Data Reporting Doubts About The Antibiotic Resistance Atlas

Last week, I provided some links to the CDC’s Antibiotic Resistance Patient Safety Atlas (there’s an interactive version here).

First, the CDC should be given kudos for putting this resource together. This isn’t a criticism of their effort. But the tl;dr version is that once you start going through the data you realize just how poor and inconsistent our infectious disease reporting is. Before I discuss some of the red flags, there’s a general point about how hard this is.

If we want to calculate the percentage of antibiotic infections, we need to get good estimates of the numerator (the number of antibiotic resistant infections) and the denominator (the total number of resistant and sensitive infections). I apologize for the remedial arithmetic, but we need to realize that capturing these numbers isn’t trivial. For certain kinds of infections, many never even make it into the reporting system, such as food-borne infections or low-grade UTIs. For hospital-acquired infections (HAIs), reporting requirements and reliability can vary widely from locality to locality*. It’s even harder to capture ‘unsexy’ infections that clear up rapidly, and which might never have even made it to the clinical lab.

That said, onto the data. The Atlas reports three types of hospital-acquired infections (HAIs): Catheter Associated Urinary Tract Infection (CAUTI), Central Line Associated Bloodstream Infection (CLABSI), and Surgical Site Infection (SSI). These infections typically do have higher resistance rates than minor infections in healthy people.

Some of the data seem to pass the smell test. Nationally, methicillin-resistant Staphylococcus aureus infections (‘MRSA’), according to the Atlas, account for 46 percent of all HAI S. aureus infections, ranging from the low thirties to the low sixties depending on the state. Seems right.

But then we get to carbapenem-resistant enterobacteriaceae (‘CRE’) and everything gets weird (carbapenem-resistant infections typically are untreatable with everything we have except for colistin–and resistance to colistin is rising. Have a nice day).

If we look at Klebsiella CREs, the average frequency is 8.4%–which is to say, that 8 out of 100 procedure-related Klebsiella infections are CRE. That seems very high. New Jersey has a Klebsiella CRE frequency of 22 percent, while New York is at sixteen percent. Keep in mind, these are supposedly state-wide estimates, so some facilities will be worse than this.

This strains credulity, to say the least. With these kinds of rates, hospitals–or at least wards–should be shutdown on a routine basis. Meanwhile, if we look at neighboring Connecticut, the Klebsiella CRE frequency is 2.2 percent, one tenth that of New Jersey’s. Massachusetts (adjacent to New York) has the same frequency as CT. Similar comparisons between D.C. and Maryland or Virginia yield similar disparities. Do we really think these massive differences reflect the spread of Klebsiella CRE, patient composition, or infection control issues? It’s possible, but there’s a more parsimonious explanation–data issues.

When we look at the E. coli CRE, things are much more where we expect. Nationally, the rate is 0.7%, and New Jersey has a two percent rate. While Klebsiella CRE are more frequent than E. coli, the Klebsiella numbers suggest to me that we have a very long way to go in getting accurate, standardized reporting that is consistent from state-to-state.

Maybe instead of ‘disrupting’ healthcare**, some smart computational types could figure out how to accurately capture hospital lab data. At the same time, we need unified, national reporting standards and requirements. Otherwise, despite the CDC’s best efforts, we won’t really have a good handle on even the most basic contours of the antibiotic resistance problem–especially its most critical components.

*We don’t have a public health system; we have many, many health systems.

**I would argue our healthcare is disrupted enough, thank you. Don’t like that word around healthcare at all, though ‘improving’ isn’t a sexy ‘entrepreneur’ word.

This entry was posted in Antibiotics, Public Health. Bookmark the permalink.

1 Response to Data Reporting Doubts About The Antibiotic Resistance Atlas

  1. Pingback: Data Assistant Vacancy at Aids Healthcare Foundation (AHF) - Latest Nigeria Jobs & Vacancies

Comments are closed.