Why AST results need review: data from Russian laboratories
Table of Contents
An antimicrobial susceptibility test becomes useful to clinicians when the measurement is correctly matched to interpretive criteria and the resulting category is consistent with the organism’s known properties. In practice, this takes more than selecting a row in a table: the standard’s version, the organism–drug combination and cases without applicable breakpoints all matter.
In an original CMAC study, Vinogradova and colleagues compared susceptibility categorization using aggregated data from Russian microbiology laboratories in 2023 and 2024. The study did not assess treatment outcomes; it examined the quality of laboratory reporting and interpretation.
Read the article at the journal · Download the Russian-language PDF
What was compared
The analysis included data from 494 laboratories in 2023 and 520 in 2024. The authors considered the interpretation standards and versions in use, microorganisms and federal districts. They examined the proportion of results flagged for review and how those flags were distributed by type.
The paper calls these “errors/inaccuracies”, but they are an analytic category, not confirmed errors in every individual report: the authors used aggregate data without reviewing original test records or clinical circumstances. In 2023, 125,435 of 1,694,993 tested results (7.4%) were placed in this category; in 2024, 244,461 of 2,506,844 (9.8%). These proportions should not be applied without qualification to every laboratory or interpreted as the proportion of incorrect clinical decisions.
What kinds of findings were flagged
The category covered four types of situations warranting expert review. The figures below are proportions of all tested results in each year:
| Situation | 2023 | 2024 |
|---|---|---|
| Expected resistance | 3.2% | 4.4% |
| Rare phenotype | 3.1% | 3.1% |
| Susceptibility category assigned without interpretive criteria | 1.9% | 2.4% |
| Susceptibility-category correction needed | 1.2% | 1.4% |
These are not four mutually exclusive groups of results: the row percentages must not be added together. Cases of expected resistance assigned a category other than R are particularly important to review because of the possible risk of false susceptibility. Aggregate data do not establish that every flagged case was an error.
Why the two percentages do not prove uniform deterioration
The aggregate proportion was 2.35 percentage points higher in 2024, but the laboratories and results included in the two years were not identical. The authors therefore compared 377 laboratories represented in both periods. That paired analysis did not find a statistically significant change in one direction across the comparable laboratories.
In short, the aggregate comparison shows a difference between datasets; it does not show that flagged findings increased in every comparable laboratory. Evaluating a trend requires a like-for-like comparison.
Where expert review matters most
The highest proportions of errors and inaccuracies were found among laboratories using only MUK 4.2.1890-04: 31.3% in 2023 and 41.0% in 2024. In multivariable analysis, a higher proportion was also associated with standards whose version year was unknown. These are associations in observational data, not proof that the choice of standard alone explains every discrepancy.
The article’s conclusion extends beyond keeping breakpoint tables current. Quality control also requires reviewing expected resistance, rare phenotypes and organism–drug combinations for which interpretive criteria are unavailable. The standard’s version should be known, and ambiguous results should be available for expert assessment.
Why the organism matters
The overall proportion conceals substantial differences among species. The proportion of tested results placed in the errors/inaccuracies category was:
| Organism | 2023 | 2024 |
|---|---|---|
| Staphylococcus aureus | 30.6% | 42.8% |
| Enterococcus faecium | 36.5% | 32.8% |
| Streptococcus pneumoniae | 16.0% | 19.0% |
| Acinetobacter baumannii | 16.9% | 11.7% |
| Pseudomonas aeruginosa | 10.5% | 9.4% |
| Escherichia coli | 0.9% | 2.8% |
| Klebsiella pneumoniae | 1.8% | 2.1% |
Each proportion uses the number of tested results for that species in that year as its denominator. These figures are not a ranking of laboratory quality: the organism–drug combinations tested, standards used and composition of the data differed.
The complete Table 1 below shows not only overall proportions but also the type of finding, absolute counts and cases assigned a category other than R. For S. aureus, a susceptibility category was assigned without interpretive criteria in 10.07% of tested results in 2023 and 16.81% in 2024. For E. faecium, expected-resistance flags accounted for 18.09% and 15.74%, respectively. Each flagged result calls for review in the context of the specific antibiogram.
