Numbers Matter: Why counting only resistant bacteria ultimately harms our patients

You've seen the numbers. The CDC estimates that 23,000 deaths are caused by antibiotic resistant pathogens annually and as many as 14,000 of these deaths are linked to C. difficile. Every time I look at those numbers, they make me incredibly sad. First, they include all C. difficile deaths and not just those attributed to fluoroquinolone-resistant C. difficile, for example. This tends to incorrectly overweight the importance of C. difficile relative to other pathogens. Second, and you've heard me rant about this before, they only count deaths caused by the small proportion of bacterial pathogens that happen to be resistant, as narrowly-defined. This would tend to diminish the importance of bacterial pathogens compared to other causes of death (e.g. accidents). But I'm getting ahead of myself.

The numbers I'm about to throw at you are very rough estimates. I'm using these estimates to illustrate a point and hope that others will eventually provide more accurate estimates. If you think I need to correct a specific number, let me know in the comments, and I'll do my best to make the change - but I'm not promising. 

Let's take S. aureus as an example. The CDC estimates that 80,000 infections and 11,000 deaths are attributed to MRSA each year. In 2005, they also estimated that MRSA was associated with 18,650 deaths, but I'll be conservative and stick with 11,000. Per this 2013 NHSN report, the proportion of S. aureus that were MRSA ranged from 43.8% (SSI) to 58.7% for CAUTI. I'll use the lower proportion (44%) since this allows for some mortality secondary to more community (less MRSA) infections. However, I suspect most patients that die from S. aureus infection will ultimately be hospitalized. For simplicity, I will also assume that MRSA is twice as lethal as MSSA (AKA penicillin-resistant S. aureus).

Taking the above numbers, if there were 80,000 MRSA infections, we would expect 102,648 MSSA infections. If the mortality rate for MRSA was 13.75% (11,000/80,000) then MSSA's mortality rate would be half of that or 6.875%. So there would be 7057 deaths from MSSA. If you add that to the 11,000 you get 18,057 deaths due to S. aureus. We can quibble about numbers, but I suspect that 7,000 deaths caused by MSSA, passes the so-called giggle test.

Thus, if we used the CDC rankings to fund research and prevention activities, we would rank C. difficile at the top of the report. However, if we used my ranking system, S. aureus (MSSA+MRSA) ranks ahead of CDI. Since most interventions to prevent MRSA deaths would also work against MSSA (vaccines, new antibiotics) shouldn't both types be included in burden of disease estimates? The imbalance gets worse when you look at Gram-negative infections. Do we really only care if grandma dies of the 2-12% of E. coli or Klebsiella that are resistant to carbapenems? Do we really only count the 610 deaths from CRE and 1700 from ESBL? Clearly it would be better if we counted all deaths from E. coli and Klebsiella and projected a future where almost all strains would be ESBL or CRE.  This would allow us to make better decisions regarding current and future research priorities and prevention efforts.

I suspect if the S. aureus mortality estimate jumps from 11,000 to 18,000 when counting MSSA, it's not a stretch to imagine that deaths from bacterial infections would approach 100,000 in the US. If we count attributable mortality appropriately, deaths from bacterial infections would be a top 10 cause of death in the US. Top 10 means more money for research and prevention. Let's get these numbers right - grandma is counting on us.

...at least she didn't die of MRSA

For the past six months our group at Iowa has been collaborating with colleagues at the University of Utah on a project for CDC. We are tasked with measuring the burden of MDRO using systematic reviews to inform economic models that will project MDRO incidence and attributable cost over the next 20 years. We are studying pathogens like C. difficile, VRE, MRSA, ESBL-GNR and CRE. It's a tall order, but we expect that CDC and others will use these estimates to guide funding for research and prevention efforts. Since bacterial pathogen research is so widely underfunded compared to their burden of disease, these new estimates can only help.

But, it has occurred to me that the planned approach of estimating the burden of disease using pathogens typically categorized as MDRO, such as CRE and MRSA will result in serious underestimates of the bacterial pathogen burden and lead to perpetuating the chronic underfunding of our research and prevention efforts. A simple way of demonstrating the impact of neglecting susceptible bacteria is to focus on S. aureus. One widely cited estimate of yearly MRSA mortality burden is 18,650 in-hospital deaths. Ignoring secular trends and community deaths, while assuming 50% of S. aureus infections in the US are MRSA (and thus 50% are MSSA), we might estimate that S. aureus kills 37,300 people annual in the US. This would place S. aureus (one bacteria!) ahead of traffic deaths and rank it as the #11 cause of death in the US (see table below). Imagine if we included "susceptible" (and resistant) bacteria like E coli, Klebsiella and Streptococcus in a total bacterial burden estimate!

For those that will argue that we have effective antibiotics for MSSA, so that it's unimportant, I offer several counterpoints. First, people die of "susceptible" bacteria (18,650 MSSA!) and we don't fund HIV research based on mortality burden for only protease inhibitor resistant strains. Second, even strains we call susceptible are actually resistant to numerous classes of antibiotics - most MSSA is actually PRSA and try treating enterococcus with a cephalosporin. Third, by ignoring susceptible strain burden, we underinvest in strategies that could treat or prevent all infections, not just arbitrarily defined resistant ones. For example, S. aureus vaccines could prevent both MSSA and MRSA infections. If NIH (or CDC or ECDC) uses only MRSA burden to guide funding of S. aureus vaccine research, they would underfund by 50%.

You get my point, but I will leave you with my final reason for recommending inclusion of "susceptible" strains when measuring burden of disease for bacterial pathogens. Imagine if your aunt is very sick with an MSSA prosthetic hip infection in the ICU. And then imagine if the doctor comes to inform your family that she is very sad that your aunt has passed away, but adds... "at least it wasn't MRSA." Does that make you feel any better?

beer image: source