External validation of the PROgnostic Model for Advanced Cancer for 90-day, 6-month and 1-year mortality in geriatric oncology

Introduction Prognostication is an important aspect of cancer care, particularly for older adults who often prioritise quality of life over longevity. Currently, there are few useful prognostic tools for predicting 6-month and 1-year mortality to guide hospice enrolment and interventions in geriatric oncology. Most existing models focus on short-term prognosis (days to weeks) or were developed for patients receiving active treatment, limiting their applicability for hospice eligibility assessment. This study aimed to validate the predictive performance of the PROgnostic Model for Advanced Cancer (PRO-MAC) for 90-day, 6-month and 1-year mortality in older outpatients with cancer. Research design and methods A total of 901 older adults (≥65 years) with incurable cancer, seen at a palliative care clinic in a university hospital from January 2015 to June 2020, were studied in this retrospective study. Patient demographics, survival data and PRO-MAC scores were collected. PRO-MAC incorporates five prognostic factors: Palliative Performance Scale, Edmonton Symptom Assessment System-revised, total white cell count, serum albumin and number of organ systems with metastatic disease. Missing data (30.1% of participants) were handled using multiple imputations with chain equations. Model accuracy was assessed using Harrell’s C-index, Cox calibration slope, Kaplan-Meier curve and log-rank test. Results The median age was 79 years, with 90-day, 6-month and 1-year mortality rates of 35.7%, 57.9% and 77.6%, respectively. The PRO-MAC model achieved a C-index of 0.69 (95% CI 0.66 to 0.72), 0.65 (0.63 to 0.68) and 0.63 (0.61 to 0.65) for 90-day, 6-month and 1-year mortality, respectively. Patients in the high-risk group (≥2 points) had a 3.0-, 2.4- and 2.0-times higher risk of mortality compared with the low-risk group with sensitivity of 65%, 56%, 49%; specificity of 68%, 74%, 74%; positive predictive value of 53%, 75%, 87%; and negative predictive value of 78%, 55% and 30% for 90-day, 6-month and 1-year, respectively. Conclusion This study supports the use of PRO-MAC for 90-day, 6-month and 1-year mortality prognostication to enable needs assessment, end-of-life discussion and hospice utilisation in geriatric oncology. Future research should prospectively validate the risk grouping in geriatric oncology with the eventual aim to implement PRO-MAC in routine oncology and palliative care practice.