• US wait times and mortality

    Aaron’s recent posts on waiting times in the US and Canada (and Austin’s prodding) have convinced me to add a little more data to the conversation.  Julia Prentice and I did a study of the effect of waiting times on mortality among elderly American veterans who use VA outpatient geriatric clinics.  VA medical centers offer low-cost or free medical care to qualifying veterans, so demand for care typically exceeds the available supply by a larger margin than in most of the US system.  We studied these particular patients because the VA has electronic waiting times data and because these patients had numerous health conditions that we thought would make them vulnerable to bad outcomes if they had to wait for care.

    The key figure, shown below, illustrates that risk-adjusted mortality rates increased with waiting times, but the relationship was not as strong as some might expect.  The risk of mortality increased about 20% starting at an average wait time of 31 days (the increases at the far right and far left of the figure were not significant because of sparse data).

    I should emphasize that the waiting times in this study were for all outpatient encounters with physicians, not just elective surgery, which is often the subject of comparisons between the US and Canada.  Still, there’s evidence that waiting for care is bad for you, particularly if your health is fragile and the waits for appointments are in excess of 30 days.


    • Steve
      Probably helpful to add that as a VA population, it skews male, older, more potentially unmeasured SES variables, etc., and wait times may have a greater magnified effect.

      Also possible that their presentation is delayed in the first place (“putting off care”), and the prolongation amplifies a preordained bad outcome (vs a matched control population).

      Not that data is invalid, it just needs context.

    • One should also note that 100% of their patients will smoke while waiting to see the doctor, or at least it always seemed that way.


    • I notice that differences in the “case mix” between facilities are controlled in the study by including previous mortality rates for the facility as an independent variable. If there are long-term wait-time problems in a given facility, from your results, we should expect facilities with higher wait-times to also have higher historical mortality rates. Because of this, it seems to me that the true relationship between wait-times and mortality would be stronger than your findings indicate.

      Since you had direct data on the type of clinic appointments at each facility, why didn’t you use that data to control for differences in the case mix between facilities? Why do you consider historical mortality a better proxy?

    • Scott,
      You ask a good question. When we examine the variability in waiting times it seems to us that there’s a lot of variance from month-to-month, so that helps reduce the concern about long-run waiting time effects on past facility mortality rates. Nevertheless, what you say is correct, controlling for facility effects eliminates some of the variation in wait times that could be associated with the outcome. In that way, this is a conservative estimate, but we hope it is less vulnerable to confounding by unobserved facility quality.
      You also ask why we didn’t use the mix of appointment types as a casemix control. The answer here is that we already use patient-level diagnosis codes and different facilities use clinic stop codes differently, so we didn’t think it was worth it. Good thinking though.