One of the things that we often like to say when giving data is “All things being equal…” Of course, we know that all things are not equal. There are differences between people and populations. A city like New York may be close to Philadelphia, but the two have differences that make them unique. From the racial make-up of a city to its geographic location, we need to be very careful when presenting data to make sure that we are not deceiving the public in general and those who need to know in particular.
For example, look at the current situation with Legionnaires’ disease in the Bronx, New York:
“City health officials are investigating a spike in Legionnaires’ disease in The Bronx, The Post has learned. “Providers should consider Legionnaires’ disease when evaluating patients . . . with signs of pneumonia,’’ the Health Department said in an alert to medical providers. Last month, 11 cases of the disease were reported in The Bronx, the city’s poorest borough, compared to two or three in a typical December. The Bronx recorded 61 cases of the potentially deadly disease in all of 2014, substantially exceeding the citywide average. The disease struck 4.3 out of every 100,000 Bronx residents last year, compared to 2.5 per 100,000 citywide, according to the alert. The rate in the northeast Bronx was even higher — 9.4 per 100,000 residents.”
What is it about The Bronx that makes it different than the rest of the city?
Now look at the number of yearly hospitalizations associated with Legionnaires’ disease, a form of pneumonia, in the United States as a whole. According to the Centers for Disease Control and Prevention (CDC), there are between 8,000 and 18,000 hospitalizations from LD each year in the United States. Compare this to the number of reported cases of LD in 2012: 3,688. How can it be that there are more hospitalizations than reported cases?
Conspiracy nuts will tell you that CDC is lying, or that CDC is being lied to. Most of the time they say this is to discredit data on things like vaccine adverse events or autism, or some other pet project of theirs. What is happening here is that we are looking at two sources of data — one for hospitalizations and one for reported cases — and presenting them with little explanation. In fact, you don’t really see how the hospitalization number is estimated except for a disclaimer:
“Each year, between 8,000 and 18,000 people are hospitalized with Legionnaires’ disease in the U.S. However, many infections are not diagnosed or reported, so this number may be higher. More illness is usually found in the summer and early fall, but it can happen any time of year.”
Yes, the number is more likely higher because of several reasons. First, not all cases of LD are diagnosed. This is because people who present to a healthcare provider with community-acquired pneumonia are quickly put on antibiotics. Most of the time, they will respond to antibiotics since Legionella bacteria are not known to be very resistant to common antibiotics. These people won’t be tested. Without a test, we’ll never know if they really were cases of LD. Second, there are many times when people who test positive for LD are not reported. This happens when the lab doing the test leaves the reporting to the healthcare provider at the same time that the healthcare provider leaves the reporting to the lab. I’ve seen it happen way too many times. Third, not everyone who tests positive fits the case definition of LD, so they don’t count as true cases.
Still, how did CDC come up with the 8,000 to 18,000 number for hospitalizations? Believe it or not, it comes from a 1997 study that looked at 1991 data from two counties in Ohio:
“Extrapolation from study incidence data showed the projected annual number of cases of community-acquired pneumonia requiring hospitalization in the United States to be 485,000. These data provide previously unavailable estimates of the annual number of cases that are due to Legionella species (8000-18,000), Mycoplasma pneumoniae (18,700-108,000), and Chlamydia pneumoniae (5890-49,700).”
Yes, here we are in 2015 and we’re using a hospitalization estimate from 1997. That estimate was done previous to the diagnostic technology that we have today. It was done before the current case definition of LD, which was adopted in 2005. Long story short, we need to change it.
We need to change it because we need to equip healthcare providers and public health decision-makers with a better idea of the burden of LD. Healthcare providers need to be convinced to test each and every single case of pneumonia so that we get a better idea of the burden of LD. Short of that, we need to come up with a nationally representative sample of people with pneumonia and test them. We also need policymakers to know the burden of LD — along with the risk factors — so we can mitigate some of those risks and inform the public about their risk of LD. (Smokers are at great risk, and so are people with compromised immune systems.)
We need to change it because two counties in Ohio in 1991 are not the same as The Bronx in 2015. Diseases really are more about who you are and where you live than the bug itself.
And that, dear readers, is where your favorite epidemiologist steps in. But more on that later… As my thesis proposal comes together.