4.2 Interpreting estimates

When you read and review reports or publications on the proportion and frequency of AMR infections, it is important to make sure that you understand how the numerators and denominators are estimated. When interpreting the estimates you should ask yourself the following key questions:

• Are the estimates biased towards patients with specific characteristics, such as patients who failed initial empiric antibiotic treatments?
• Are the denominators appropriately selected?
• Do the estimates represent the target population?

If you are collecting data and/or calculating and reporting proportion and frequency of AMR infections, it is useful to be aware of the tools and protocols available to support you when you are processing data and generating reports. Some examples that you might like to explore include the following:

• WHONET [Tip: hold Ctrl and click a link to open it in a new tab. (Hide tip)] , an application for management and analysis of microbiology laboratory data.
• SEDRI- LIMS (Global Laboratory eTools, n.d.), an open source laboratory information management system specialising in microbiology.
• AMASS, an open access offline application to automatically generate AMR surveillance reports using microbiology data and, if available, hospital admission data.
• MICRO (Turner et al., 2019), a checklist to guide the reporting and interpretation of clinical microbiology data.
• AMR (for R), a package within R software for AMR data analysis.

It is also useful to be aware of data-generation and reporting systems in local, national and regional AMR networks and protocols.

Whenever possible you should consider reporting:

• the numerators and denominators of your estimates
• the number and proportion of missing antimicrobial susceptibility test results for the antibiotics under analysis among the cases with the pathogen of interest identified
• stratified data or estimates by community-origin, healthcare-associated and hospital-origin, because each of these has distinct population characteristics, transmission dynamics and burden of AMR, hence would require different strategies to control spread of AMR
• stratified data or estimates by sex and age groups – you can find out more about the importance of disaggregated data in the Gender and equity in AMR surveillance course
• metadata that could reflect the data sources (i.e. number and type of healthcare facilities that serve the target population from which data was retrieved), catchment population size, microbiology testing service utilisation rate (i.e. number of culture sets sent for microbiology testing over the reporting period), total patient days over the reporting period, total number of discharges over the reporting period, etc.