AST insights

Redefining Accuracy: Precision for Improved Sepsis Patient Management

doctor and sepsis patient

Accuracy is about predicting patient outcomes – not correlation with legacy methods. 

Antimicrobial susceptibility testing (AST) should not only be fast and precise but also accurate. However, AST accuracy is hard to define on its own because there is no such thing as an inherently “true MIC” for a specific bacterial strain.

Current standards based on legacy AST methods like broth microdilution (BMD) and disk diffusion have been used for a long time to define AST accuracy. A result equal to the reference is, by definition, accurate.  Conversely, a result that deviates from the reference is, by definition, inaccurate. This is despite the fact that the clinical breakpoints defined using these reference standard methods are not optimal for predicting patient outcomes and guiding treatment. Yet, they are our measuring stick for accuracy – our meter stick in Paris. In reality, this means that all new AST methods are evaluated based on their correlation to the older method. Consequently, they are tied to the performance of that reference method. This presents a significant challenge when evaluating new AST methods, as their resolution and clinical utility may be higher than the existing reference.

Next-generation AST means faster, more precise – and more accurate.

A new next-generation AST system should ideally not only be faster and more precise than our current AST solutions but also have a higher accuracy. Is that even possible when a correlation to current and, by definition, an inferior reference method is required, even expected? A next-generation AST method with increased resolution could offer information about the areas “in between” the traditional 2-fold dilution steps. Offering this new information leads to a more accurate method than the reference standard.

True accuracy goes beyond agreement with legacy methods. It’s about guiding and predicting clinical treatment outcomes. Rather than treating legacy methods as absolute truths, we question their role in driving improvement. By prioritising clinical outcome prediction over legacy agreement, we pave the way for innovation and advancement in AST methods.