity. Further exploration for causes of diagnosis discrepancy is needed as diagnosis discrepancy is statistically and clinically significantly associated with hospital practice outcome measures.
See more of: The 33rd Annual Meeting of the Society for Medical Decision Making