Article Text

Download PDFPDF
Poster Numbers 185 – 241 – People & places: Poster No: 217
Predicting prognosis: experience of a tertiary referral cancer centre end-of-life care service
  1. Leanne Boyle,
  2. Clare Smith,
  3. Anna-Marie Stevens and
  4. Julia Riley
  1. The Royal Marsden NHS Foundation Trust, London, UK


Background The UK End-of-Life Care Strategy advocates providing choice for terminally ill patients. People want to die at home however 58% of people die in hospital. The Royal Marsden Hospital introduced the Hospital2Home (H2H) Service to bridge the gap for patients who are no longer receiving anticancer treatment by conducting a community based case conference. The aim to facilitate patient choice by encouraging participation in advanced planning. One barrier to planning at the end-of-life is the difficulty in predicting prognosis.

Aim This audit examines the accuracy of the clinically predicted prognosis in patients referred to the H2H service.

Method This was a retrospective audit analysing data collected in the routine clinical service. Patients referred to the service are assigned a predicted GSF prognostic indicator (A=Years, B=Months, C=Weeks or D=Days). This was compared to survival.

Results 308 patients have had a H2H case conference, 270 have now died. The median overall survival is 29 days (min 1, max 653). The accuracy of predicting prognosis per group is: • Group D (Days): 8 patients were given a prognosis of days. 6/8 (75%) died within days. • Group C (Weeks): 132 patients were given a prognosis of weeks. 60/132 (45.5%) died in weeks, 42/132 (31.8%) in months and 30/132 (22.7%) in days. • Group B (Months):122 patients were given a prognosis of months. 79/122 (64.8%) died in months, 11/122 (9.0%) in days, 30/122 (24.6%) in weeks and 2/122 (1.6%) survived over 1 year. • 8 patient's data missing. Overall clinicians were accurate in 53.7% of cases, over optimistic in 26.3% and under estimate prognosis in 17%.

Conclusion Prognostication is difficult. It is understandable that professionals have concerns about predicting prognosis. If anything we overestimate. Therefore if patients are to be told prognosis then the uncertainties around this prediction should be clearly explained.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.