Carbapenemase types shifted in hospital CRKP infections, retrospective study finds

A decade-long retrospective study of 324 clinically isolated carbapenem-resistant Klebsiella pneumoniae strains found a marked change in the enzymes associated with carbapenem resistance at a tertiary hospital in Shanxi Province, China. NDM carbapenemase predominated from 2015 through 2020, accounting for an average annual proportion of 70.24% of isolates. From 2021 through 2024, KPC became dominant, reaching an average annual proportion of 84.70%, according to the study published in Frontiers in Public Health.

The analysis, conducted at the Second Hospital of Shanxi Medical University, a 3,200-bed facility, included unique inpatient isolates collected from January 2015 through December 2024. Researchers defined CRKP as K. pneumoniae resistant or intermediate to meropenem, imipenem or ertapenem. They used the modified carbapenem inactivation method, or mCIM, and the EDTA-modified carbapenem inactivation method, or eCIM, to assess carbapenemase production, while polymerase chain reaction and sequencing were used to detect genes encoding KPC, NDM, IMP, VIM and OXA-48. The first isolate from each patient was included, and duplicate isolates from the same infection site were excluded.

Among all 324 isolates, 202, or 62.35%, were positive for KPC and 108, or 33.33%, were positive for NDM. Ten isolates did not produce a detectable carbapenemase, while four produced both KPC and NDM. A multiple regression discontinuity analysis placed the transition at 2021 and adjusted for ward distribution, specimen type and length of hospital stay. At that breakpoint, the proportion of KPC-producing isolates increased by an estimated 41.53 percentage points, although that estimate was not statistically significant at the conventional threshold, with a 95% confidence interval of −7.16 to 90.18% and a p value of 0.077. The NDM proportion fell by an estimated 46.06 percentage points, with a 95% confidence interval of −87.73% to −4.39% and a p value of 0.037. The authors interpreted the results as evidence of a population-level epidemiological transition rather than simply a transient clonal fluctuation, but the absence of whole-genome molecular typing prevents definitive confirmation of that interpretation.

Respiratory infections predominated in the cohort. Sputum was the source of 206 isolates, representing 63.58% of the total, followed by urine with 39 isolates, or 12.04%, and blood with 21, or 6.48%. Pulmonary infection was likewise the most common clinical infection category, accounting for the same 206 patients, while bloodstream infection accounted for 39. Infection types other than the major categories each had fewer than five annual cases, and their combined proportion showed an increasing annual trend. Although the study linked each organism to the patient’s primary infectious focus to help distinguish infection from colonization, respiratory dominance does not by itself establish transmission dynamics or the source of the observed shift.

The intensive care unit accounted for the largest departmental group, with 78 CRKP patients, or 24.07% of the cohort. The annual proportions of patients admitted to neurology and nephrology increased over the study period, while no significant temporal change was detected across the seven specimen categories. Sepsis and septic shock accounted for 59 patients, or 18.21%, and severe pneumonia accounted for 62, or 19.14%. The proportion classified as sepsis increased significantly over time, with a trend-test p value of 0.029, indicating a temporal increase in the share of recorded CRKP infections classified as sepsis. The observational design could not determine whether changes in patient risk, treatment practices, screening or transmission contributed to that pattern.

All isolates were resistant to meropenem and imipenem, with minimum inhibitory concentrations of at least 16 mg/L for both agents. Resistance was also detected in 184 isolates, or 56.79%, for cefoperazone-sulbactam; 134, or 41.35%, for amikacin; 85, or 26.23%, for aztreonam; 29, or 8.95%, for tigecycline; and four, or 1.23%, for colistin. Most isolates remained susceptible in vitro to colistin, tigecycline and aztreonam, but susceptibility results do not by themselves establish clinical effectiveness in an individual infection. The researchers noted that the enzymes involved differ in their response to available agents: the beta-lactam and beta-lactamase inhibitor ceftazidime-avibactam inhibits KPC and some OXA-48-like enzymes but not metallo-beta-lactamases such as NDM, while aztreonam retains stability against metallo-beta-lactamases. The study did not test patient responses to particular drugs or compare clinical treatment strategies.

