Introduction There is recognition that the nature of palliative care may be changing, as cancer survival improves and long-term conditions increase the complexity of the case load. Previous work has examined the accuracy of expert prognostication in patients with a life expectancy of less than a year. This piece of work also considers the accuracy of prognostication in patients thought to have a longer prognosis, and who may therefore be largely excluded from referral to specialist palliative care services, based on current referral patterns. It questions whether the palliative care net should be widened to reflect uncertainties in prognostication, and thereby extend valuable end of life services to a broader range of patients.

Aim(s) and method(s) A ‘snapshot’ cohort of the total number of patients accessing a palliative care day therapy unit on a given date was taken. Patients were categorised into the groups below using the ‘Surprise Question’ (Gold Standards Framework) by two experts in Palliative Care (one clinical nurse specialist/ one physician) and triple-read by a second palliative care consultant. End of life care=likely to have less than 12 month prognosis (malignant & non-malignant diagnoses). Cancer Survivorship= malignant diagnosis with a probable life expectancy of greater than 12 months. Non Malignant Survivorship=Patient has a non-malignant diagnosis with a probable life expectancy of greater than 12 months.

The patient group was re-analysed one year later, to assess the accuracy of prognostication.

Results See Abstract P 37 Table 1.

Conclusion(s) Our end of life predictions mirror other studies for those patients predicted to have a life expectancy of less than one year. We have also demonstrated that the cancer survivorship population (ie greater than 12 months life expectancy) had higher than predicted mortality rates within the first year. Is the surprise question an optimal tool? Does the surprise question promote ‘cherry picking’; should specialist palliative care be focusing on total symptom burden and complexity rather than prognostication?