380 - Hospitalized Heart Failure Incidence is Significantly Underestimated by Diagnosis-Related Group Codes

Abstract

Background: Hospitalized heart failure remains a major driver of cost, healthcare utilization, and morbidity. Diagnosis-related groups (DRGs) have commonly been used to estimate hospitalized heart failure incidence. Furthermore, payors use DRGs as the basis for novel reimbursement models, such as the Bundled Payments for Care Improvement (BPCI). However, DRGs group payments, not diagnoses, and thus are inherently limited as a tool for ascertainment of incidence. Therefore, relying upon DRG data alone could lead to significant underestimates of the burden of hospitalized heart failure. Objective: To prospectively compare the incidence of hospitalized heart failure as detected by DRG data to expert chart review of algorithmically detected cases. Methods: Hospitalized heart failure cases were detected using a sensitive enterprise data warehouse query, which identified patients based on clinical and diagnosis-related parameters, including BNP level and administration of intravenous diuretics. The EDW query was run daily; expert clinicians verified hospitalized heart failure based on comprehensive chart review. During the same study period, the number of patients coding into heart failure DRGs (291, 292, or 293) was compiled. Results: During the study period of 207 days, a total of 6915 charts were screened (33.4 charts per day). A total of 1010 patients were identified by chart review as having heart failure requiring active management. During the study period, only 334 patients (33%) received a heart failure DRG, while the remainder received 1 of 146 other DRGs. Of the patients not coding into a heart failure DRG, 362 received another cardiac DRG, of which 256 were procedural DRGs. Pulmonary DRGs accounted for 86/1010 cases (8.5%), while renal DRGs accounted for a further 70/1010 cases (6.9%). Conclusions: Hospitalized heart failure is a common clinical syndrome, but relying solely on DRG data leads to large underestimates of disease burden. Conversely, because of clinical co-morbidities and the intricacies of the DRG grouping algorithm, clinically apparent heart failure can code into an astonishing number and variety of DRGs. A more comprehensive estimate of hospitalized heart failure incidence has broad process of care, workflow, and public health implications.