So we have written a number of things about practices from manufacturing or other service settings could impact the health care industry. Now comes word that the National Science Foundation is funding research on porting ideas from the airline industry to medical clinics (Reducing Health Care Costs, Improving Care, Sept 20, U.S. News & World Report). This might strike you as good news. After all, US airlines do a remarkable job managing complex operations while avoiding dramatic failures. I am not saying that every flight is on time, but the number of crashes and fatalities is remarkably low. Perhaps they could help physicians and care givers avoid catastrophes such as medication mix ups.
While researchers are looking at that, the NSF is also funding work that relates more to customer service. That’s right, someone is looking to airlines on work that impacts customer service.
Let that sink in for a moment.
Specifically, they are looking at overbooking, the practice of selling more tickets than you have seats on the plane. For an airline, flying an empty seat is costly. The marginal cost of having one more passenger is essentially zero so foregoing the chance to sell one more ticket leaves several hundred dollars on the table. For a medical clinic, idle capacity is similarly costly.
The center is using mathematical algorithms and other models to simulate patient flow, in order to replicate real life situations, and experiment with different ways of doing things, including “strategic’’ overbooking. “Like with an airline, patients don’t always show up,’’ he said. “That’s a waste of the doctors’ and nurses’ time. No-shows back up the queue, and it’s very costly. So some amount of ‘optimal’ overbooking might be part of a solution.’’
For example, the Harvard Vanguard Medical Associates obstetrics and gynecology clinic in Quincy, Mass., is trying Northeastern’s approach on a trial basis. Last year, its no-show rate topped 18 percent, and the clinic lost an estimated $446,000 in wasted clinical time and appointment delays. “When you ask health care professionals what is causing them misery, almost every one of them will talk about no-shows,’’ Benneyan said.
In this instance, a Northeastern engineering team used computer modeling to identify the types of patients most likely to skip their appointments, which here turned out to be those scheduled for an annual exam, and to figure out how much to overbook each day. The team then used computer simulation to experiment with various ways of overbooking, such as clustering the double-booked appointments at certain times of the day, or spreading them throughout the day.
So I have not seen the research that the folks at Northeastern have produced but I have seen a number of other papers on this topic and the whole problem of overbooking in health care is really, really hard. It’s a relatively hard problem for airlines, as well, but there are some key differences that make overbooking in a clinic more challenging. For example, think about who you deem to be overbooked. Suppose an airplane has 100 seats and has sold 101 tickets to passenger who then all have the nerve to show up. Who do you boot to off the flight? The airline likely doesn’t do this on a first-come-first-served basis. If first-come-first-served is defined by when customers bought tickets, it is likely that those buying later paid more so you may want to keep them happy and not risk losing their business now or in the future. If first-come-first-served is based on when they checked in, it is likely that at some point you stop giving out seat assignments so you have a choice of whom to bump. Which customer incurs the cost of overbooking is then a choice and may even be handled by asking for a volunteer.
What is clear, however, is that only one customer is going to be inconvenienced — taking one for the team as it were. At a medical clinic, the entire team may have to take the hit. Overbooking now means scheduling more patients then you have capacity to handle in a time slot. If that time slot occurs early in the session and everyone shows up, the clinic then runs late all day. All patients are then inconvenienced.
This points to another difference between airlines and clinics. Airlines have a pretty good idea of what bumping a passenger costs. There are travel vouchers for volunteers, payments to other airlines and government mandated reimbursements. There is also potentially long-term goodwill costs for passengers who perceive that they have been abused. (At the other extreme I knew grad students who planned travel so they could get bumped and score vouchers.) At a clinic, the cost of overbooking depends on how the clinic values the waiting time of customers since there are no clear alternatives such as sending the patient to a different office or offering vouchers for later care. That is, the clinic is trading off real costs of overtime or idle nurses with nebulous customer waits.
The problem is that the optimal solution will depend on these waiting costs in a nontrivial way. If the costs are sufficiently low, there is a simple solution: Book every patient for the earliest slot of the session. Sure customers will wait a lot, but resources will only be idle if every patient for the session has been seen. Obviously, this is not a long-term solution. If patients have any choice, they will find another clinic. Even if patients don’t have choice, no one in the clinic will want to be dealing with patients who have had to wait hours for routine care.
This is not to say that clinics shouldn’t overbook. There is a legitimate need to hedge against idle resources and if no shows are a fact of life, overbooking is pretty much the only available answer. The other alternative, of course, is to reduce no shows. If annual exams are the most likely appointments to be skipped, why not change how you take those appointments — say only letting them be made two weeks out or having aggressive procedures to confirm these appointments? Yes, this could add cost but it may be simpler and more effective than overbooking.