Routine data can estimate clinical outcomes in Head and Neck cancer: Results of a pilot study


Session type:

Matt Williams1,4, Zi Wei Liu2, Heather Fitzke3
1Dept. Clinical Oncology, University College Hospital, London, UK, 2Dept. Otolaryngology Head and Neck surgery, Colchester General Hospital, Colchester, UK, 3Dept. Imaging, University College Hospital, London, UK, 4Dept. Computer Science, University College London, London, UK


For patient with head & neck cancer, control of loco-regional disease is key. Recurrent disease may be curable, but at considerable cost. At present, national audits of H&N cancer can report overall survival, but not relapse rates1. At a national level, there are routinely collected data on hospital admissions, radiotherapy and chemotherapy (via HES, RTDS and SACT). We conducted a pilot study to evaluate the use of routinely collected healthcare data to describe patterns of treatment in H&N cancer patients.


We identified 7 patients at high risk of recurrence from local MDT lists and extracted a record of their treatment from hospital notes. We obtained a combination of local HES, SACT and RTDS data on the same patients, and assessed our approach by measuring comparative identification of key healthcare events and time-to-event data.


All 7 patients had primary squamous cell carcinoma of the oropharynx or oral cavity, and all received radical treatment. In 6 cases, this was with chemo-radiotherapy; 1 had primary radical surgery. 4 patients had further radical/adjuvant treatment for residual disease or close surgical margins. 1 had no further treatment and 2 had further palliative treatment. The four patients who had further radical/ adjuvant treatment, and the 1 who had no further treatment, were all correctly identified using routine data. Only 1 of the 2 patients who had subsequent palliative treatment were correctly identified. The estimates provided by the routine data for event-free survival were within 0.85 - 1.15 of the actual figure for 3 of the 5 patients.


We have developed a method of assessing clinically relevant outcomes for patients with H&N cancers using routinely collected data. This small pilot study demonstrates that in radically treated patients it accurately captures patterns of initial and subsequent care.