Unmeasured Factors Are Associated with the Use of Completion Lymph Node Dissection (CLND) in Melanoma

BackgroundCompletion lymph node dissection (CLND) was the standard treatment for patients with melanoma with positive sentinel lymph nodes (SLN) until 2017 when data from the DeCOG-SLT and MLST-2 randomized trials challenged the survival benefit of this procedure. We assessed the contribution of patient, tumor and facility factors on the use of CLND in patients with surgically resected Stage III melanoma. MethodsUsing the National Cancer Database, patients who underwent surgical excision and were found to have a positive SLN from 2012 to 2017 were included. A multivariable mixed-effects logistic regression model with a random intercept for the facility was used to determine the effect of patient, tumor, and facility variables on the risk of CLND. Reference effect measures (REMs) were used to compare the contribution of contextual effects (unknown facility variables) versus measured variables on the variation in CLND use. ResultsFrom 2012 to 2017, the overall use of CLND decreased from 59.9% to 26.5% (p < 0.0001). Overall, older patients and patients with government-based insurance were less likely to undergo CLND. Tumor factors associated with a decreased rate of CLND included primary tumor location on the lower limb, decreasing depth, and mitotic rate <1. However, the contribution of contextual effects to the variation in CLND use exceeded that of the measured facility, tumor, time, and patient variables. ConclusionsThere was a decrease in CLND use during the study period. However, there is still high variability in CLND use, mainly driven by unmeasured contextual effects.