Geographical variations in the use of cancer treatments are associated with survival of lung cancer patients

Abstract

Introduction: Lung cancer outcomes in England are inferior to comparable countries. Patient or disease characteristics, healthcare-seeking behaviour, diagnostic pathways, and oncology service provision may contribute. We aimed to quantify associations between geographic variations in treatment and survival of patients in England.

Methods: We retrieved detailed cancer registration data to analyse the variation in survival of 176,225 lung cancer patients, diagnosed 2010-2014. We used Kaplan-Meier analysis and Cox proportional hazards regression to investigate survival in the two-year period following diagnosis.

Results: Survival improved over the period studied. The use of active treatment varied between geographical areas, with inter-quintile ranges of 9%-17% for surgical resection, 4%-13% for radical radiotherapy, and 22%-35% for chemotherapy. At 2 years, there were 188 potentially avoidable deaths annually for surgical resection, and 373 for radical radiotherapy, if all treated proportions were the same as in the highest quintiles. At the 6 month time-point, 318 deaths per year could be postponed if chemotherapy use for all patients was as in the highest quintile. The results were robust to statistical adjustments for age, sex, socio-economic status, performance status and co-morbidity.

Conclusion: The extent of use of different treatment modalities varies between geographical areas in England. These variations are not attributable to measurable patient and tumour characteristics, and more likely reflect differences in clinical management between local multi-disciplinary teams. The data suggest improvement over time, but there is potential for further survival gains if the use of active treatments in all areas could be increased towards the highest current regional rates.

Keywords: lung cancer; lung cancer chemotherapy; radiotherapy; surgery; survival.

Figure 1

Inclusions and exclusions in lung cancer datasets 2005–2014, 2010–2014 and 2010–2013.

Figure 2

One-year (blue), 2 year (red) and 5 year (green) survival of lung cancer patients in England, diagnosed 2005–2014.

Figure 3

Kaplan-Meier survival functions for lung cancer patients, categorised by the quintiles of treatment rates in their area of residence. A: Survival of Stage I-II patients in relation to quintile of surgical resection rate. B: Survival of Stage III patients and patients with no record of stage, in relation to quintile of radical radiotherapy rate. C: Survival of Stage IV patients in relation to quintile of chemotherapy rate.

Figure 4

Plots of the HR for death in 15 equal-sized groups of patients from geographical areas, from the lowest to the highest geographical treatment rates. A: Survival of Stage I-II patients in relation to surgical resection rate. B: Survival of Stage III patients and patients with no record of stage, in relation to radical radiotherapy rate. C: Survival of Stage IV patients in relation to chemotherapy rate. Quadratic functions were fit to the 15 data points and are presented with 95%CIs.