The art of predicting “broad brush estimates” of people dying from AMR

Sometimes you read something you wished you had authored. For me, that is the essay “Will 10 Million People Die a Year due to Antimicrobial Resistance by 2050?” published yesterday in PLoS Medicine. At last, a scientific response to the highly praised AMR review from Jim O’Neill. Summarized in a few words: scrutinized to the bone, hardly anything remains.

In todays’ world AMR is considered one of the major threats of a healthy future. The fire was lit by the reports from ECDC and CDC in 2009 and 2013, and exploded by the O’Neill report earlier this year. Since then, many scientifically intended publications (and even more layman texts) start with the prediction that in 2050 we will see 10.000.000 deaths per year caused by AMR. This message has been adopted as the new reference standard, and astonishingly few have raised questions on the methods leading to these estimates.

In this essay de Kraker, Stewardson and Harbarth address the most important methodological issues. The “science” of the AMR report was provided by 2 consultancy firms (RAND and KPMG), but less clear was how numbers of infections and attributable deaths were calculated. De Kraker nicely explains the major (and for every peer reviewer fatal) flaws.

Yet, the most hilarious part of the methods is the 4 future scenarios: “an absolute rise in resistance levels of 40% for all species under study or 100% resistance, with both of these scenarios combined with either stable or doubled infection rates.” The authors correctly state that “To date, there is no empirical data supporting any of these scenarios.” Very politely they say – I think – that this work would not have passed scientific peer review and I fully agree with their statement that this report “potentially undermines, rather than supports, the fight against a post-antibiotic era.”

For activists this report probably is the best thing since the invention of sliced bread. For those trying to scientifically quantify the consequences of AMR in order to better understand its dynamics and to develop better and cost-effective measures to cope with this global problem, these analyses are a disgrace. It is a sign of our time, in which big words mean more than facts. For years, I have tried (and sometimes succeeded) to convince colleagues and (some) policy makers that AMR (and investigating related aspects) is relevant. If they ask me what I think of this report, I honestly, can only repeat the contents of this essay. And that doesn’t help.

The AMR Review even acknowledged “that the reported numbers are “broad brush estimates,” (for which you obviously can’t quantify confidence areas) that “more detailed and robust work will no doubt be done by academic researchers,” and that there is a lack of data, urging for improvement of infection surveillance.

I am preparing new questions for trivial pursuit.

What happens in 2050?

  1. The first Rhino will be born on the North pole
  2. Geneva will be renamed Geneva sur Mer
  3. Ivanka Trump, the 1st female US president, will be re-elected for the 4th time in a row
  4. 10.000.000 people will die from AMR

4 thoughts on “The art of predicting “broad brush estimates” of people dying from AMR

  1. Not wishing to contradict an expert like yourself and not having read the PLOS Medicine paper, I can only say that we actually don’t really know the true morbidity and mortality of AMR by country, region and continent. In the UK for example numbers of deaths from AMR are quoted as 5,000 pa and for the EU as a whole 25,000 pa. Our charity has a lot of circumstantial evidence (not scientific) that there is considerable under-reporting of AMR infections. We know of circumstances where patients have died of a nosocomial AMR infection but there is no mention of this on their death certificate. How many pneumonia and sepsis deaths arise from AMR infections? Perhaps the data is out there but I haven’t seen it. So if we don’t know the true numbers of deaths now, how good are we at predicting the future? 10 million does sound a lot and hopefully we will find new ABX before then but as O’Neill said the problem of AMR needs to be tackled holistically and globally. You are doing a great job in the Netherlands controlling ABX use but this needs to be rolled out everywhere. What O’Neill has done is put the AMR issue firmly in the public eye and we all owe him a debt of gratitude for that.


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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s