Objective To evaluate the accuracy and impact of clinicians’ estimates of prognosis (CEP) in patients referred for hospice inpatient care.

Methods Retrospective review of 12 months’ referrals to a London hospice unit. Data extracted included date of referral, admission and death and CEP.

Results N=383. Mean age 72 years (range 24–101). CEP accuracy: Median survival where CEP was ‘days’ (n=141) was 7 days (0–164); CEP ‘weeks’ (n=167) was 14 days (1–538); CEP ‘months’ (n=75) was 32 days (2–507). Kaplan-Meier survival curves showed significant difference between CEP of ‘months’ and ‘weeks’ (p<0.0001); ‘months’ and ‘days’ (p<0.0001); but not ‘days’ and ‘weeks’ (p=0.1). CEP impact: admission waiting time increased with increasing CEP: CEP ‘days’ (n=105) median 1 day (0–14); CEP ‘weeks’ (n=154) median 2 days (0–46); CEP ‘months’ (n=69) median 3 days (0–46). No significant difference was demonstrated in the number of discharge planning conversations between groups (0.9/patient).

Conclusions CEP was accurate in over half of the cases but did not adequately discriminate between those with prognoses of days or weeks. CEP may affect the prioritisation given to patients by hospices. Inaccurate CEP on referral forms may influence other aspects of care; however, further research is needed.