Determinants of Variation in the Use of Adjuvant Chemotherapy for Stage III Colon Cancer in England

Adjuvant chemotherapy (ACT) for stage III colon cancer is well-established. This study aimed to explore the determinants of ACT use and between-hospital variation within the English National Health Service (NHS).

In total, 11 932 patients (diagnosed 2014-2017) with pathological stage III colon cancer in the English NHS were identified from the National Bowel Cancer Audit. Records were linked to Systemic Anti-Cancer Therapy and Hospital Episode Statistics databases. Multi-level logistic regression analyses were carried out to estimate independent factors for ACT use, including age, sex, deprivation, comorbidities, performance status, American Society of Anaesthesiologists (ASA) grade, surgical urgency, surgical access, TNM staging, readmission and hospital-level factors (university teaching hospital, on-site chemotherapy and high-volume centre). A random intercept was modelled for each English NHS hospital (n = 142). Between-hospital variation was explored using funnel plot methodology. Fully adjusted random-intercept models were fitted separately in young (<70 years) and elderly (≥70 years) patients and intra-class correlation coefficients estimated.

60.7% of patients received ACT. Age was the strongest determinant. Compared with patients aged <60 years, those aged 60-64 (adjusted odds ratio [aOR] 0.76, 95% confidence interval 0.63-0.93), 65-69 (aOR 0.63, 95% confidence interval 0.54-0.74), 70-74 (aOR 0.53, 95% confidence interval 0.44-0.62), 75-79 (aOR 0.23, 95% confidence interval 0.19-0.27) and ≥80 years (aOR 0.05, 95% confidence interval 0.04-0.06) were significantly less likely to receive ACT. With adjustment for other factors, ACT use was more likely in patients with higher socioeconomic status, fewer comorbidities, better performance status, lower ASA grade, advanced disease, elective resections, laparoscopic procedures and no unplanned readmissions. Hospital-level factors were non-significant. The observed proportions of ACT administration in the young and elderly were 46-100% (80% of hospitals 74-90%) and 10-81% (80% of hospitals 33-65%), respectively. Risk adjustment did not reduce between-hospital variation. Despite adjustment, age accounted for 9.9% (7.2-13.4%) of between-hospital variation in the elderly compared with 2.7% (1.2-5.7%) in the young.

There is significant between-hospital variation in ACT use for stage III colon cancer, especially for older patients. Advanced age alone seems to be a greater barrier to ACT use in some hospitals.