Patient Demand Management through Prioritized Access with Time Windows Access Protocol

Problem definition: We consider appointment scheduling workflows for providing access to new patients seeking appointments at specialty group-practice departments and offer a novel demand management mechanism that enables the best use of care capacity through prioritization of requests. Relevance: Considering the increasing demand for healthcare, it is imperative that capacity-constrained specialty departments intentionally manage demand, particularly during the new patient intake process. Methodology: The novel Time Windows Access Protocol (TWAP) prioritizes appointment requests considering the corresponding patient population’s sensitivity to access delays. Under TWAP, scheduling agents are provided, for each patient priority class, a distinct contiguous portion of the booking horizon (i.e., time window) that they can use to search for an available appointment upon a request from Priority n. High priority classes are incentivized to book and attend appointments with low wait offers, while sufficient dilution of requests from lower priority classes are induced with higher (but still, medically safe) levels of wait. Results: We develop a closed-form, computationally efficient model to determine the optimal set of time windows to use for given delay-dependent appointment realization probabilities. We demonstrate the calculation and use of TWAP under strict prioritization and compromised prioritization objectives for a number of experimental settings and a real-life case study. Managerial Implications: TWAP allows direct control of access delays for different priority classes. Due to patients’ sensitivity to wait (characterizable through available data) this is equivalent to controlling fill rates of different priority classes, allowing us to manage patient demand and match it to care capacity. Under TWAP, slot utilization or overbooks are observed, and can be used as signals to self-correct. Other reservation-type prioritization mechanisms are generally effective, but they are hard to compute. They are also hard to implement since they attempt to control or protect capacity and observe access delays or fill rates.