Purpose: Evidence on health disparities among patients treated with multimodality therapy protocols is still limited. We aimed to evaluate the associations between patient-level and system-level factors and the receipt of guideline-concordant therapy among patients with esophageal adenocarcinoma (EA).
Methods: This is a national cohort study of patients with stage I-III EA in the National Cancer Database (2006-2018) treated with either upfront resection or neoadjuvant therapy followed by surgery. Clinical and pathologic staging data were used for defining guideline-concordant therapy. Logistic regression models were built to identify independent associations between deviations from treatment guidelines and clinical, sociodemographic, and hospital-related factors.
Results: Among 18,803 patients with EA treated at 1,049 hospitals, 4,511 had an endoscopic resection, 4,866 had upfront resection, and 9,426 had neoadjuvant therapy followed by surgery. A total of 16,002 patients (85.1%) received guideline-concordant therapy. Patients who were age 70 years or older (odds ratio [OR] 1.35; 95% CI, 1.14 to 1.60), had a Charlson-Deyo modified score of ≥ 1, and were treated at hospitals with a safety-net burden of ≥ 10% (OR 1.13; 95% CI, 1.02 to 1.25) had higher risk of deviations from guidelines, whereas treatment at high-volume facilities (OR 0.84; 95% CI, 0.74 to 0.95) and diagnosis after 2011 decreased this risk. Relative to cT0-1N0 patients, those with cT2N0 disease had the highest risk of deviations from guideline-concordant therapy (OR 3.76; 95% CI, 3.30 to 4.29). Among patients treated at high-volume facilities, safety-net burden over 10% increased the risk of deviations (OR 1.26; 95% CI, 1.01 to 1.57).
Conclusion: The delivery of guideline-concordant therapy among patients with EA varies significantly across clinical stage groups and is associated with several sociodemographic disparities. This knowledge should be a resource for future quality improvement strategies intended to address inequitable care within vulnerable populations.