The laboratory team reinterpreted existing disk diffusion inhibition zone diameters recorded in the Buea Regional Hospital WHONET database for the period 2023–2025. Raw zone diameters for clinical isolates were the only input reclassified against two breakpoint systems: EUCAST 2025 and CLSI 2025. The cohort comprised 1,198 isolates in total: 748 Enterobacterales, 367 Staphylococcus and 83 Pseudomonas. This was a retrospective, laboratory-based analysis of stored antimicrobial susceptibility data; the source did not report any prospective clinical intervention or dosing changes tied to the reclassification.
Investigators evaluated concordance between the two standards using Cohen’s kappa (κ). The study team treated κ values below 0.61 as indicating poor agreement. Major interpretation switches—defined as categorical flips between susceptible and resistant—were counted for each antibiotic-organism pair. The authors also used the exact McNemar test to determine whether one standard systematically yielded more non-susceptible outcomes than the other.
Agreement varied across organism groups. For Enterobacterales, 4 of 12 antibiotics (33.3%) showed poor agreement; the lowest κ reported was for ofloxacin at 0.48. For Staphylococcus, 5 of 10 antibiotics (50%) fell below the κ threshold, with levofloxacin demonstrating the weakest concordance (κ = 0.08). The greatest discordance was seen with Pseudomonas, where 5 of 7 antibiotics (71.4%) had poor agreement; cefepime and imipenem both had κ = 0.24, the lowest values reported for that group. The source lists specific κ values for these low-agreement pairs but did not include granular per-isolate raw data in the abstract beyond aggregate counts.
Major interpretation switches were documented across all three organism groups. Among Enterobacterales, three antibiotics showed major susceptible↔resistant flips: amox/clavulanic acid (1.1% of isolates), ceftriaxone (0.6%), and ofloxacin (15.6%). The investigators observed more frequent major switches among Staphylococcus isolates: six of ten antibiotics demonstrated sizable flips, notably minocycline (21.1%), gentamicin (15.9%), rifampicin (14.4%), and levofloxacin (13.7%); clindamycin and tetracycline also contributed smaller percentages (2.7% and 4.5%, respectively). For Pseudomonas, two antibiotics had major switches: cefepime (20%) and ciprofloxacin (3%).
These switches reflect cases in which identical zone diameters cross different susceptibility thresholds under the two standards and therefore would lead to opposite categorical interpretations for the same isolate.
The exact McNemar test applied by the study team indicated a systematic difference in classification. Data indicated that EUCAST was significantly more likely to classify isolates as non-susceptible compared with CLSI (p < 0.001). The abstract reports this directional bias without specifying per-antibiotic p-values in the summary; the reader would need to consult the full manuscript or supporting files for a breakdown by drug and organism.
Discordant breakpoint interpretation can produce two practical problems. First, clinicians relying on local laboratory reports could receive opposite categorical advice—susceptible versus resistant—for the exact same disk diffusion result, depending on whether the laboratory applies EUCAST or CLSI. Second, aggregated national and regional antimicrobial resistance (AMR) surveillance datasets assembled from laboratories using mixed breakpoint systems may be internally inconsistent, degrading trend analyses and policy decisions.
The authors recommend adoption of a single national breakpoint standard in Cameroon to reduce interpretative heterogeneity and to improve the reliability of AMR surveillance. The source frames this recommendation on the frequency of major interpretation switches and the observed bias toward more non-susceptible calls when using EUCAST 2025 in this dataset.
This report is a medRxiv preprint and has not undergone peer review; the authors explicitly state that the findings should not yet guide clinical practice. The study is retrospective and single-centre, limited to the Buea Regional Hospital WHONET records from 2023–2025. The abstract does not provide detailed per-isolate tables or methodological minutiae in the summary—such as the full list of antibiotics included per organism beyond those named, or how intermediate categories were handled—so readers must consult the full manuscript or supporting information for those specifics. The authors disclose ethical oversight from the South West Regional Ethics Committee for Human Health Research.
Outstanding, reportable questions documented in the source include the operational impact of switching national laboratories from mixed breakpoint use to a harmonized standard and the effect such harmonization would have on reported AMR prevalence estimates. The source recommends a national policy decision but does not present prospective outcome data showing how harmonization would change clinical outcomes or stewardship metrics; that question remains open for future, ideally multicentre, prospective evaluation.