PRedicting In-hospital mORtalITY in older patients using general practice data: the PRIORITY retrospective cohort study
Background and relevance
In an emergency situation, the GP can refer the older patient to the hospital. However, the risk of in-hospital mortality is increased. The GP will discuss with the patient and his relatives whether hospitalization is desirable in emergency situations. An estimation of the risk of in-hospital mortality helps with this choice. Our project will provide a reliable prediction model that can be used directly by the GP in the discussion with the patient and his relatives about the risk of in-hospital death after referral. Such a reliable model is not yet available.
Problem definition and goals
The care for the older patient leads to increasing pressure on the general practice. Older patients often have multiple chronic diseases and medications and may have cognitive impairment. This makes care complex. Emergency situations that may require hospitalization are common. However, older patients have an increased risk of mortality during hospitalization. The Dutch College of General Practitioners has placed research into risk factors with which in-hospital mortality can be predicted high on its research agenda. The aim of our project is therefore to identify risk factors that are independent predictors of in-hospital mortality in the older patient. We use data from general practitioner files.
Plan of approach
We propose a retrospective cohort study of eighteen months. We use data from more than one and a half million patients from two large Dutch General Practice research networks. We combine this data with information from hospitals and mortality figures from Statistics Netherlands. The cohort consists of people aged 65 and older who were hospitalized between January 1, 2012 and January 1, 2022. Our research team consists of general practitioners, data analysts, a geriatrician and a representative of the Dutch Patient Federation. The project consists of 4 work packages (WP). In WP1 (months 1-6) we link the different data files and compile the retrospective cohort. In WP2 (months 1-6) we conduct a literature study and conduct group interviews with GPs and patients. We want to gain a better insight into potential risk factors and the usability of these factors in daily care. In WP3 (months 7-18) we analyze the cohort data. We describe patient characteristics and take into account factors such as age, gender and socio-economic status. We then calculate hazard ratios with 95% confidence intervals. Finally, we build prediction models through traditional logistic regression analysis and advanced analysis based on artificial intelligence. In WP4 (months 1-18) we bring our results to the attention of colleagues through (inter)national journals, conferences and social media. We also share the results with our professional association, general practitioner training courses and universities of applied sciences.
Our study population consists of general practitioners and patients aged 65 and older.
WP1 provides the database of the retrospective cohort. After WP2, we have a clinically relevant overview of potential risk factors associated with in-hospital mortality, based on literature and the opinion of GPs and patients. After WP3, independent risk factors for in-hospital mortality have been identified and the most appropriate predictive model determined.
Our results directly impact the daily practice of the GP, because the prediction model helps in the discussion with the patient about whether or not to refer to the hospital. In particular, we bring our results to the attention of the Dutch College of General Practitioners and general practitioners with expertise in elderly care and palliative care, so that they can include them in clinical guidelines and education.