Faculty/Clinicians/Staff: Amy Cohn,  Amy Rothberg
Students: Henry Ballout, Cara Cheshire, Moses Chan, Erick Dagenais, Daniel Huang, Joanna Fleming, Nate Janes, Madalina Jiga, Jared Kott, Anna Learis, John Li, Cameron Misko, Alex Mize, Paige Mollison, Matt O’Callaghan, Elizabeth Olin, Yiming Qiang, Abhilash Rao, Nina Scheinberg, Pranjal Singh, Amber Smith, Joseph Sorenson, Haitian Sun, George Tam, Hwon Tak, William Yang

Project Contact:

Project Synopsis:

Timely access to outpatient healthcare, and in particular to specialty care, is a significant challenge nationally. Although insufficient numbers of providers/appointments can contribute to this, existing capacity may also be scheduled inefficiently. Our objective was to develop a methodology to more deeply understand the relationship between scheduling policies, patient cancellation behaviors, and capacity utilization.

To do so, we partnered with the Metabolism, Endocrinology, and Diabetes (MEND) clinic at the University of Michigan. MEND has more than twenty-eight physician providers, eight diabetes educators, diabetes nurses, and other support personnel. In collaboration with the MEND clinic, we developed a unique SQL (“structured query language”) database specially designed to track schedule changes longitudinally. Each record in the database corresponds to an appointment slot and its status on a specific date (available for schedule or assigned to a specific patient).

By collecting daily schedule ‘snapshots’ of this form, we are able to observe both how appointment slots evolve over time (when are they filled, how often are these cancelled and refilled, etc.) and how patients behave (when do they schedule, with how much appointment lead time, how often do the cancel, when do they cancel, etc.). These insights can, in turn, lead to new scheduling policies to better utilize clinic capacity.

In addition to developing this database structure, we conducted a case study using scheduling data from the MEND clinic, comprised of three providers, over three thousand patients, and more than twenty-five thousand appointment slots over the course of three years. Key observations from the analysis include the following:

  • Only a limited number of appointments are available for short-term scheduling
  • Appointments are canceled (and sometimes but not always rescheduled) with high frequency
  • Rescheduling takes place close to the appointment date and increases in likelihood as a function of lead time (i.e. how far into the future the appointment was booked)
  • As a result, patients are often seen far later than their originally-requested date because rescheduling incurs significant additional delays
  • Nonetheless, appointment slots often go ultimately unfilled.

These results suggest the potential value of modified scheduling policies, including scheduling closer to the desired time of appointment and better wait-list management, to ensure that patients are seen closer to the desired date of appointment. Although our case study was specific to the MEND clinic, our methodology is easily extendible to other outpatient specialty clinic environments.

Papers, Posters, & Presentations: