Evidence-based benchmarks for use of cancer surgery in high-income countries: a population-based analysis.

BACKGROUND

Estimating a population-level benchmark rate for use of surgery in the management of cancer helps to identify treatment gaps, estimate the survival impact of such gaps, and benchmark the workforce and other resources, including budgets, required to meet service needs. A population-based benchmark for use of surgery in high-income settings to inform policy makers and service provision has not been developed but was recommended by the Lancet Oncology Commission on Global Cancer Surgery. We aimed to develop and validate a cancer surgery benchmarking model.

METHODS

We examined the latest clinical guidelines from high-income countries (Australia, the UK, the EU, the USA, and Canada) and mapped surgical treatment pathways for 30 malignant cancer sites (19 individual sites and 11 grouped as other cancers) that were notifiable in Australia in 2014, broadly reflecting contemporary high-income models of care. The optimal use of surgery was considered as an indication for surgery where surgery is the treatment of choice for a given clinical scenario. Population-based epidemiological data, such as cancer stage, tumour characteristics, and fitness for surgery, were derived from Australia and other similar high-income settings for 2017. The probabilities across the clinical pathways of each cancer were multiplied and added together to estimate the population-level benchmark rates of cancer surgery, and further validated with the comparisons of observed rates of cancer surgery in the South Western Sydney Local Health District in 2006-12. Univariable and multivariable sensitivity analyses were done to explore uncertainty around model inputs, with mean (95% CI) benchmark surgery rates estimated on the basis of 10 000 Monte Carlo simulations.

FINDINGS

Surgical treatment was indicated in 58% (95% CI 57-59) of newly diagnosed patients with cancer in Australia in 2014 at least once during the course of their treatment, but varied by site from 23% (17-27) for prostate cancer to 99% (96-99) for testicular cancer. Observed cancer surgery rates in South Western Sydney were comparable to the benchmarks for most cancers, but were higher for some cancers, such as prostate (absolute increase of 29%) and lower for others, such as lung (-14%).

INTERPRETATION

The model provides a new template for high-income and emerging economies to rationally plan and assess their cancer surgery provision. There are differences in modelled versus observed surgery rates for some cancers, requiring more in-depth analysis of the observed differences.

FUNDING

University of New South Wales Scientia Scholarship, UK Research and Innovation-Global Challenges Research Fund.