Abstract
Background: Dabigatran and rivaroxaban are common causes of harm, with significant harm associated with inappropriate dosing. Therapeutic drug monitoring can improve clinical outcomes by reducing pharmacokinetic variance. Routinely collected prescribing data provides an opportunity to efficiently evaluate dosing and therapeutic drug monitoring but requires specialist skills to link and extract data sets for pharmacoepidemiological purposes.
Aims: To describe the inpatient dosing and laboratory monitoring of dabigatran and rivaroxaban with respect to clinical guidelines; to describe the impact of laboratory monitoring on inpatient dosing; to identify factors associated with dosing and monitoring; and to visualise the results for clinical governance purposes.
Methods: Azure Data Studio was used to extract data from the local district data warehouse. Results were analysed in SPSS and visualised in Microsoft Power BI. Inpatient prescriptions of dabigatran and rivaroxaban from June 2016 to March 2023 were extracted and linked to relevant laboratory and patient data. For analysis, the final prescription of each admission was used. Results were manually validated to ensure the validity of the retrieval code, and to understand the effect of inpatient therapeutic drug monitoring on prescribed anticoagulant doses.
Results: Data pertaining to 22,937 prescriptions for 10,155 individuals were extracted. Overall, the doses of 7,412 (43%) dabigatran prescriptions and 3,764 (66%) rivaroxaban prescriptions were consistent with guidelines. 8,259 (48%) of dabigatran prescriptions had doses which were higher and 1,560 (9%) lower than recommended; in comparison, for rivaroxaban, 285 (5%) and 1,581 (28%) of prescribed doses were, respectively, higher and lower than recommended. Of the 3,197 (19%) of dabigatran prescriptions with therapeutic drug monitoring, 1,321 (8%) had prior monitoring and 2,241 (13%) subsequent. For rivaroxaban, 543 (10%) prescriptions were associated with therapeutic drug monitoring, with 225 (4%) having prior monitoring and 347 (6%) subsequent. Age, ethnicity, renal function, service, and co-prescribed inhibitors and inducers of p-glycoprotein or CYP3A4 were all associated with both dosing in accordance with guidelines and use of therapeutic drug monitoring in multivariate analysis. Extremes of renal function and co-prescribed inhibitors and inducers had the strongest association for both dosing and use of laboratory monitoring. These results were visualised in a dashboard with the capacity to filter results and consider various subpopulations in isolation.
Conclusions: In Canterbury, inpatient dosing of dabigatran had lower concordance with clinical guidelines than most international studies and patients were mostly at risk of overdosing. Concordance of rivaroxaban dosing with guidelines was similar to other studies and patients were more at risk of underdosing. Both anticoagulants had therapeutic drug monitoring with greater frequency in patients at greatest risk, but the overall rates of therapeutic drug monitoring were low. The study provides proof of concept for the ability to link data sets in the local data warehouse and visualise the results, providing a platform for more efficient local audits and further pharmacoepidemiological research.