ESTIMATING HEALTHCARE COSTS FOR DECISION-ANALYTIC MODELS IN ATRIAL FIBRILLATION USING ADMINISTRATIVE DATA

Purpose: To obtain high-quality estimates of direct health care costs of patients with atrial fibrillation (AF), using administrative data, to inform decision-analytic models.

Method: We identified 109,002 patients diagnosed with AF, between January 1, 2003 and June 30, 2011 in Ontario using administrative data housed at the Institute for Clinical Evaluative Sciences. Patients were identified based on hospital admission for AF (ICD-10-I48). Patients were then followed forward until the occurrence of a relevant clinical event (i.e. minor ischemic stroke/transient ischemic attack (TIA), moderate to severe ischemic stroke, myocardial infarction, extracranial hemorrhage, intracranial hemorrhage, multiple events, death from an event or death from other causes), based on an existing state-transition model for AF. Patients who died within 30 days of experiencing an event were assumed to have died as a result of that event (i.e. a fatal event). These costs were pooled together for all events to yield an aggregate cost of dying from any clinical event. Costing of resources was based on a previously validated algorithm. Costs associated with each health state were based on a per day cost of being in that health state and stratified as acute (<30 day) and long-term (>30 day) costs. Average 30-day costs compatible with the state-transition model cycle length were then calculated from per-day costs. All costs are reported in 2012 Canadian dollars.

Result: The average cost associated with experiencing a fatal clinical event was $42,700. The cost associated with dying from all other causes was much smaller ($12,750). The clinical event associated with the highest acute cost was an intracranial hemorrhage ($22,260), followed by a moderate to severe ischemic stroke ($19,860). The clinical events associated with the lowest acute costs were minor ischemic strokes/TIA ($12,470) and extracranial hemorrhages ($12,210). Patients with AF who did not experience a clinical event, cost $1,560 per 30 days, on average. Trends in long-term costs associated with the clinical events were similar to those seen with acute costs. However, long-term costs associated with moderate to severe ischemic strokes were higher than those for intracranial hemorrhages.  

Conclusion: This study provides an estimate of acute and long-term costs associated with major clinical events experienced by AF patients, based on real-world data. These costs can also be used to inform state-transition models for AF.