Safety of antithrombotic care in Dutch hospitals

Antithrombotic drugs belong to the most used medications in healthcare. About one in ten Dutch inhabitants uses them. Antithrombotic drugs reduce the risk of clot formation caused by pathologies such as atrial fibrillation or venous thromboembolism. Mitigating this risk comes at the cost of a small increase in the risk of bleeding. Often this trade-off is clear-cut and favours the use of antithrombotic drugs. However, in some cases these drugs can result in adverse events. Antithrombotic related adverse events (ARAEs) gained increasing attention in adverse event studies in the Netherlands in the past decades but the exact magnitude remained unknown. This thesis aims to systematically measure the safety of antithrombotic care provided in Dutch hospitals to guide future improvement efforts. The following research questions are discussed: 1) How common are (preventable) antithrombotic related adverse events in Dutch hospitals and what are the circumstances in which they occur? In our research we found that ARAE incidence in deceased patients decreased from 1.20% in 2008 to 0.54% in 2015/2016. Although this is an encouraging result, other findings were concerning. These are: a lacking decline of ARAEs in discharged patients, the relatively high preventability of ARAEs (28.54%) and the increasing trend of involvement of antiplatelet and/or combined antithrombotic therapy in ARAEs over time. 2) How reliable is perioperative antithrombotic management (PAM) and administering VTE prophylaxis in Dutch hospitals? Reliability was operationalised as guideline compliance in this thesis. We concluded that deviations from antithrombotic guideline recommendations were common. Most of these recommendations can be regarded as process measures. When comparing them as such with measures from other clinical domains, compliance is low. Additionally, we observed significant variation in compliance, up to 70 percentage points, between hospitals. This is indicative of a heterogeneous adoption of antithrombotic CPG recommendations in practice. Regarding PAM we found that several surgical characteristics (type, urgency) demonstrated a correlation with the use of non-recommended bridging therapy. 3) How is perioperative antithrombotic management conducted in everyday practice (work-as-done) and how does this relate to predefined procedures (work-as-imagined)? Using the FRAM methodology, we concluded that preoperative PAM is organised differently in practice than what is described in CPGs and related documents. Especially formulating the preoperative PAM policy was organised differently. Various downstream functions, such as self-developed checklists, were set to signal a missing PAM policy, reducing ARAE probability. In the general discussion future perspectives regarding the safety of antithrombotic care are drafted. We suggest investigating the use of FRAM methodology for prospective hazard identification. This is a recognized application of FRAM in other research domains but remains relatively underused within the healthcare domain. Furthermore, we propose progression towards a data driven platform, adopted and recognized by medical proffesionals, to support learning and improvement of antithrombotic care. To get there, barriers concerning unambiguous and conflicting clinical practice guidelines should be tackled and a broad spectrum of medical specialties have to align. We conclude this thesis by making recommendations. Regarding clinical practice, we recommend periodical monitoring of process measures, explore the use of FRAM models for hazard anticipation, and the implementation of antithrombotic case-management in analogy with antimicrobial stewardship. Regarding healthcare policy, we recommend continuing periodical measurement of ARAEs. Furthermore, we recommend the revision of relevant CPG’s and to revise the CPG development process. Regarding further research, we recommend to further investigate the processes of INR measurement in hospitals, since a majority of ARAE’s occurred during out-of-range INR’s. Lastly, we recommend exploring additional or additive methodologies that help quantifying the results of FRAM analyses in healthcare.