Medicine is hard because patients are complex

A decade ago, I heard a cardiologist ask, "A third of our patients who have heart attacks aren't put on aspirin. Now, how often does McDonald's serve you a hamburger without a pickle? Why are they so much better at what they do than we are?"

Survival of a heart attack is greatly enhanced if a patient takes aspirin within 24 hours of the attack. This is a simple and safe intervention, and the cardiologist was astonished that emergency departments failed to carry it out routinely. (I think that EDs do this better now.) The force of the cardiologist's complaint, though, still holds: given that lives are at stake, why is the quality of health care so much lower than food service, or automobile manufacturing.

I fully endorse the goal of standardization of routine health care. But there is an obvious answer to the question: It is hard to standardize care because patients are more complex than hamburgers. Here are some data that illustrate the point.

For one of my research projects, I have data on 44,604 neonatal intensive care unit (NICU) patient stays, in US NICUs over the past three years. Many of these infants were born prematurely, so this is a particularly complex group of patients. But it is interesting to see how complex they are. The data include the diagnoses (possibly not all of them) they received during their stay. Here is a histogram of the number of diagnoses each patient received. NICU patients received an average of 10.4 diagnoses, and you can see that there is a tail of patients to the right with 30 or more. 

But what really surprised me was the distribution of the number of patients who had each diagnosis. Overall, the 44,604 patients had 2773 different diagnoses. Now, 2773 overstates the complexity to some degree, because many of the diagnoses are variants of basic classes of disorders (for example, there are codes for several levels of severity of bleeding in the brain). Nevertheless, physicians think that these distinctions are worth making, so it could well be that the distinctions matter for decisions they have to make about treatment. If you look at the second Figure, you will see that there are a very few diagnoses that are common among NICU patients (that is, you would find them in a third to a half of patients). These are mostly problems with the functioning of lungs, which are incompletely developed in very premature infants. However, half of the diagnoses were found in 5 or fewer patients (that is, about 0.01% of patients seen), indicating that the typical NICU patient will often have one or more relatively uncommon problems.

So, it is really important to see that well-proven standards -- like aspirin following a heart attack -- are carried out routinely. But developing these standards is much harder than you think, because patients have complex constellations of problems.

 

Comments

[how to]

The sooner a person seeks medical attention for symptoms of an impending heart attack, the more likely it is that the person will survive. Do not try to tough it out, or wait and hope that the discomfort that you are feeling will go away. There is never any harm in seeing the doctor, and having her tell you that you are fine. However, if you don’t see a physician, there is a fairly good chance that the heart attack will kill you. Don’t risk it.

[KPajer]

One of the issues that contributes to the complexity of patients is the instability of clinical presentation. Diagnoses are made cross-sectionally, but disease and disorder often change over time. The customer who wants a pickle one week usually has the same presnetation the next week.

[Bill Gardner]

Very good point, Dr. Pajer.