Accuracy of clinician perceptions in estimating patient survival in general medicine

Patients and families often rely on clinicians' predictions of life expectancy to make critical treatment and personal decisions.

To evaluate the accuracy and risk intelligence of clinicians in predicting 1-year survival of general medicine patients after hospital discharge.

Doctors, nurses, allied health professionals and medical students completed 125 questionnaires (70% response rate) distributed in the intensive care unit (ICU) and at a medical grand round at Christchurch Hospital. Survival estimates for four case vignettes were compared to an evidence-based survival score (EBSS) derived from over 120 000 local hospital discharges. Accuracy was defined as estimates within ±10% and ±20% of the EBSS. Confidence was assessed based on the reported range for each estimate.

Median survival estimates were widely variable and frequently underestimated the EBSS, with 27% and 48% of estimates falling within ±10% and ±20% of the EBSS respectively. Confidence was moderate (median range ±15%), although 52% of estimates were outside ±20% of the EBSS. No significant differences in accuracy were observed between ICU and general medicine respondents, doctors and other healthcare professionals, or senior and junior doctors.

Clinician estimates of 1-year survival in vignette-based scenarios varied widely and differed from model-based estimates. These findings highlight variability in prognostic estimation and support the potential role of evidence-based tools to complement clinical judgement.