A stratified analysis of the perioperative outcome of 17623 patients undergoing major head and neck cancer surgery in England over 10 years: Towards an Informatics-based Outcomes Surveillance Framework.

Objectives: To perform a national analysis of the perioperative outcome of major head and neck cancer surgery to develop a stratification strategy and outcomes assessment framework using hospital administrative data.

Design: A Hospital Episode Statistics N = near-all analysis.

Settings: The English National Health Service.

Main Outcome Measures: Local audit data were used to assess and triangulate the quality of the administrative dataset. Within the national dataset, cancer sites, morbidities, social deprivation, treatment, complications, and in-hospital mortality were recorded.

Results: Within local audit datasets, the accuracy of assigning newly-derived Cancer Site Strata and Resection Strata were 92.3% and 94.2%, respectively. Accuracy of morbidities assignment was 97%. Within the national dataset, we identified 17 623 major head and neck cancer resections between 2002 and 2012. There were 12 413 males and mean age at surgery was 63 ± 12 years. The commonest cancer site strata were oral cavity (42%) and larynx-hypopharynx (32%). The commonest resection site was the larynx (n = 4217), and 13 211 and 11 841 patients had neck dissection and flap-based reconstruction, respectively. There were prognostically significant baseline differences between patients with oromandibular and pharyngolaryngeal malignancy. Patients with pharyngolaryngeal malignancies had a greater burden of morbidities, lower socio-economic status, fewer primary resections, and a sixfold increased risk of undergoing their major resection during an emergency hospital admission. Mean length of stay was 25 days and each complication linearly increased it by 9.6 days. There were 609 (3.5%) in-hospital deaths and a basket of seven medical and three surgical complications significantly increased the risk of in-hospital death. At least one potentially lethal complication occurred in 26% of patients. The risk of in-hospital death in a patient with no potentially lethal complication was 1.1% and this increased to 6% with one potentially lethal complication, and to 15.1% if two potentially lethal complications occurred in one patient. Complex oral-pharyngeal resections and pharyngolaryngectomies had the highest risks of complications and mortality.

Conclusion: Mortality following head and neck cancer surgery shows variation across different resection strata. We propose an Informatics-based Framework for Outcomes Surveillance (IFOS) in Head and Neck Surgery for perpetual quality assurance, using the local hospital coding data or its collated destination, the national administrative dataset.