Patient-level analyses found that infection severity, disease location, admitting department and specimen type were associated with selected measures of hospitalization, cost and outcome. In unadjusted comparisons, hospital stay differed by disease and specimen classification, while total drug cost differed by severity, department and disease classification. Antimicrobial drug cost differed by severity, disease and specimen type, and treatment outcome differed by severity. After multivariable adjustment for age, comorbidities, infection severity, ICU admission, invasive procedures, source control, antimicrobial therapy, carbapenemase type and study year, invasive procedures were associated with a longer hospital stay, with a hazard ratio of 1.74 and a 95% confidence interval of 1.22 to 2.48. Disease category was no longer independently associated with length of stay, while specimen category showed only a marginal association, with a p value of 0.077.

Adjusted analyses identified severity as the only classification independently associated with discharge-defined treatment outcome. Patients with milder infections had an odds ratio of 0.32 for an adverse outcome compared with patients in the sepsis group, with a 95% confidence interval of 0.16 to 0.66 and a p value of 0.002. The model showed acceptable fit, with a Hosmer-Lemeshow p value of 0.12, and all variance inflation factors were below 5. These results refer to care at discharge: cured and improved outcomes were classified as effective treatment, while death, non-improvement, worsening, non-medical discharge and abandonment were classified as ineffective treatment. The observational analysis supports an adjusted association between severity classification and outcome, rather than establishing that severity alone caused the differences.

The median total drug cost was 99,198.35 Chinese yuan, with an interquartile range of 45,770.65 to 197,347.13 yuan, and the median antimicrobial drug cost was 6,932.51 yuan, with an interquartile range of 2,521.48 to 17,441.01 yuan. After adjustment, ICU admission was associated with 1.37 times the total drug cost of non-ICU admission, with a 95% confidence interval of 1.03 to 1.83. Sepsis was associated with 1.54 times the total drug cost of milder infections and 2.82 times the antimicrobial drug cost. Pulmonary infection carried 2.39 times the total drug cost and 5.64 times the antimicrobial drug cost of urinary tract infection. The investigators did not adjust the cost data for inflation because China’s zero-profit drug policy in 2017 and centralized procurement from 2019 changed prices during the study period, limiting comparisons across years.

The study is a single-center retrospective analysis of selected inpatient infections, so its results may not represent other hospitals, regions or patient populations. It did not examine every nonenzymatic resistance mechanism, and no whole-genome typing was available to trace lineages or determine whether the 2021 change reflected clonal replacement, multiple introductions or other transmission processes. Antimicrobial testing methods were updated in line with standards used at the relevant times, although laboratory identification and routine screening intensity were reported as stable. Taken together, the findings document a shift from NDM toward KPC among CRKP isolates at one Shanxi hospital, widespread resistance to two carbapenems, predominantly respiratory infection, a large ICU burden and an increasing sepsis proportion. The authors called for independent multicenter validation, while the study’s main evidentiary value lies in documenting a local molecular and clinical pattern that requires broader surveillance to determine its reach.

Subject of Research: Retrospective molecular epidemiology and clinical distribution of carbapenem-resistant Klebsiella pneumoniae at one hospital in Shanxi, China

Article Title: Trend of carbapenemase changes and epidemiological distribution of carbapenem-resistant Klebsiella pneumoniae clinically isolated in a hospital: a single-center retrospective study from 2015 to 2024

Article References: Wang, Z., Hu, H., Wang, S., Yin, D., Duan, J., & Kang, J. (2026). Trend of carbapenemase changes and epidemiological distribution of carbapenem-resistant Klebsiella pneumoniae clinically isolated in a hospital: a single-center retrospective study from 2015 to 2024. Frontiers in Public Health, 14, Article 1944879. https://doi.org/10.3389/fpubh.2026.1944879

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1944879

Keywords: CRKP, carbapenem resistance, Klebsiella pneumoniae, KPC carbapenemase, NDM carbapenemase, hospital epidemiology, antimicrobial susceptibility, clinical outcomes, hospital costs, Shanxi China, Trend, carbapenemase