Validation of Medication Proxies for the Identification of Hospitalizations for Major Adverse Cerebro-Cardiovascular Events

PURPOSE: To date, complete and long-term registrations of diseases and events are not available in every situation. As a useful proxy, medication usage data is very promising. For instance, real-world dispensing data from pharmacies are attractive because of the high validity of drug dispensing data, large sample sizes, and long-term registration. However, before application as a proxy, validity must be assessed. Therefore, in this study, we aim to assess the validity of various medicines used as a proxy for major adverse cardio-cerebrovascular events (MACCE), that is, to identify an incident or previous hospitalization for a MACCE.

PATIENTS AND METHODS: Using the claims database of a large Dutch healthcare insurer, we estimated the concordance between hospitalization claims for MACCE and specific claims for dispensings to treat MACCE in a cohort of patients on primary preventive antihypertensive and/or antihyperlipidemic therapy between 2013 and 2020.

RESULTS: In a cohort of more than 110,000 patients, a dispensing of either vitamin K antagonists, platelet aggregation inhibitors, or nitrates was predictive of an incident hospitalization for a MACCE between 2013 and 2020, with a sensitivity of 71.5% (95% CI: 70.4-72.5%) and specificity of 93.2% (95% CI 91.1-93.4%), and any history of hospitalization for a MACCE (prevalence) with a sensitivity of 86.9% (95% CI: 86.5-87.3%) and specificity of 81.9 (956% CI: 81.6-82.1%), while positive predicted value remains low. Sensitivity analyses across age, sex, and patients with asthma/COPD or diabetes showed a similarly good performance.

CONCLUSION: Claims for the dispensings of vitamin K antagonists, platelet aggregation inhibitors, and/or nitrates can be validly used as a proxy for new and previous hospitalization for MACCE in patients on primary preventive therapy. Further study is required to assess the validity of such dispensing claims for non-hospitalized cerebro-cardiovascular events and whether the results are generalizable in non-Dutch cohorts.