As medicines become more numerous and often more expensive, healthcare professionals seek new ways to understand which treatments are most likely to yield the best outcomes at sustainable costs.
The so-called learning healthcare system paradigm, an approach to using healthcare data to personalize care, reduce costs, and optimize resources, has been the subject of increased attention in recent years, and in 2012, a group of teaching hospitals in the Netherlands undertook construction of such a system as a means to benchmark the prescribing of biosimilar medicines and to assess how novel therapies impact survival in oncology patients in their system.
The group of 7 hospitals, which together provide care to approximately 13% of the Dutch population, extracts data every week from electronic health records that are used in routine clinical care.
By 2017, the database had accumulated records of 18,741,766 medication orders for 798,632 unique patients with prescriptions.
These records allowed the hospitals to create benchmark reports that show the level of biosimilar prescribing in newly diagnosed patients with inflammatory bowel disease among the hospitals, and also allow users to see how many patients who are treated with anti-tumor necrosis factor therapies switch to other agents. This provided insight that switching due to ineffectiveness is a rare phenomenon.
The database also allows users to generate interactive heat maps that show how each hospital performs in terms of it’s prescribing; individual colors in the generated heat maps demonstrate how much individual hospitals deviate from the overall population in terms of the drugs prescribed.
Although recently established, we believe that having these data timely and in a uniform format largely broadens the opportunities for direct observations on trends in in-hospital medicine utilization and studying these trends concerning relevant patient outcomes.