PATIS – A patient safety intelligence system and framework

Recent research showed that medical errors occur quite often. The issue of patient safety has thus found strong interest in scientific discussions in the last year. One challenge is to define appropriate indicators for patient safety within a health care institution. Here, the secondary use of clinical data will play a prominent role in the future. Secondary use of clinical data describes the technical and methodological means to select, integrate, aggregate and present data from the routine medical and nursing documentation in the form of so-called indicators. The vision of the PATIS project is the development and evaluation of such technical and methodological means in order to develop meaningful indicators for patient safety. First, we will develop a framework and a methodology for the secondary use of clinical data. Here we will refine and integrate earlier work in the area of indicator development and secondary use. The framework will describe a systematic process how to develop patient safety indicators from clinical routine data and how to establish a so-called patient safety intelligence system. In a clinical case study, the framework and methodology will be implemented and validated. Together with medical and nursing partners, we will develop and implement a patient safety intelligence system. The PATIS project will contribute to patient safety in healthcare institutions. In particular, the PATIS project will help to develop and monitor patient safety indicators.

Within the PATIS project, we contributed methodologically to patient safety. Patient safety means that the risk that patients are harmed during their stay in a health care institution is minimized as far as possible. Patient safety is an essential issue for health care, as around 10% of patients in a hospital are subject to such risky situations or even harmed by errors (e.g., medication errors). We approached this important topic from the perspective of medical informatics. Thus, we first investigated which indicators are helpful to measure patient safety. We then conducted case studies in various fields such as neurology or gerontopsychiatry. We found here that a large amount of data is already available that may be helpful to monitor patient safety indicators without the need to collect additional data. Then, we investigated in more detail for a selected set of indicators whether it is indeed practically possible to used valid routine patient data that allow monitoring these indicators. Here, we used routine clinical data from patients in neurology or gerontopsychiatry. We were able to show that for some patient safety indicators, we can indeed use routine clinical data. Routine data has the advantage that it is already available so that patient safety could be monitored regularly. However, routine clinical data is not always of sufficient quality. We thus performed some methodological work on the question of the quality of routine clinical data. For example, routine data from neurological monitoring of an intensive care patient comprises many data generated every second. This data is automatically monitored, but manual activities (such as injection of medication) may have an unnoticed impact on this data. Thus, routine clinical data must be synchronized with documented manual activity to avoid drawing false causal conclusions. Our cooperation partners in neurology used the clinical routine data for in-depth analysis on specific clinical questions based on our methodological work. For example, they found that some patient outcome indicators might worsen when a particular standard therapy is applied in a specific sub-group of patients. Our project results show that a careful analysis of available clinical routine data can help monitor health care quality and thus contribute to patient safety. Overall, our work contributes in a methodological way to the idea that available clinical routine data can be used to derive meaningful information for quality management and patient data. Given that more and more routine patient data is available electronically and standardized, we expect that application areas for using this data will further grow.