Facility contextual effects influence the use of adjuvant immunotherapy in stage III melanoma.

Abstract

e18758 Background: There are now numerous effective adjuvant immunotherapy options for surgically resected stage III melanoma including novel checkpoint inhibitors and targeted therapies. Current guidelines recommend that the decision to treat stage III melanoma with adjuvant immunotherapy should be individualized and based upon disease burden, patient goals and anticipated therapy tolerance. We sought to assess the contribution of patient, tumor and facility factors on the implementation of immunotherapy in patients with surgically resected stage III melanoma. Methods: Using the National Cancer Database (NCDB), patients from 2012-2017 that underwent excision and were found to have a positive sentinel node were identified. A multivariable mixed effects logistic regression model with a random intercept for site was used to determine the effect of patient, tumor, and facility variables on the probability of immunotherapy. Reference Effect Measures (REM) were used to estimate the variation in immunotherapy use due to unmeasured facility factors (contextual effects) after adjusting for measured patient, tumor, and facility variables. Results: From 2012 to 2017, the percent of patients with stage III melanoma treated with adjuvant immunotherapy increased from 23.7% to 38.5% (p < 0.05). Overall, younger patients and patients with private insurance were more likely to receive immunotherapy. Tumor factors associated with increased use of adjuvant immunotherapy included increasing depth, mitotic rate ³1, ulceration, lymphovascular invasion (LVI), and undergoing a completion lymph node dissection (CLND). Additionally, treatment at a facility with a surgical volume <190 cases/year was associated with increased immunotherapy use. However, the width of the 90% REM range for unmeasured facility effects exceeded that of the measured facility, tumor, time, and patient demographics suggesting that contextual effects had a higher contribution to the variation in immunotherapy use. Conclusions: Our analysis suggests that uninsured patients and patients with government insurance (Medicaid and Medicare) are not receiving immunotherapy at the same frequency as privately insured patients with the same tumor characteristics treated at the same facility. Lastly, compared to known patient, tumor and facility factors, institutional contextual effects were the major drivers of the implementation of immunotherapy.[Table: see text]