Variation in chemotherapy prescribing rates and mortality in early breast cancer over two decades: a national data linkage study.

BACKGROUND

Regional variation in clinical practice may identify differences in care, reveal inequity in access, and explain inequality in outcomes. The study aim was to measure geographical variation in Scotland for adjuvant chemotherapy use and mortality in early-stage breast cancer.

PATIENTS AND METHODS

In this retrospective cohort study using population cancer registry-based data linkage, patients with surgically treated early breast cancer between 2001 and 2018 were identified from the Scottish Cancer Registry. Geographical regions considered were based on NHS Scotland organisational structure including 14 territorial Health Boards as well as three regional Cancer Networks. Regional variation in the proportion receiving chemotherapy, breast cancer mortality and all-cause mortality was investigated. Inter-regional comparisons of chemotherapy use were adjusted for differences in case mix using logistic regression. Comparison of breast cancer-specific mortality and all-cause mortality used regression with a parametric survival model. Time trends were assessed using moving average plots.

RESULTS

Chemotherapy use ranged from 35% to 46% of patients across Health Boards without adjustment. Variation reduced between 2001 and 2018. Following adjustment for clinical case mix, variation between cancer networks was within 3 percentage points, but up to 10 percentage points from the national average in some Health Boards. Differences in breast cancer mortality and all-cause mortality between cancer networks were modest, with hazard ratios of between 0.933 (95% confidence interval 0.893-0.975) and 1.041 (1.002-1.082) compared with the national average. Survival improved over the time period studied.

CONCLUSION

With adequate case mix adjustment, variation in adjuvant chemotherapy use for early breast cancer in Scotland is small, with a trend towards greater convergence in practice and improved mortality outcomes in more recent cohorts. This suggests very limited regional inequity in access and convergence of clinical practice towards risk-stratified treatment recommendations. Outliers require assessment to understand the reasons for variance.