Session Item

Hypofractionated palliative radiotherapy (PallRT) aims to improve symptom control and, in limited circumstances, survival whilst minimising treatment burden. Anecdotally, wide variation in use persists. The national radiotherapy dataset (RTDS), collected by PHE, provides a unique opportunity to assess variation in use and early mortality (a marker of futility) following palliative radiotherapy.

All radiotherapy episodes delivered 2 years in the English NHS were extracted from the RTDS, linked to cancer registration and admissions data. Treatment intent was defined using clinically determined algorithms. Site-treated using PallRT was defined using anatomical codes. Variation in fractionation patterns of PallRT was assessed by provider organisation. 30 day mortality (30DM) was determined and predictors of 30DM assessed using multi-variable logistic regression models. Variation in 30DM between providers was presented using unadjusted and adjusted funnel plots.

Over 100,000 palliative radiotherapy treatments were delivered across the English NHS in 2014-15. Treatment to bone lesions accounted for the largest proportion (41.7% of delivered prescriptions), followed by treatments for soft tissue (39.5%), brain and base of skull metastases (10.5%) and the head and neck (2.9%). The fractionation patterns delivered varied widely between provider organisations. For example, 65% of non-emergency bone and spinal treatments were delivered using a single fraction (range 37.7-90.3%), see figure 1. Over 8000 patients died within 30 days of a first treatment episode in the cohort (10.4%). Multiple factors were significantly associated with 30DM. These included fractionation pattern (10 fractions vs single fraction OR 0.23 (95%CI 0.21-0.26)), travel time (>60 minutes vs <20 minutes OR 0.85(95%CI 0.76-0.96)), urgency of treatment (emergency vs routine OR 2.02 (95%CI 1.87-2.18)) and inpatient status at treatment (inpatient vs outpatient OR 2.44 (95% CI 2.16-2.77)). The variation in 30DM was presented using unadjusted and adjusted funnel plots (see figure 2).

Wide variation in the use of fractionation in PallRT was demonstrated alongside variation in 30DM. The benefit delivered to those who die very soon after treatment is likely to be very limited. Better understanding of the benefits gained by these very poor prognosis patients, and validated prognostic models, are needed to help inform decision-making. This may help to avoid fractionation in this frail population and, where appropriate, guide use of alternative strategies, such as holistic palliative care.

The data quality and ascertainment in the RTDS are improving rapidly and there is now a pressing need for these data to be presented to treating clinicians to allow individuals to better understand how practise within their centre compares to peers. This will support ongoing professional development and improvement of services where appropriate.