Study objective: To estimate the costs of hospitalization for neutropenia among chemotherapy-treated patients with newly diagnosed non-Hodgkin's lymphoma and to assess baseline patient factors associated with these costs.
Design: Retrospective cohort study.
Data source: Linked Surveillance, Epidemiology, and End Results Program-Healthcare Cost and Utilization Project databases for Iowa from 1993-1998.
Patients: Patients with newly diagnosed non-Hodgkin's lymphoma who received all inpatient care at Iowa hospitals during their first course of chemotherapy.
Measurements and main results: Neutropenia-related hospitalization costs were estimated from discharge abstracts found within the earliest of the following: 6 months after the diagnosis month, the date of bone marrow transplantation, or date of death. We performed univariate tests of differences in neutropenia-related hospitalization costs in all patients in the sample, as well as tests for neutropenia-related hospitalization costs, length-of-stay, and cost/inpatient day for patients with at least one hospitalization for neutropenia. We modeled total inpatient charges over the period for patients with at least one neutropenia-related hospitalization (multiple regression). A total of 1636 patients with non-Hodgkin's lymphoma had chemotherapy in Iowa and met inclusion criteria; of these, 316 had at least one hospitalization for neutropenia. The 316 patients had 418 stays. Patients with advanced stage (vs limited stage), previous anemia (vs no anemia), positive Charlson comorbidity score (vs score of 0), and diffuse large cell histology (vs follicular) had higher mean neutropenia-related hospitalization cost/patient with non-Hodgkin's lymphoma (p<0.05). Among those with neutropenia-related hospitalizations, a longer length of stay was associated with nonfollicular histologies, previous anemia, and positive Charlson score (p<0.05).
Conclusion: When estimating expected payments for neutropenia-related hospitalization in patients with non-Hodgkin's lymphoma, payers need to be aware of the distribution of clinical characteristics in these patients.