Are prognostic scores better than clinician judgment?

A prospective study using three models

Background: Prognostic scores are commonly used, but how they compare to subjective clinician prediction of survival (CPS) isn’t well understood. CPS is often overly optimistic and estimated to have only 20-30% accuracy compared to actual survival. Prognostic scores have been reported to have 65-85% accuracy in various situations and populations, but prior studies were mostly in patients who survived for months. The authors sought to better understand the prognostic performance of the Palliative Performance Scale (PPS), Palliative Prognostic Index (PPI), and Palliative Prognostic Score (PaP) in patients who survive for weeks.
This study was chosen study because there are many scales available to us in palliative care, and utility of calculating, compared to subjective clinical assessments. 

Design and Participants: Multi-center prospective observational cohort study in Japan. Patients were adults with locally extensive or metastatic cancer admitted to one of 22 palliative care units (PCUs). PPS, PPI, PaP, and CPS were all determined at time of enrollment. Patients were classified into low-, intermediate-, and high-risk groups for each of the 3 calculated scores, and into groups of days, weeks, or months for CPS. Median overall survival was then calculated for each group by each prognostic score. Area under the receiver operating characteristics curve (AUROC) was calculated to assess discrimination ability of the prognostic scores.

Results: 1896 patients were enrolled during Jan-Dec 2017. 50.9% were men and 49.1% were women, and median age was 72.4 years. Median survival time was 19 days (with 95% CI 2-140.2 days). 87 clinicians participated, 70% of whom specialized in palliative care. The median length of clinical experience was 11.2 years in total, and 5.5 years in palliative care.

Based on AUROC, all tools had moderate-high discriminatory power, but the PaP and CPS were statistically significantly higher than the PPS and PPI at all time frames. CPS accuracy was highest in patients with weeks of survival.

Commentary: Many factors affect clinician prediction of survival, including patient population, physician characteristics and training/experience, and survival timeframe. Past literature is conflicted about performance of standardized prediction tools such as PPS, PPI, and PaP compared to CPS. In this population of palliative care patients who survived weeks, CPS and PaP performed slightly better than PPS and PPI, though all tools performed well. Notably, the greatest proportion of the PaP is CPS. One major limitation is that this was conducted in palliative care units in Japan, so may be difficult to generalize to other countries and practice settings.

Bottom Line: In palliative care patients with advanced cancer, clinician prediction of survival may be sufficiently accurate to predict short-term survival prognosis, though prognostic tools such as PPS, PPI, and PaP may improve prognostic accuracy particularly for inexperienced clinicians. 

Source: Hiratsuka Y, Suh SY, Hui D, et al. Are prognostic scores better than clinician judgment? A prospective study using three models. J Pain Symptom Manage. 2022;64(4):391-399. doi:10.1016/j.jpainsymman.2022.06.008

hiratsuka 2022.pdf