Medical Decision Making
“Our mission is to contribute to optimal evidence-based and personalized decisions in healthcare…. We apply our methods in collaboration with multiple clinical groups within and outside Erasmus MC to directly improve patient care."
Medical decision-making aims to support patients, clinicians and healthcare policymakers in making the best decisions about diagnostic, therapeutic and other medical interventions. Our mission is to contribute to optimal evidence-based and personalized decisions in healthcare. We work on the development and application of methods to measure and quantify quality and effectiveness of care, and to translate group-based information to decision making in individuals. Specific interests include prediction modelling, design of randomized trials, and comparative effectiveness research (CER). We apply our methods in collaboration with multiple clinical groups within and outside Erasmus MC to directly improve patient care.
Highlights
Comparative effectiveness research: methods and applications
Our methodological work includes methods to exploit existing between-center heterogeneity in treatment approaches for comparative effectiveness research. In diseases with limited evidence on optimal treatment, treatment choices are usually based on experience and local protocols. Such practice variation is in principle undesirable, but also can be exploited for comparative effectiveness research. We study methods to compare outcomes between centers with different standard treatment approaches to identify best practices. In a simulation study we showed that these methods allow estimation of causal treatment effects in observational data (1). We apply our methods in various clinical fields, including Traumatic Brain Injury (TBI). For example, using data from CENTER-TBI study, a large European prospective observational study, we found that there is significant variability in fluid management of TBI patients across centers, with more positive fluid balances being associated with worse outcomes (2). These results suggest that aiming for neutral fluids balances contributes to improved outcome, and thus directly can inform clinical care.
COVID outcome prediction in the emergency department (COPE)
We developed COPE for prediction of in-hospital death and need for intensive care when patients with suspected COVID-19 present at the Emergency Department. Developed in patient data from the first wave of the pandemic, based on six quickly and objectively obtainable predictors – age, respiratory rate, LDH, CRP, albumin and urea – COPE discriminated well and was well-calibrated in patients admitted to hospitals in the first and second wave of the pandemic, both for predicting in-hospital death and for ICU admission.
The development of COPE is published Open Access in BMJ Open (doi: http://dx.doi.org/10.1136/bmjopen-2021-051468). COPE is implemented in a web app (and in mobile apps apple / google play. COPE discriminated well in 13 clinics of Northwell Health in New York City.
From interviews, we learned that health care providers appreciate the added value of models like COPE for decisions about hospital admission, if these models are externally validated, implemented in electronic apps, and giving interpretable results. Patients and their loved ones appreciate the information about the severity of the disease, to potentially inform treatment decisions.
