Factors associated with colon cancer fragmentation in the Keystone state.

120 Background: Fragmentation of cancer care delivery across multiple providers can impact care quality and patient outcomes. Management of non-metastatic colon cancer often involves adjuvant chemotherapy, which increases the risk of potentially deleterious care dispersion. This study assesses the extent of care fragmentation across different inpatient facilities and examines patient characteristics associated with inpatient care fragmentation. Methods: We adapt the modified Herfindahl-Hirschman market concentration index to construct a Fragmentation Index (FI) capturing the distribution of a patient’s care across inpatient facilities. For patients who underwent oncologic surgery for colon cancer from 2010-2016, we calculate FI for the 15 months post-diagnosis if at least 2 admissions occur during the study period. We use Pennsylvania Cancer Registry data linked to all-payer statewide facility discharge records to estimate the relationship between FI and patient demographics, rurality, area deprivation index (ADI), and payer status using adjusted multivariable linear regression models. Results: For the operative cohort (28.6% Stage I, 34.7% Stage II, 36.7% Stage III, n = 7803), 86% of patients lived in urban areas, 9.5% in large towns and 5% in rural areas. 60% had Medicare, 6.1% Medicaid and 7.1% were uninsured or had unknown payer status. 52% of patients were female and 87% were non-Hispanic White. The overall mean FI was 0.192; 60% (n = 4693) had an FI of 0, indicating all admissions were at the same hospital, and among those with FI> 0 (n = 3110), mean FI was 0.483. The mean FI for patients potentially eligible for adjuvant chemotherapy (Stage II, III) was 0.188: 61% of patients had an FI of 0 and among those with FI > 0 (n = 2173), mean FI was 0.482. Stratification by stage reveals similar proportions of FI > 0.5 (stage I: 24.6%, stage II: 22.3%, stage III: 21.3%). Increased care dispersion was observed in patients from large towns (ΔFI 0.04, CI 0.03—0.06) or rural areas (ΔFI 0.04, CI 0.02—0.08) relative to urban areas, as well as for patients with relatively higher number of hospitalizations (ΔFI 0.04, CI 0.03—0.04). Less care dispersion was seen in patients with stage III disease (relative to stage I ΔFI -0.02, CI -0.03 — -0.01), uninsured status (relative to private insurance ΔFI -0.03, CI -0.05 – -0.01) or residence in the most disadvantaged ADI quartile (relative to Q1 ADI ΔFI -0.02, CI -0.04 — 0.00). Conclusions: Among patients with operatively managed colon cancer, slightly more than half did not experience inpatient care fragmentation. As expected, a higher number of total admissions and non-urban residence were associated with increased inpatient care dispersion. However, our findings of more concentrated care among patients with stage III disease, higher levels of relative economic disadvantage, and uninsured status suggest potentially complex and divergent forces driving hospital selection, to be explored in future work.