Prediction modelling in older adults: Insights from the COVID-19 pandemic

Older individuals are significantly impacted by the COVID-19 pandemic, highlighting the need for accurate prognostication in this high-risk population. This thesis aims to systematically evaluate COVID-19 prognostic models in older individuals across different healthcare settings and pandemic waves. Additionally, we investigated the added value of important older population relevant predictors to improve risk prediction in older adults.
Chapter 2 described a peer reviewed protocol to systematically evaluate the performance of COVID-19 prognostic models and scores for mortality risk prediction in older populations (≥ 70 years of age) across three health-care settings: hospitals, primary care, and nursing homes. across hospital, primary care and nursing home settings.
Chapter 3 presented the results of a retrospective external validation study with 14,092 older individuals in six validation cohorts (three hospital, two primary care and one nursing home cohort) in the Netherlands. Based on a living systematic review of COVID-19 prediction models using Prediction model Risk Of Bias ASsessment Tool for quality and risk of bias assessment and considering predictor availability in validation cohorts, six prognostic models predicting mortality risk were validated (GAL-COVID-19 mortality, 4C Mortality Score, National Early Warning Score 2-extended model, Xie model, Wang clinical model, and CURB65 score). All prognostic models performed poorly particularly in terms of calibration. The 4C Mortality Score appeared as the most discriminative (C-statistic in hospital cohorts: 0.66 to 0.74) and well calibrated (calibration slopes: 0.82 to1.15, calibration-in-the large: 0.78 to 0.03) for predicting in-hospital mortality after COVID-19 infection among the validated COVID-19 prognostic models.
During the COVID-19 pandemic, dynamic factors such as governmental policies, improved treatment, prevention options, and viral mutations changed the incidence of outcomes and possibly changed the relation between predictors and outcomes.
Chapter 4 assessed the influence of dynamic context of the pandemic on the predictive performance of mortality predictions over time in older patients hospitalized for COVID-19. A prediction model for in-hospital mortality was developed during the in the first wave and temporally validated it in subsequent waves. Compared to the first wave, a slight decrease in discrimination and calibration was observed in the second and third waves, while a much larger decrease occurred in the fourth wave. These findings highlight the importance of ongoing data collection, performance monitoring, and model updates to main predictive accuracy during a pandemic.