Herd Immunity: The fight against superbugs starts with a doctor’s pen

We keep hearing that we are heading toward an immunity apocalypse, where all of our best antimicrobial drugs fail to defend us against drug-resistant bacteria, viruses, and other microbes that can sicken, maim, and basically devour us in a matter of hours.

And yet we know very little about the true extent of this trend, which hampers our ability to craft strong public and private policy responses. This ignorance also hampers our ability as individual patients to make the best choices about our own treatment regimens, and as consumers deciding whether to buy household products that are loaded up with antimicrobials, from bathroom caulk to toy ducks. 

In the Herd Immunity series, I have highlighted the difficulties of mapping the spread of superbugs in hospitals across the country. The most obvious problem is the lack of facility-specific tracking and mapping of outbreaks of things like MRSA and C. difficile in health facilities.

But mapping presumes the people who have been diagnosed with these infections have been accurately diagnosed. A certain percentage of these people end up dying, and so the mapping also presumes that their cause of death is accurately reported. Diagnoses from health facilities, diagnostic coding in insurance claims, and causes of death in vital registration systems are bedrock datasets for many of the health trends that you see being reported by local, state, and federal agencies. 

In a recent Reuters series, a team of reporters exposed the knowns and unknowns about superbugs and highlighted a huge hole in our knowledge: the inaccuracy of death certificates.

Ryan McNeill, Deborah J. Nelson, and Yasmeen Abutaleb named their must-read series “The Uncounted.” They reviewed death certificates, hospital records, state reporting requirements, and other documents and found things that should make people more than a little nervous:

1. Even when hospitals accurately diagnose a drug-resistant infection that leads to a death, physicians are leaving that diagnosis off the death certificates, making it impossible to accurately track the deaths.

2. The number of deaths that are recorded on death certificates is substantial. “A Reuters analysis of death certificates found that nationwide, drug-resistant infections were mentioned as contributing to or causing the deaths of more than 180,000 people” from 2003 to 2014.

3. If you take into account that about 29.9 million people died in the U.S. during those years, that amounts to about 0.6 percent of all deaths. That percentage likely would go up were the true toll being reported. By the reporters’ estimates, “tens of thousands of deaths from drug-resistant infections – as well as many more infections that sicken but don’t kill people – go uncounted because federal and state agencies are doing a poor job of tracking them.”

4. Most people live in places where their state agencies are ignoring the drug-resistant microbe problem. “Twenty-four states and the District of Columbia – an area comprising 3 of every 5 Americans – said they do not regularly track deaths due to antibiotic-resistant infections. In contrast, all 50 states require reporting of AIDS-related deaths.”

One of the things I most admire about this series is how fearless the reporters were in partnering with experts to do their work while still providing an unvarnished account of the problem. The team worked directly with the CDC’s National Center for Health Statistics’ Division of Vital Statistics to comb through the center’s database of death records and find deaths that were likely caused by drug-resistant microbes.

And yet, the reporters wrote an entire piece calling into question the CDC estimates that are regularly cited as the best count of drug-resistant microbial infections and deaths.

Reuters took a close look at how the agency arrived at its numbers and made a surprising discovery: They are based on so little hard data that they could be off by more than 30 percent – more than 10,000 people – in either direction.

The one quibble I would make with that statement is that the accuracy of an estimate is not tied merely to the amount of data underlying it, as strange as that may seem. It’s also based on the statistical modeling strategy being used. For example, you may have literally zero data for a town in Indiana but a few data points for a neighboring town. With the right modeling strategy, you can make accurate estimates for the neighboring town that would get you closer to the truth than 30 percent one way or the other.

In all, though, the reporters performed a tremendous public service by carefully documenting the gaps in our knowledge about this significant public health problem.

**