Medical Decision Making
to help improve patient care
Diagnostic and treatment options continue to increase, both in number and in complexity. We aim to support patients, clinicians and healthcare policymakers in making the best decisions on the use of diagnostic tests and therapeutic interventions. Medical decision sciences concerns the quantitative study of decision problems in individual patient care, and can provide that support, particularly when it is carried out within a cost-effectiveness framework. The field is multidisciplinary, with contributions from economy, mathematics, epidemiology, bio-statistics, and psychology. A specific feature of our work is the application of regression-based prediction models, which predict the presence of disease, or the outcome of a disease process, given patient and/or care characteristics. We also focus on epidemiological, clinical and ethical aspects of end-of-life decision-making.
Medical decision-making is relevant throughout life. One example concerns the very beginning of life. When women wish to become pregnant but do not have a normal ovulatory cycle, several treatments are available. We devised a patient-tailored, cost-effective treatment algorithm, based on individual patient characteristics, that predicts ongoing pregnancy chances. Another example comes from our work on medical decision-making at the end of life. We studied the practice of deeply sedating severely ill patients during their dying process. The distinction between sedation and euthanasia, that is the active ending of life, is sometimes hard to make. But our careful study of both practices has shown that sedation is most often used as a final means to relieve severe symptoms, whereas euthanasia is predominantly a ‘way out’ for patients who fear the loss of autonomy and dignity at the end of their lives.
Public Health Impact
Our periodic studies on euthanasia and other end-of-life decisions play an important role in the debate on public regulations in this field. Insight in the typical characteristics of various medical practices at the end of life, such as euthanasia, palliative sedation, and the use of high dosages of morphine, has proven to be very helpful in policymaking and the development of guidelines. We have been the first to describe the characteristics of the practice of sedating patients with severe symptoms prior to death, a practice that is currently amply discussed in the media. The distinction between euthanasia and palliative sedation, and whether or not both practices are complementary or exchangeable, are among the key issues in this debate. The results of our study have been amply used in an analysis of the Health Council of the ethical questions that are related to this practice. The Royal Dutch Medical Association also used our results in the development of a guideline for palliative sedation.
Research Highlight: Individualized ovulation induction treatment
(Eijkemans et al., Hum Reprod 2005)
When women wish to become pregnant but do not have a normal ovulatory cycle, conventional treatment starts with a simple and cheap drug (Clomiphene Citrate, CC). If this treatment is not successful, hormonal treatment may be considered next (exogenous gonadotrophins, FSH), and finally in-vitro fertilization (IVF), the most burdensome and costly treatment. However, for patients who have a poor response to the initial treatment step(s), another strategy might be better. In this study we devised a patient-tailored cost-effective treatment algorithm, based on individual patient characteristics, that predicts ongoing pregnancy chances.
We created 16 prognostic groups among 240 women visiting a specialist academic fertility unit, who were followed in a prospective cohort study. The 16 groups were defined according to the absence or presence of four predictive patient characteristics: age > 30 years, absence of ovulation, elevated androgen levels and obesity. The chances of response with each of the three treatments were calculated using regression models that predict ongoing pregnancy. Treatment costs were based on the data of the same 240 patients.
Overall, we found that the current strategy (CC, followed by FSH and IVF) generated more pregnancies against lower costs than any other strategy. However, for women with relatively low pregnancy chances cost per pregnancy were over € 200,000, and it was more efficient to skip the FSH treatment step when CC had failed.
This type of research demonstrates the value of prognostic modeling: the cost-effectiveness of treatment differs according to individual patient characteristics, and the preferred treatment strategy depends on the prognosis. Prognostic models are essential tools for such more individualized assessments of treatment efficiency. This study supports a wider application of prognostic models in clinical practice.