This article originally appeared in Oncology Practice Management.
We often compare scheduling infusion appointments to playing a game of Tetris, as most of the effort is focused on fitting patients with appointments of varying lengths into all available resources in the center. Once every resource is scheduled into and there are no more gaps on any of the resources (ie, the Tetris board is completely covered), then no additional patients can be booked.
However, as anyone involved with infusion center operations can tell you, things will likely not go exactly as planned on any given day. In fact, centers often see fewer patients than were originally scheduled. This can happen for a variety of reasons, including the following:
- There can be very little advance notice regarding same-day changes in treatment regimens.
- It is nearly impossible to predict how many adverse reactions to therapy may occur.
- The team may know Mr Smith will likely show up late for his appointment and Mrs. Jones will arrive exactly 10 minutes early, but it is anyone’s guess as to how many other patients will show up early, late, or on time.
All of these factors, and more, can affect how long patients need to utilize resources in the center.
Therefore, even when an infusion center’s schedule appears full at the beginning of the day, when the day is over, it turns out that there were empty chairs that could have been used to treat additional patients. If you combine this standard approach to scheduling patients—even when it is done exceptionally well—with the fact that it very expensive and time-consuming to expand the physical footprint of an infusion center to accommodate growing demand/patient volume, the end result is appointment volumes that are stifled by the physical limitations of a center.
Figure 1 shows an example of a 32-chair infusion center. The height of the bars indicates that this center was able to schedule into the majority of its chairs for most of the day and was even slightly overbooked above the 32-chair capacity at some intervals during lunch hours. The height of the pink line represents the actual patient volume at each interval during the day. Notice that this reaches a peak of 27 chairs by mid-afternoon, and then hovers near the low to mid-20s for the rest of the key treating hours. Although it appeared that the center was going to be full and consistently busy on this particular day, the schedule ended up being quite manageable, and the center had capacity to accommodate additional patient appointments.
What can be done to fully utilize existing resources?
Most electronic health record systems are configured in a way that supports scheduling patients into available resources; however, this is not the most advanced or most efficient method. Changing this scheduling approach and utilizing advanced analysis available through machine learning can help infusion centers fully utilize their existing resources. This analysis can be used to effectively accommodate an infusion center’s patient patterns and to understand several other things about operations within the center, including the following:
- How large the reduction in patient load will be due to variability throughout the day. This can be measured not just by appointment volume, but also by chair hours.
- Where the reduction in patient load can be expected during the day. Will this effect be seen more often in the morning or afternoon? Are long appointments or short appointments more likely to cancel? Which appointment types typically run short? Which appointment types run long, which could offset this reduction?
- What effect will this reduction in patient load have on resources within the center?
With this information, infusion centers can adjust their approach to scheduling to reduce lost capacity and fully utilize their available resources. One way to schedule differently is to strategically schedule over a center’s chair limit at key times during the day, based on the following factors:
- Analysis of patient arrival patterns
- Net same-day cancel/no-show versus same-day add-on rates
- Actual versus expected treatment cycle times.
As shown in Figure 2, the center experiences a fairly consistent volume of same-day cancellation/no-show appointments throughout the day; however, many of these are offset by same-day add-on appointments in the afternoon. In this scenario, the center could more aggressively book over chair capacity in the morning and be less aggressive with this approach in the afternoon. In addition, although it makes sense to those working in infusion centers that their same-day add-on appointments usually come in the afternoons, the number of these appointments that actually occur and net offset to same-day cancellation/no-show appointments is best calculated using deeper data science techniques, which will determine how this information can be applied to the center’s scheduling strategy.
The result of a more strategic and targeted approach to scheduling
The result of being more strategic with a targeted approach to scheduling is that infusion centers can achieve a higher utilization rate for their existing resources. This is generally realized by increasing the volume of appointments treated at the center. Since more patients can be treated within the center without the need to expand the center’s physical footprint via center redesign and/or construction, this is an effective way to increase revenue without increasing costs.
Take the example of the same 32-chair center shown in Figure 3. On this day, the center scheduled patients beyond its chair limit for most of its key treating hours. This is especially true during the late-morning/early-afternoon hours where analysis of the net same-day cancel/no-show versus same-day add-on rate showed this booking strategy would be most effective. The center was able to achieve a peak of 31 chairs a few times during the day and hovered in the high-20s for the rest of the key treating hours.
What is the end result difference? Figure 1 has 87 appointments scheduled with a shrink rate of 11% resulting in 77 appointments completed, whereas Figure 3 has 117 appointments scheduled with a similar shrink rate of 10% resulting in 105 appointments completed. This is an increase of 28 completed appointments using the exact same number of chairs.
Although the example used here is a relatively large infusion center, machine learning can be used to analyze the same metrics and apply the strategic, targeted scheduling practices to any size center.
With consistent pressure to do more with less and large cost/time barriers to increasing physical capacity, infusion centers are forced to examine how to best utilize their resources. Harnessing the power of advanced data science and machine learning can allow these centers to schedule their patient volume more strategically, which in turn unlocks capacity that would have otherwise gone unused. The more efficient use of existing resources ultimately results in infusion centers being able to handle more appointments and provide better access to new and existing patients.
