Understanding Hospital-Level Patterns of Nonoperative Management for Low-risk Thyroid and Kidney Cancer

IMPORTANCE There is a growing trend toward conservative management for certain low-risk cancers. Hospital and health-system factors may play a role in determining how these patients are managed. OBJECTIVE To explore the contribution of hospitals on patients' odds of nonoperative management for low-risk cancer. DESIGN, SETTING, AND PARTICIPANTS In this cross-sectional study, individuals with low-risk papillary thyroid cancer and solitary kidney masses were identified, and those receiving nonoperative management vs surgery were compared. Patients with low-risk thyroid cancer and kidney cancer from 2015 to 2017 eligible for nonoperative management according to National Comprehensive Cancer Network guidelines within the National Cancer Database were included. Data were analyzed from October 2021 to March 2022. MAIN OUTCOMES AND MEASURES For each facility, the proportion of these patients who received operative and nonoperative management was calculated. A mixed-effects logistic regression model with a hospital-level random effects term was used to calculate factors associated with nonoperative management. Between-hospital variability was assessed using ranked caterpillar plots. RESULTS There were 19 570 individuals with low-risk thyroid cancer (15 344 women [78.4%]; mean [SD] age, 51.74 [95% CI, 51.39-52.08] years) and 41 403 with kidney cancer (25 253 men [61.0%]; mean [SD] age, 61.93 [95% CI, 61.70-62.17] years). In the group with low-risk thyroid cancer, 2.1% (419 patients) received nonoperative management, and in the group with kidney cancer, 9.5% (3928 patients) received nonoperative management. This varied between hospitals from 1.1% (95% CI, 1.0%1.1%) in the bottom decile to 103% (95% CI, 8.0%-12.4%) in the top decile for low-risk thyroid cancer, and from 4.3% (95% CI, 4.1%-4.4%) in the bottom decile to 24.6% (95% CI, 22.7%-26.5%) in the top decile for small kidney masses. For both cancers, age was associated with increased odds of nonoperative treatment. The hospital-level odds of nonoperative management of thyroid and kidney cancer using unadjusted probabilities (observed proportions) were minimally correlated (Spearman rho = .33; P < .001). CONCLUSIONS AND RELEVANCE The findings of this study suggest that although health systems factors may be associated with the tendency to pursue nonoperative management, hospital-level factors may differ when comparing unrelated cancers.