With the nationwide shortage of primary care physicians (PCPs), practices are finding it increasingly difficult to deliver two vital and conflicting patient-centered metrics: 1) Timeliness i.e. ensuring patients get access to care as quickly as possible; 2) Patient-physician continuity, i.e. ensuring patients see their own PCP as much as possible. This proposal will develop new stochastic optimization approaches for addressing timeliness and continuity through the lens of physician flexibility - the ability of a physician to see patients of other physicians. Physicians in primary care are inherently flexible; however, the management and efficient use of this flexibility has to consider the balance between timeliness and continuity. The research will explore the design and optimal use of flexibility at different but interlinked levels of the capacity planning hierarchy involving two streams of uncertain demand, non-urgent and urgent, each with different timeliness and continuity requirements.
If successful, the results will be used to develop optimal policies and insights on capacity allocation and dynamic patient scheduling under different flexibility configurations, as well as to develop heuristics that can be easily implemented in practice. Since primary care has the broadest possible reach, improved access and continuity has the potential to benefit the population as a whole. Furthermore, the findings should be applicable to other healthcare and general service settings where timeliness and continuity are essential. Close collaboration with a primary care physician trained in operations research and a wide range of clinics, rural and urban, academic and non-academic, will ensure the quick transfer of results. The case study of a community clinic in Western Massachusetts will be used to disseminate research findings in undergraduate and graduate courses.
With the nationwide shortage of primary care physicians (PCPs), practices are finding it increasingly difficult to deliver two vital and conflicting patient-centered metrics: 1) Timeliness i.e. ensuring patients get access to care as quickly as possible; 2) Patient-physician continuity, i.e. ensuring patients see their own PCP as much as possible. This proposal developed stochastic optimization approaches for addressing timeliness and continuity through the lens of physician flexibility - the ability of a physician to see patients of other physicians. Physicians in primary care are inherently flexible; however, the management and efficient use of this flexibility has to consider the balance between timeliness and continuity. The research explored the design and optimal use of flexibility at different but interlinked levels of the capacity planning hierarchy involving two streams of uncertain demand, non-urgent (prescheduled) and urgent (same-day/acute), each with different timeliness and continuity requirements. Two other questions were addressed: 1) how a practice should schedule and sequence patient appointments (routine checkups for adults and children; chronic condition follow-ups; and acute same-day requests), requiring different and uncertain amounts of time with provider and nurse, to minimize in-clinic patient waiting and provider idle time; and 2) how an outpatient teaching clinic should allocate its patients with different diagnoses and chronic conditions (case-mix), to medical residents to promote consistency in educational training while promoting access and continuity in the long-term. To explore these questions, mathematical models including stochastic, dynamic and integer programming formulations were developed as appropriate for the setting at hand, and efficient computational methods were created to find the optimal solutions. The practical insights obtained through this approach are as follows: 1) When providers flexibly pool their capacity so as to satisfy short-notice requests (for which continuity is presumed to be less important) access to same-day appointments improves; at the same time, patient-provider continuity for prescheduled requests is maintained. Interestingly, flexibility may or may not increase access for prescheduled patients. When the practice is overloaded, pooling same-day capacity results in reduced slots for prescheduled patients, while the opposite happens for underutilized practices. Using providers’ inherent flexibility to serve the prescheduled patients as well, however, yields minimal additional benefits in access while depriving patients with chronic conditions from a much needed strong patient-provider bond. 2) The earlier in the workday prescheduled appointments are scheduled, the better a practice’s ability to satisfy same-day requests during regular work hours; continuity for same-day patients is also maximized. Since it is unrealistic for all prescheduled appointments to be scheduled in a single block in the morning (All-Morning), a two blocks approach – one block early in the morning, and one block early in the afternoon – is more feasible and also performs almost as well as All-Morning. Furthermore, simple rules-of-thumb that dynamically assign patients to same-day slots are shown to provide near optimal solutions. 3) While patient face times with nurses and providers vary greatly even under the same ailments, a comprehensive time study of the flow of 400 patients in a family practice reveals that a simple, practical classification into three groups—prescheduled high complexity, prescheduled low complexity, and same-day—captures large differences among these patient requirements, and can be used for improved scheduling. 4) In creating schedules for a workday, considering a two-step model of patient flow – i.e. accounting for time spent by the patient with nurse as well as provider – significantly outperforms schedules based on provider time only. Furthermore, "flexible" appointment sequences that allow adequate time-of-day options for patients, when optimally spaced, can provide an adequate balance between patient in-clinic waiting and provider idle time (which is the key to the practice’s financial viability). An Excel tool to facilitate scheduling incorporating these aspects can be found at http://blogs.umass.edu/hyunjuno/2013/04/24/instructions/. 5) In outpatient teaching clinics, a simple sorting based assignment of patients to residents (the most complex patient assigned to the resident with the least total case-mix complexity) achieves equity in training experience. To facilitate adequate transition in care (i.e., avoid loss of continuity) for reassigned patients, a staggered and incremental implementation of the assignment algorithm is shown to be effective. Thus the guidelines that emerge from this research have the potential to impact the management of outpatient primary care practices. Furthermore, the findings should be applicable to other healthcare and general service settings where timeliness and continuity are essential. Results have been disseminated in both engineering and clinical journals and conferences, and through the PI’s website, people.umass.edu/hbalasub. Findings have either influenced or are currently in the process of being implemented at the Atkinson Family Practice (Amherst, MA) and at the internal medicine teaching clinic at the Massachusetts General Hospital (Boston, MA). In addition to supporting two female doctoral students and two masters’ students, results of this research were disseminated in the PI’s graduate and undergraduate courses, including a special graduate course on healthcare operations research.
