The cost-effectiveness of endovascular therapy (EVT) compared to tissue plasminogen activator (tPA) alone for acute ischemic stroke (AIS) has been established in the literature. However, decision-makers still face challenges of how to best deliver EVT in a timely manner to maximize patient outcomes while minimizing the burden to the healthcare system, given that AIS has time-dependent treatment outcomes. The objective of this presentation is to report an optimization approach for improving health system value and outcomes for patients with AIS who are eligible for EVT in Alberta.

An economic model was developed to compare combinations of “mothership” (transport directly to a comprehensive stroke center [CSC] to receive tPA and EVT) and “drip-and-ship” (transport to a primary stroke centre to receive tPA, followed by transport to a CSC to receive EVT) methods across Alberta. The model considered geographical variation and searched for the best delivery methods through a pairwise comparison of all possible strategies. The controlled variables including in the model were population densities, disease epidemiology, time/distance to hospitals, available medical services, treatment eligibility and efficacy, and costs. Patient outcomes were measured by functional independence. The model defined optimal strategies by identifying the transport methods that produced the highest probability of improved health outcomes at the lowest cost.

The analysis produced an optimization map showing optimal strategies for EVT delivery. The lifetime cost (standard deviation [SD]) per patient and likelihood (SD) of good outcomes was CAD 291,769 (CAD 11,576) [USD 226,207 (USD 8,975)] and 41.82 percent (0.013) when considering optimal clinical outcomes, and CAD 287,725 (CAD 4,141) [USD 223,097 (USD 3,211)] and 41.67 percent (0.016) when considering optimal economic efficiency.

Our model reduces the gap that exists between health technology implementation and cost-effectiveness analysis; namely, neither fully addresses relative efficiency driven by geographical variation, which may misrepresent system value in local settings. Implementation strategies generated in our model capture full values in terms of patient outcomes and costs.