Background/aim: NEAT is a validated prognostic model that calculates survival estimates based on the number of active tumors, ECOG performance status, albumin, and primary tumor site. Since models are imperfect, we hypothesized that experienced clinicians could predict the survival of patients with metastatic cancer better than a validated prognostic model alone, thereby quantifying the previously unmeasured value of clinical judgment.
Patients and methods: This prospective, single-institution cohort study conducted at a large community hospital recruited 73 patients with metastatic cancer referred to radiation oncology between October 2016 and December 2017. The consulting nurse and physician were prospectively surveyed on whether the patient would survive a longer or shorter duration than the calculated NEAT survival estimates. The accuracy of predictions between groups was assessed using the McNemar's chi-squared test.
Results: The median survival for enrolled patients was 9.2 months. Nursing and physician predictions were similarly accurate (61.6% vs. 60.3%, p=0.85). The accuracy of confident clinical predictions was similar to less confident predictions (64.2% vs. 58.2%, p=0.46). Radiation dose intensity was informed by predicted survival, and median survival was significantly higher in patients receiving an EQD2≥40 (17 months vs. 2 months, p<0.001).
Conclusion: Experienced clinicians, both nurses and oncologists, have insight that modestly supplements the accuracy of a validated model to predict survival in patients with advanced cancer.
Keywords: Metastasis; artificial intelligence; palliative care; prognosis; radiation oncologists.
Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.