# Comparative Accuracy Analysis of the In Vitro Activity of Colistin Against Multidrug-Resistant (MDR) Non-Fermenters Using Broth Microdilution, Disk Elution, and Agar Dilution

## TL;DR

This study compares different methods for testing colistin effectiveness against drug-resistant bacteria, finding that one method is reliable for a specific type of bacteria.

## Contribution

The study evaluates CBDE and CAT as feasible alternatives to BMD for colistin susceptibility testing in MDR non-fermenters.

## Key findings

- CBDE showed 90.5% categorical agreement with BMD for colistin susceptibility testing.
- Pseudomonas species had better diagnostic performance with both CBDE and CAT compared to Acinetobacter.
- CBDE demonstrated substantial agreement (kappa 0.79) with BMD and is a viable alternative for Pseudomonas aeruginosa.

## Abstract

Objectives: Colistin is a crucial treatment approach for multidrug-resistant (MDR) gram-negative bacterial infections but its susceptibility testing is challenging for clinical laboratories. Therefore, the study was planned to evaluate the diagnostic accuracy of the colistin broth disk elution (CBDE) method and the colistin agar test (CAT) in determining colistin minimum inhibitory concentrations (MICs) among MDR non-fermenting gram-negative bacilli (NF-GNB). The performance of these methods is compared to the gold standard broth microdilution (BMD) technique, with the goal of identifying reliable, resource-feasible alternatives for routine clinical laboratory use.

Material and methods: A total of 74 MDR non-fermenter isolates (37 Pseudomonas aeruginosa, 24 Acinetobacter baumannii complex, eight Stenotrophomonas maltophilia, three Elizabethkingia meningoseptica, and two Chryseobacterium indologenes) were tested by BMD in polystyrene microtiter plates. CBDE and CAT were performed and deciphered as per Clinical and Laboratory Standards Institute (CLSI) interpretive criteria (M100; 32nd Edition).

Statistical analysis: The diagnostic performance characteristics of CBDE and CAT were assessed by comparing them to the gold standard BMD. Agreement statistics, including categorical agreement (CA), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), were calculated. Kappa statistics with 95% CI were used to determine the level of agreement.

Results: CBDE and CAT showed CAs of 90.5% and 85.1%, respectively, with respect to BMD. The overall sensitivity, specificity, PPV, and NPV for CBDE, with reference to BMD, were 88.8%, 91.4%, 85.7%, and 93.4%, respectively, while for CAT, these were 88.8%, 82.9%, 75%, and 92.8%, respectively. Agreement statistics (kappa) for CBDE was 0.79 (95% CI: 0.56 to 1), and for CAT, it was 0.69 (95% CI: 0.46 to 0.97), both indicating substantial agreement. Species-wise stratified analyses for Pseudomonas and Acinetobacter showed that the diagnostic properties and kappa values were better for Pseudomonas species with both CBDE and CAT.

Conclusion: Various susceptibility testing methods for colistin give diverse results. CBDE can be used as an alternative to BMD for Pseudomonas aeruginosa in a constrained resource setting.