Clinical prediction survival of advanced cancer patients by palliative care: a multi-site study

Aims: This study examined (1) accuracy of clinician prediction of survival (CPS) by palliative practitioners on first assessment with the use of standardised palliative tools, (2) factors affecting accuracy, (3) potential impact on clinical care. Methods: A multi-site prospective study (n=1530) was used. CPS was divided into four time periods ( 2 to 6wks, >6 to 12wks and >12wks). Multivariate analysis was assessed on six predictor variables. Results: Overall, median survival of the sample was only 5 weeks. CPS category was accurate only 38.6% of the time, with 44.6% patients dying before the predicted time period. Of six candidate variables, on multivariate analysis only (i) the clinical time periods themselves and (ii) Palliative Performance Scale <=50 predicted for prognostic accuracy. Conclusion: CPS, even by palliative practitioners, remains overly optimistic with the existence of the horizon effect. This raises the question in that these individuals may have been potentially overtreated.