The clinical applicability of inpatient readmissions for cancer patients.

e18661 Background: The 30-day rate of readmission, a standard metric for many common acute medical conditions, has also been proposed as a quality measure in cancer. However, hospitalizations for cancer patients are more heterogeneous than traditional groups (pneumonia, heart failure patients). We analyzed the pattern and timing of inpatient readmissions for a population-based sample of adults newly diagnosed with cancer. Methods: We identifiably linked adults diagnosed with cancer (2008-2018) from the UT Southwestern tumor registry to comprehensive hospital visits from a regional all-payer health exchange covering a 100-mile radius of Dallas, TX. Our primary outcome was first inpatient readmission within 30 days of discharge among patients hospitalized within 1 year of cancer diagnosis. Subsequent index periods began > 30 days from index discharge to create non-overlapping readmission windows. We grouped primary diagnosis codes into 15 clinically relevant conditions, noting concordance between index and readmission diagnosis. We separately grouped hospital diagnoses deemed preventable by Medicare’s systemic therapy (chemo or immunotherapy) quality measure. We defined early readmission as occurring on days 0-7 and late as days 8-30. Results: Our cohort of 19,608 patients (49.5% female, 71.0% commercial/Medicare insurance) were most diagnosed with gastrointestinal (14.7%), breast (13.9%), and lung cancer (13.6%), with 28.9% advanced stage at diagnosis. Of those, 53.3% had received systemic therapy for their initial treatment. Patients generated 15,066 index admissions and 1,783 readmissions (11.8% 30-day readmission rate). The reason for readmission was the same as index hospitalization in 22.9% of early readmissions and 15.3% of late readmissions. Neoplasm and infection were the most common conditions; their frequencies were similar in early readmissions (12.7% and 12.8% respectively), but infections were higher in late admissions (12.0% and 15.7%). The next most common were gastrointestinal symptoms - modestly more frequent in systemic therapy patients vs. non-systemic therapy (11.4% and 9.3%); and complications of medical care, modestly more frequent in early readmissions vs. late (9.8% and 7.5%). The proportion of early readmissions for preventable diagnoses was 21.7% following receipt of systemic therapy vs 18.0% not receiving systemic therapy, and 29.2% vs. 22.5% respectively for late readmissions. Conclusions: The 30-day readmission rate in patients with cancer was 11.8% (lower than many inpatient conditions) and was usually due to reasons different than the index admission. More work is needed to validate/refine the concept of readmissions to cancer patients as an indicator of quality of care due heterogeneity of patients with cancer, the reasons they are hospitalized and rehospitalized, and relatively low rates of potential preventability.