Econometrician Merel van Diepen (LUMC) will develop predictive models related to the quality of life of kidney patients. These models can provide patients with information about what to expect in terms of disease-related aspects that they themselves consider important. Van Diepen received a Kolff grant from the Kidney Foundation for this research.
In the Clinical Epidemiology department, Van Diepen has been developing models for predicting clinical outcomes in individual patients for some time now. In the first place, the aim is to be able to inform patients about what to expect in terms of treatment or disease course. “That is important information for the patient. In addition, prediction models can identify high-risk groups that may benefit from an intervention.”
More and more models
Because computers can do more and more, the number of models is growing. These are becoming increasingly complex, but according to Van Diepen, a clear clinical question is often missing. It also happens that impractical methodological choices are made during the development, making a model hardly applicable. “More and more models are also being developed within nephrology. But few of them are actually implemented. That’s too bad. Because if doctors have access to data and use it to enter into a conversation with patients, this has great potential for, for example, patient-oriented care and shared-decision making. Data is certainly not intended to replace the doctor, because he knows the patient best and has a clinical view. But data can provide support for conversations or decisions. That is why I hope to get more models into practice.”
Not just clinical measures
The application of models starts with external validation, in other words testing them in new patients. Van Diepen has conducted studies in recent years in which she has compared models. However, she gradually realized that models often focus on clinical outcome measures, such as decline in renal function, initiation of dialysis or death. “But there is much more to patients than just clinical measures, such as symptom burden and quality of life. More and more attention is being paid to patient-related outcomes, i.e. outcomes that are important for the patient. Patients can use PROM questionnaires to indicate how they feel and what their quality of life is. For example, if complications of a treatment occur, this has more impact for one patient than for another. It is good if such aspects are discussed in the parlor. Fortunately, that is happening more and more.”
Give more control
With the current grant from the Kidney Foundation, Van Diepen will develop prediction models for patient-related outcomes with the aim of giving patients an idea of what awaits them in different aspects of their disease and treatment. This can give them a little more control over their situation and perhaps reduce stress about their future. “And maybe we can give them an idea about the quality of life when, for example, whether or not to start dialysis. That can help with their choice,” says Van Diepen.
Van Diepen is now supervising two PhD students. They have started developing a questionnaire for patients about outcomes that are important to them: what information do patients need and at what point in the care process? This questionnaire will be distributed in the short term via the patient association. “In addition, we map out which types of models are already available, which are still missing and which options are available for implementation. We also want to validate existing models. It will be a big job, because there are hundreds of models.”
New techniques
The next step will be to create models that are still missing. That requires techniques that are not yet widely used. “Clinical outcomes are often a yes-no answer to a question, for example whether or not to start dialysis. Outcomes for quality of life are much broader,” explains Van Diepen. “Our research is therefore also methodological: how can we make predictive models like that? And if we have and apply such a model, will it also have an impact on clinical practice? Do we see an improvement in quality of life or health somewhere? Such impact studies are hardly ever done.”
The research project started in September 2021 and will last four years. There is also good cooperation with the patient association. Van Diepen finds this enriching: “We held discussions with patients to develop our questionnaire. They make the research more lively and provide inspiration. The experiences of patients, that’s what you do it for.”
What is and does an econometrician do?
Econometrics is a mix of statistics and economics. Merel van Diepen explains: “Econometricians design models for economic processes. Initially I focused on consumer behaviour, but I preferred to do something with more social relevance. At the Department of Clinical Epidemiology of the LUMC I was able to develop models for kidney patients. It was new for me to work with doctors, but I’ve been doing that for about twelve years now. With my methodological and statistical knowledge and experience, I really feel part of this field.”