AMR deaths in Europe

“In case of an emergency check your own pulse first”, that’s one of the rules of the House of God. More than 33.000 deaths due to AMR in Europe per year, as reported yesterday, definitely is an emergency. Therefore, I tried to disentangle what that means for my small country that so vividly tried to keep these superbugs out of the country.

The estimates on the number of deaths in Europe were derived with a method, developed in the EU-funded Burden of Communicable Diseases in Europe (BCoDE) project, that has been used for other infections. It transparently provides estimates not only of deaths, but also on Disability-Adjusted Life Years (DALY’s) and Years Lived with Disabilities (YLD). All estimates are presented with confidence intervals, so in this respect the methods are much more scientific than, for instance, the widely cited AMR report from Jim O’Neill.

Yet, as with all previous attempts the model depends on the input: the estimates of invasive infection per country, extrapolated from that we get the estimated number of non-invasive infections and then the estimates of attributable mortality per resistant bug and infection site. This is the Achilles heel of such studies, as these extrapolations start and continue with incomplete data. One of the most important unknowns probably is (although I didn’t check) the “attributable mortality factor”, which is one of the most difficult things to estimate. Only few studies have done a reasonable job in including relevant confounders that influence the association between an infection with a resistant bug and patient outcome. A very important factor in this association probably is the local situation in which patients are treated: a setting where adaptation of inappropriate empiric therapy occurs after 24 hours versus 96 hours (or even longer), where source control is achieved early rather than late, etc.. Unfortunately, we don’t have country-specific estimates of attributable infection, so the investigators had to work with “global” figures. For clarity: this could  mean that this factor was over- and underestimated.

Now, to my lovely small country that has built an imaginary wall against invading superbugs (and paid for it). According to the study each year 206 people die because of AMR in the Netherlands. If you want to know where this comes from you have to dive deep into the supplement and don’t forget your microscope for reading!

I found that each year 24 patients die from infections caused by colistin and/or carbapenem resistant E. coli (out of 277 being infected). Strange, as to the best of my knowledge we don’t have identified such bacteria in infected patients so far. We had a nursing-home outbreak, but without infections.  Another 107 patients apparently die from E. coli being resistant to 3rd generation-cephalosoprins (and 3.500 being infected). Although we do regularly see such infections (not surprisingly as 5% of the population has detectable carriage), our estimates of attributable mortality due to these infections is close to zero (presented at ECCMID last year, paper being finalized). Probably because we either start empiric therapy with a carbapenem (yes we do, with increasing frequency) or we adjust to it once resistance to cephalosporins has been demonstrated (usually within 48 hours). With that, two deaths per week due to these infections seems high. The same holds for the 30 deaths (and 471 infections) due to colistin and/or carbapenem or 3rd generation-cephalosoprin resistant K. pneumoniae. In the Gram-positive area our numbers are 63 VRE infections with 4 deaths; 249 MRSA infections and 12 deaths; 60 penicillin-resistant pneumococci and 3 deaths. A recent estimate based on our national database yielded 12 and 71 patients with VRE or MRSA bacteremia per year. The 23 deaths attributable to resistance in P. aeruginosa comes – to me – closest to our situation, although we lack good studies in which the role of confounders was meticulously incorporated in estimating attributable mortality for these bugs. My own experience is that – sometimes – we run out of options in patients with ultimately fatal underlying disease conditions, such as haematological malignancies or lung transplantation. The pan-resistance in P. aeruginosa usually results from the successive selection of resistance in endogenous strains, rather than cross-transmission. Whether such patients succumb due to antibiotic resistance or untreatable underlying disease is then a matter of debate.

To summarize: there is an emergency, but our pulse is still regular. The study definitely adds to our interpretation of the resistance problem in Europe and highlights the enormous (and for many I assume) unacceptable risk difference between countries. Don’t focus too much on the absolute numbers as they may not be very precise. We (and the investigators) would benefit from estimates for attributable mortality due to antibiotic-bug combinations derived with the same methodology in multiple countries, and allowing estimates per country (thus a large study). For the Netherlands, we are very comfortable in the lower part of Figure 3.

One thought on “AMR deaths in Europe

  1. Dear Marc, interesting points! Thanks for acknowledging the scientific approach and transparency. Large population studies carry some limitations, which are acknowledged in the article. As mentioned, one crucial limitation is attributable mortality, for which country-specific data are not available. In the absence of country-specific data on attributable mortality that would reflect different treatment approaches, this study used the best available evidence, i.e. systematic reviews and critical extrapolation of results from studies.
    Regarding the results for the Netherlands (stated EARS-Net country population coverage of 65%), the study estimates 16 cases of bacteraemias due to VRE (similar to your estimation of 12 cases) and 43 cases of bacteraemias due to MRSA (less than your estimation of 71 cases). Moreover, the Netherlands reported 22 cases of colistin-resistant E. coli bacteraemias to EARS-Net for 2015. All these cases were reported as being susceptible to carbapenems. Laboratories in most EU/EEA countries have difficulties with colistin susceptibility testing and this was certainly true back in 2015. For future studies, we are considering omitting the category “colistin-resistant” altogether when translating surveillance data into burden estimates.
    If you have any further questions about this study, you are welcome to send them to
    Alessandro Cassini for the Antimicrobial Resistance and Healthcare-associated Infections Disease Programme at ECDC


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s