Exploration of how the distributions in MIC vary by key population groupings (such as age or infection type)

Expression of Interest Title: Data Challenge – Exploration of how the distributions in MIC vary by key population groupings (such as age or infection type)
Team 9056, led by Jacob Wildfire of London School of Hygiene and Tropical Medicine, UK with Gwen Knight (LSHTM), Naomi Waterlow (LSHTM), Naomi Fuller (LSHTM) and Alastair Clements (LSHTM)

Summary of Research:

Our research will exploit the wealth of data available in this challenge on minimum inhibitory concentration (MIC) values to explore how the distributions in MIC vary by key population groupings (such as age or infection type). We will develop a research methodology that quantifies and characterizes these distributions across user-defined groupings and over time to gain insights into different infecting strains and selection pressures. We hypothesize that variation in MIC distribution across different population groupings can provide insight into commonalities or variation in transmission, infection and selective pressure by subpopulations. This will support improved public health practice and better targeting of antibiotic stewardship programmes. As a test case, we will explore how resistance prevalence varies by patient age and sex. Antibiotic exposure, healthcare contact and immune system function all vary by age and sex. While preliminary research suggests that there is substantial variation in the trends in antibiotic resistance by demographic factors, it has not been quantified globally. We will develop a stepwise analysis that will firstly test for resistance prevalence differences by age and sex across bacteria and antibiotics. Secondly, we will compare MIC distributions with the hypothesis that if MIC distributions within a bacteria for a certain antibiotic vary by age or sex, then the bacteria causing the infections or the resistance mechanisms vary by age suggesting different transmission sources or antibiotic effects. Thirdly, we will explore MIC creep within subpopulations over the years of data provided. We will highlight our findings in a limited number of bacteria-antibiotic combinations (likely within S. aureus and E. coli). Finally, we will generate a generalized framework, with open-source code that would allow users to explore MIC distributions across their own desired characteristics (such as infection type). Such analysis would enable more detailed analysis of the most commonly available phenotypic AMR data.

Requested Datasets:

Data Contributor: Pfizer Inc.

Bedaquiline Drug Resistance Assessment in MDR-TB (DREAM)
Data Contributor: Johnson & Johnson

GEARS (Global Evaluation of Antimicrobial Resistance via Surveillance)
Data Contributor: Venatorx Pharmaceuticals

Data Contributor: Paratek Pharmaceuticals, Inc.

Data Contributor: Shionogi

SOAR 201818
Data Contributor: GSK