Scroll horizontally to see the full table. Cells show the percentage and count (n/N); “category ≠ R” is a subset of findings of that type.
| Organism | Finding type | 2023 | 2024 | Difference, pp | |||
|---|---|---|---|---|---|---|---|
| All findings | Category ≠ R | All findings | Category ≠ R | All findings | Category ≠ R | ||
| Acinetobacter baumannii | Expected resistance | 14.04% 12 000 / 85 496 | 4.2% 3 587 / 85 496 | 10.73% 9 087 / 84 666 | 4.86% 4 117 / 84 666 | -3.31 | 0.66 |
| Acinetobacter baumannii | Category without criteria | 2.86% 2 443 / 85 496 | 1.85% 1 581 / 85 496 | 0.96% 815 / 84 666 | 0.54% 461 / 84 666 | -1.90 | -1.31 |
| Enterococcus faecium | Expected resistance | 18.09% 13 376 / 73 956 | 14.47% 10 699 / 73 956 | 15.74% 11 636 / 73 948 | 13.11% 9 694 / 73 948 | -2.35 | -1.36 |
| Enterococcus faecium | Category without criteria | 12.21% 9 030 / 73 956 | 9.26% 6 851 / 73 956 | 10.79% 7 979 / 73 948 | 6.5% 4 806 / 73 948 | -1.42 | -2.76 |
| Escherichia coli | Expected resistance | 0.92% 6 755 / 737 640 | 0.54% 4 011 / 737 640 | 2.80% 32 317 / 1 153 379 | 2.08% 24 042 / 1 153 379 | 1.88 | 1.54 |
| Klebsiella pneumoniae | Expected resistance | 1.78% 8 294 / 466 789 | 0.86% 3 996 / 466 789 | 2.09% 12 745 / 610 187 | 0.93% 5 693 / 610 187 | 0.31 | 0.07 |
| Pseudomonas aeruginosa | Expected resistance | 7.45% 9 528 / 127 972 | 6.22% 7 962 / 127 972 | 8.10% 14 146 / 174 688 | 7.07% 12 359 / 174 688 | 0.65 | 0.85 |
| Pseudomonas aeruginosa | Category without criteria | 3.04% 3 895 / 127 972 | 2.6% 3 332 / 127 972 | 1.29% 2 253 / 174 688 | 1.05% 1 838 / 174 688 | -1.75 | -1.55 |
| Staphylococcus aureus | Expected resistance | 1.84% 2 935 / 159 577 | 1.41% 2 245 / 159 577 | 6.30% 18 839 / 299 001 | 4.53% 13 549 / 299 001 | 4.46 | 3.12 |
| Staphylococcus aureus | Category without criteria | 10.07% 16 065 / 159 577 | 8.04% 12 832 / 159 577 | 16.81% 50 248 / 299 001 | 11.15% 33 333 / 299 001 | 6.74 | 3.11 |
| Streptococcus pneumoniae | Expected resistance | 3.45% 1 432 / 41 478 | 1.95% 809 / 41 478 | 9.26% 10 244 / 110 668 | 5.98% 6 619 / 110 668 | 5.81 | 4.03 |
Adapted from Table 1 in the original article (Vinogradova et al., CMAC, 2026; CC BY 4.0). Only organism–finding combinations reported by the authors are shown. A flagged finding does not mean that every individual report contained a confirmed error.
Interactive dashboard
The journal provides an interactive dashboard. You can open it in a separate window or try viewing it on this page. It helps explore the data presented by the authors, but does not replace the study’s methods and limitations.
This dashboard is hosted by the CMAC journal and loads only when you open this panel. If the embedded view does not work, open it on the journal site.
Show the interactive dashboard
What this means for microbiology results
For laboratories and clinicians, obtaining an MIC or inhibition-zone diameter is not the end of the process. The susceptibility category reported in an antibiogram needs to be justified, and exceptions or unusual combinations need review before the result enters the microbiology report. If these data are later used for AMR surveillance, the quality of interpretation also affects the analytic dataset.
This is the part of the workflow addressed by expert review in ABioGram: the system helps check AST results while retaining professional oversight of complex cases. This describes the product’s scope; it does not imply that the study evaluated ABioGram’s effectiveness.
A related question – how to assess the risk of a resistant pathogen before microbiology results arrive – is covered in our review of a systematic study of risk factors.
Source: Vinogradova A.G. et al. Clinical Microbiology and Antimicrobial Chemotherapy. 2026;28(2):253–263. DOI: 10.36488/cmac.2026.2.253-263.