Analysis and Presentation of Cumulative Antimicrobial Susceptibility Test Data--The Influence of Different Parameters in a Routine Clinical Microbiology Laboratory

Abstract

Introduction: Many clinical microbiology laboratories report on cumulative antimicrobial susceptibility testing (cAST) data on a regular basis. Criteria for generation of cAST reports, however, are often obscure and inconsistent. Whereas the CLSI has published a guideline for analysis and presentation of cAST data, national guidelines directed at clinical microbiology laboratories are not available in Europe. Thus, we sought to describe the influence of different parameters in the process of cAST data analysis in the setting of a German routine clinical microbiology laboratory during 2 consecutive years.

Material and methods: We developed various program scripts to assess the consequences ensuing from different algorithms for calculation of cumulative antibiograms from the data collected in our clinical microbiology laboratory in 2013 and 2014.

Results: One of the most pronounced effects was caused by exclusion of screening cultures for multi-drug resistant organisms which decreased the MRSA rate in some cases to one third. Dependent on the handling of duplicate isolates, i.e. isolates of the same species recovered from successive cultures on the same patient during the time period analyzed, we recorded differences in resistance rates of up to 5 percentage points for S. aureus, E. coli and K. pneumoniae and up to 10 percentage points for P. aeruginosa. Stratification by site of care and specimen type, testing of antimicrobials selectively on resistant isolates, change of interpretation rules and analysis at genus level instead of species level resulted in further changes of calculated antimicrobial resistance rates.

Conclusion: The choice of parameters for cAST data analysis may have a substantial influence on calculated antimicrobial resistance rates. Consequently, comparability of cAST reports from different clinical microbiology laboratories may be limited. We suggest that laboratories communicate the strategy used for cAST data analysis as long as national guidelines for standardized cAST data analysis and reporting do not exist in Europe.

Fig 1. Resistance estimates dependent on the handling of screening isolates.

Resistance rates were calculated either with inclusion (black columns) or exclusion (grey columns) of screening isolates, as described in the text. Further details are given in the supporting information (S1 Table).

Fig 2. Resistance estimates dependent on the method of duplicate isolate removal.

Resistance rates were calculated using different methods of duplicate isolate removal, as described in the text. Further details are given in the supporting information (S2 Table).

Fig 3. Resistance estimates dependent on the time-point of isolate recovery.

Resistance rates were calculated with respect to early isolates (black columns) or late isolates (grey columns), as described in the text. Further details are given in the supporting information (S3 Table).

Fig 4. Resistance estimates dependent on the patient location.

Resistance rates were calculated with data stratification according to the patient location (hospital and ward), as described in the text. Further details are given in the supporting information (S4 Table).

Fig 5. Resistance estimates dependent on the specimen type.

Resistance rates were calculated with data stratification according to the specimen type (black columns: isolates recovered from blood cultures, grey columns: isolates recovered from other body sites), as described in the text. Further details are given in the supporting information (S5 Table).

Fig 6. Resistance estimates dependent on organism’s resistance characteristics.

Resistance rates were calculated with data stratification according to cefotaxime-resistance (all isolates: dark grey columns, only resistant strains: black columns, only susceptible strains: light grey columns), as described in the text. Further details are given in the supporting information (S6 Table).