More Denominators
In the last issue of PSQH, I wrote about different takes on the concept of numerators and denominators. Kerry O’Connell leaves the comfortable world of uneventful patients (denominators) when he experiences preventable harm and joins the numerators. National Coordinator for Health IT Farzad Mostashari, on the other hand, thinks of denominators as patients whose individual characteristics—including chronic and underlying conditions in need of treatment—may be invisible until brought to light through electronic records and analytics.
Another use of numerators/denominators has come to my attention. In “The Big Denominator,” his Editor’s Note to the spring 2002 issue of the Journal of Clinical Engineering, William A. Hyman questions taking comfort in “big denominator logic,” which argues that some number of adverse events (numerators) may be acceptable if that number is small enough compared to a big denominator. Hyman points out that this is cold comfort for the numerators:
Most device-related patient injuries, whether technical failures or use error, are relatively rare events as measured by the number of incidents divided by the total number of times the device was used, or the total number of patients on which it was used.
Thus the large denominator of total exposures almost always results in a low incident rate, although perhaps not as low as that reported by other industries such as aviation, or the facts leading up to the Explorer-Firestone tire recalls. One recent medical example of big-denominator logic is the following quote from Baxter following a jury verdict that Luer locks should be used rather than friction connections: “These types of products using the friction-fit connection have been provided by Baxter and other companies for more than 40 years and have been used with hundreds of millions of patients without problem.”
A second example is the recent report that few hospital bed injuries stem from entrapment, compared to fall injuries, although in this case the comparison is at least among injuries as opposed to all bed use. While the big denominator may give some sense of relative importance, it is of little consolation to the patients constituting the numerator, i.e., those that have been injured. Moreover, when the big denominator argument is used as part of a defense (e.g., we’ve sold millions of these devices, and only a few people have been killed), there is perhaps the implication that the small number of adverse events is “acceptable,” although acceptable to who remains a question.
Personalizing the issue, consider the stop sign-controlled intersection near your home that you pass at least twice a day, having done so for many years. Perhaps you have safely negotiated this intersection 3,000 times until the day you are late, and worried about your leaking roof, and the meeting you’re missing, and you fail to stop at that stop sign and strike another vehicle. Is your responsibility reduced because your demonstrated incidence rate is one in 3,000? Will the owner of the other vehicle say it’s OK because you have properly stopped many times in the past? Even “no-fault” insurance does not negate responsibility; it merely simplifies compensation under limited circumstances.
While the consequences of failure or error are singular events, prevention is a multi-faceted problem. Given the reality of limited resources, and predictions of cost-effectiveness of preventing a particular hazard, prioritization often does have to consider incidence rates among other factors. But even a seemingly rational decision that a proven hazard has not risen to the level of a specific correction plan does not relieve—and perhaps enhances—the responsibility for the consequences of that hazard manifesting itself.