This week’s The Numbers Guy column in the Wall Street Journal looks at how long patients wait for care (Long Medical Waits Prove Hard to Cure, May 25). The setting to have in mind is not how many minutes past your appointment time you spend in the waiting room. Rather, focus on actually getting an appointment to be seen for an ailment or to get a procedure scheduled.
That seems like a fairly straightforward question. One just needs to track how many patients have been referred for a procedure, when they are referred, and when the procedure is performed. The problem is that no one wants to just know what the wait is; they want to manage that wait. That means that targets will be set and comparisons made. That’s where things get tricky.
If you measure how long patients coming off a waiting list have spent on that list, a hospital has little incentive, while under evaluation, to clear those who already have been waiting longer than average. As soon as they are cleared, the hospital’s numbers get worse.
Measure the percentage of patients seen within, say, 48 hours, and those who can’t be seen in that time might instead find themselves waiting much longer, as earlier slots are saved for patients who call up later and can be slotted in the time frame, thus boosting a health provider’s numbers.
Count how many people are on a waiting list for a specialist appointment or nonelective surgery, and the provider being evaluated might change the definition of how long patients have to wait to be included on the waiting list.
Such responses have stymied efforts to cut waiting times and to determine if changes meant to alleviate delays have been effective.
“Any waiting-time measure can be thwarted or misrepresented,” says Michael Davies, an internist and acting director of high reliability systems and consultation at the U.S. Department of Veterans Affairs.
Manipulations of waiting times are not just hypothetical. We had a post back last summer about a hospital in the UK that got busted for not taking patients off the waiting list in a first-come, first-served fashion. They were being evaluated on what fraction of patient received treatment in less than a targeted timed. They resorted to letting some patients take one for the team when they had slipped past the target.
Note that airlines in the US face a similar incentive. The FAA reports what fraction of time a flight is delayed. When a mechanical problem forces an aircraft out of service, should the airline pull a plane off another trip? There is some notion of fairness in doing that. However, it would make a second flight late.
In any event, as we noted in our earlier post there is a way of addressing this: Use two measures. For example, one could evaluate performance both on what fraction of patients get treated under some targeted time but then also add a measure on what the average waiting time beyond the target is for those who are not served in a timely fashion. If one is doing well on the former, relatively few patients contribute to the latter. Making any of them wait for an extended period hammers the second performance measure.