Last November, Wellcome launched the “Wellcome Data Re-use Prize for AMR surveillance” as a challenge for teams around the world to “generate a new insight, tool or health application” from data held in the AMR Research Initiative - a partnership between the Open Data Institute and the Wellcome Trust. We are pleased to announce that a team from LSHTM won the first prize, which includes a financial award of £15,000. The team was invited to attend the ECCMID conference in Amsterdam last weekend where the winners were formally announced. This “Antibiotic Resistance: Interdisciplinary Action” (AR:IA) team is comprised of Gwen Knight, Francesc Coll, Quentin Leclerc, Nichola Naylorand Alex Aiken.
 
As surveillance of antibiotic resistance improves, vast amounts of data are being collected on the prevalence of antibiotic resistance worldwide (e.g. WHO GLASS and ECDC data). A new kid on the block is the ATLAS dataset, released as part of the prize into the AMR Research Initiative. This dataset, collected by Pfizer, contains high-quality antibiotic susceptibility data, including ‘raw’ minimum inhibitory concentration (MIC) data, for over 600,000 bacterial clinical isolates collected from 77 countries and spanning 14 years.
 
The AR:IA team focused on integrating these large surveillance datasets into something that could potentially be more clinically relevant than just prevalence of resistance to drug X in bug Y in setting C. To do this the team decided to address the problem of empiric therapy. Unfortunately, worldwide, the vast majority of antibiotics are still prescribed empirically, which means they are given before the infecting organism and any resistances it may have, are known. Guidelines on what antibiotics to prescribe empirically for what infection, should be informed by resistance data but often they are not.
 
AR:IA took open access, global data on levels of antibiotic resistance and converted them into empiric therapy recommendations for common infection syndromes. This required knowledge of what bugs cause an infection syndrome (such as sepsis), their level of resistance (by country) and the likely therapies available to a country as well as their cost. This information was combined into a syndrome-level composite index, with maps to show where recommendations would be for first, second or even higher-level therapies in different countries. Unfortunately, this work highlighted the bias of using convenience sampling: many of the isolates were highly resistant (e.g. most S. aureus were MRSA), and despite the high number of isolates, once split down by syndrome, recommendations could only really be made for the syndromes sepsis, pneumonia and cellulitis/skin abscess. By pulling together such a varied set of data, the team hopes to have generated a reproducible framework that can be used to inform empiric guideline design in the future.
 
For more information, you can check out the report here or play with the online, user-friendly app. that AR:IA developed and watch this space for what they do with the £15,000 prize money!