Prognostic Significance, Accuracy and Usefulness of Oncologists Estimates of Survival Time for Patients Starting First Line Chemotherapy for Advanced Non-Small Cell Lung Cancer (ANSCLC) (#49)
Aim: To determine the accuracy and prognostic significance of oncologists’ estimates of survival time for patients with ANSCLC.
Methods: Medical oncologists recorded the “expected survival time in months” for individual patients with ANSCLC prior to randomisation in a trial of first-line platinum-based chemotherapy. Estimates within 0.75-1.33 times observed survival were deemed precise. We expected 50% of patients to live longer (or shorter) than their oncologist’s estimate (calibration), 50% to live from half to double their estimate (typical scenario); 5-10% to live ≤¼ of their estimate (worst-case scenario); and, 5-10% to live ≥3 times their estimate (best-case scenario). Associations between estimated and observed survival times were assessed with Cox proportional hazards regression.
Results: Estimates of survival were available for 244 (98%) of the first 250 patients randomised. After a median follow-up of 21 months there were 172 deaths (69%). The median (interquartile range) for observed survival was 10 months (5-20) and for estimated survival was 11 months (9-12). Oncologists’ estimates were imprecise (22% from 0.75-1.33 times observed) but well calibrated (53% lived longer than expected). The proportion of patients with an observed survival time: ≤1/4 of their estimated survival time was 10%; half to double their estimated survival time was 53%; and ≥3 times their estimated survival time was 13%. The oncologist’s estimate of survival time at baseline was the strongest predictor of observed survival in both univariate analysis (HR 0.90, 95% CI 0.86-0.95, p<0.001) and multivariate analysis (HR 0.90, 95% CI 0.86-0.95, p<0.001) accounting for all other independent significant predictors, namely: estimated neutrophil-lymphocyte ratio >5 (HR 3.15, 95% CI 1.76-5.64, p<0.001); haemoglobin <120g/L (HR 1.93, 95% CI 1.3-2.9, p=0.001) and total white cell count >11x109/L (HR 1.55, 95% CI 1.05-2.27, p=0.03).
Conclusions: Oncologists’ estimates of survival time were well-calibrated, independently associated with observed survival and a reasonable basis for estimating worst-case, typical and best-case scenarios for survival.