Development and Performance of a Population-Based Benchmark Model for Use of Cancer Surgery in High-Income Countries

Get Permission

In a study reported in The Lancet Oncology, Perera et al developed a population-based benchmark model for guideline-recommended use of surgery for primary cancers in high-income countries. Actual use of surgery was often consistent with model predictions but sometimes varied markedly.

Study Details

The study involved evaluation of the most recent clinical guidelines from high-income countries (Australia, United Kingdom, European Union, United States, and Canada) to map surgical treatment pathways for 19 malignant cancer sites among those that were notifiable in Australia in 2014, with the mapping intended to capture models of care in high-income regions. The models included only primary malignancies. In the benchmark models, the optimal use of surgery was considered as an indication for surgery in which surgery is the treatment of choice for a given clinical situation based on the guideline analysis. Population-based data on cancer stage, tumor characteristics, fitness for surgery, and other epidemiologic factors were derived from Australia and other high-income settings for 2017. The probabilities of surgery use in clinical pathways for each cancer were used to estimate population-level benchmark rates of cancer surgery, with the benchmark model estimates being compared with actual rates of cancer surgery in the South Western Sydney Local Health District from 2006 to 2012.

Key Findings

The overall surgery rate for all cancers predicted by the benchmark model for Australia in 2014 was 58%, meaning that surgery would be indicated in 58% of newly diagnosed patients at least once during the course of illness.

By cancer site, the benchmark-predicted rates varied from 22% for prostate cancer to 99% for testicular cancer.

In a comparison of benchmark-predicted rates with actual surgery rates in the South Western Sydney Local Health District from 2006 to 2012, the greatest positive and negative deviations from the model were an absolute increase of 29% for prostate cancer and an absolute 14% decrease for lung cancer.

For all 19 individual cancer sites analyzed, the benchmark-predicted vs actual rates of surgery (absolute difference) were:

  • 22% vs 51% (+29%) for prostate cancer
  • 49% vs 65% (+16%) for head and neck cancer
  • 49% vs 63% (+14%) for cervical cancer
  • 54% vs 63% (+9%) for stomach cancer
  • 86% vs 90% (+4%) for colon cancer
  • 78% vs 81% (+3%) for brain cancer
  • 85% vs 88% (+3%) for uterine cancer
  • 97% vs 99% (+2%) for breast cancer
  • 97% vs 99% (+2%) for melanoma
  • 81% vs 81% (0%) for kidney cancer
  • 89% vs 86% (-3%) for rectal cancer
  • 99% vs 96% (-3%) for testicular cancer
  • 28% vs 22% (-6%) for liver cancer
  • 91% vs 82% (-9%) for thyroid cancer
  • 88% vs 77% (-11%) for bladder cancer
  • 38% vs 27% (-11%) for esophageal cancer
  • 44% vs 33% (-11%) for pancreatic cancer
  • 83% vs 70% (-13%) for ovarian cancer
  • 33% vs 19% (-14%) for lung cancer.  

The investigators concluded, “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 vs observed surgery rates for some cancers, requiring more in-depth analysis of the observed differences.”

Sathira Kasun Perera, MSc, of the Collaboration for Cancer Outcomes Research and Evaluation, Ingham Institute of Applied Medical Research, University of New South Wales, Sydney, is the corresponding author for The Lancet Oncology article.

Disclosure: The study was funded by a University of New South Wales Scientia Scholarship and UK Research and Innovation-Global Challenges Research Fund. For full disclosures of the study authors, visit