Adults with diabetes whose long-term blood sugar was poorly controlled were far more likely to yield cultures of multidrug-resistant organisms at a regional teaching hospital in southern Taiwan, according to a retrospective clinical microbiology study published in Annals of Clinical Microbiology and Antimicrobials. Among culture-positive isolates linked to patients with diabetes and available glycated hemoglobin data, 32.4 percent of isolates from patients with HbA1c above 8.0 percent were multidrug-resistant, compared with 13.4 percent among patients at or below that threshold. After adjustment for age, sex, and specimen type, poor glycemic control was associated with increased odds of a multidrug-resistant culture, with an adjusted odds ratio of 2.82 (95 percent confidence interval 1.84 to 4.32; p < 0.001).
The study, conducted at Pingtung Christian Hospital, a 676-bed regional teaching hospital serving Pingtung County and surrounding areas, reviewed culture-positive bacterial records from adult outpatients and inpatients between January and December 2023. From an initial screen of 10,153 culture-positive records, the researchers applied predefined exclusions and retained 1,989 eligible non-duplicate clinical isolates, meaning each patient contributed only one isolate to the analysis. The isolates came from four specimen types: abscess or pus, urine, sputum, and blood. Of the included isolates, 902 were linked to patients with diabetes mellitus and 1,087 to patients without diabetes.
The analysis focused on six clinically monitored resistance phenotypes routinely reported by the hospital laboratory and infection-control program: methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus (VRE), carbapenem-resistant Klebsiella pneumoniae (CRKP), carbapenem-resistant Escherichia coli (CREC), carbapenem-resistant Acinetobacter baumannii (CRAB), and carbapenem-resistant Pseudomonas aeruginosa (CRPA). The authors stress that this operational definition reflects local surveillance priorities and should not be read as an organism-wide multidrug-resistance definition based on non-susceptibility across all antimicrobial classes. Notably, ESBL-producing Enterobacterales without carbapenem resistance were not among the predefined phenotypes.
Overall, 18.4 percent of the 1,989 isolates were multidrug-resistant. The proportion was 19.5 percent in the diabetes group and 17.5 percent in the non-diabetes group, a difference that was not statistically significant (p = 0.244). Diabetes status alone, in other words, did not predict a higher overall burden of resistant organisms. The picture changed, however, when the researchers stratified by specimen source. In urine specimens, the multidrug-resistant proportion was significantly higher in the diabetes group than in the non-diabetes group (61 of 399 isolates, 15.3 percent, versus 41 of 392 isolates, 10.5 percent; p = 0.043). In specimen-stratified comparisons, diabetes was associated with a modest but statistically significant increase in the odds of a resistant urine culture (odds ratio 1.55; 95 percent confidence interval 1.01 to 2.36). For abscess or pus, sputum, and blood specimens, the odds ratios were above 1.0 but the comparisons were not statistically significant.
The distribution of resistance phenotypes differed markedly by specimen type, a pattern that was statistically significant. MRSA dominated abscess and pus specimens, accounting for 102 of 112 multidrug-resistant isolates from that source, and it was also the predominant phenotype in blood, accounting for 60 of 70 resistant blood isolates. Urine told a different story: VRE was the predominant resistant phenotype there, making up 76 of 102 urinary multidrug-resistant isolates, even though Escherichia coli was the most common organism isolated from urine overall. In sputum, MRSA was the most frequent resistant phenotype by absolute count, but CRAB and CRPA were relatively more prominent in sputum than in any other specimen type. The total numbers of resistant isolates were 112 in abscess or pus specimens, 102 in urine, 82 in sputum, and 70 in blood.
The authors caution that the prominence of VRE in urine must be read against their operational definition. Because ESBL-producing Enterobacterales without carbapenem resistance were excluded from the monitored phenotypes, the predominance of VRE among urinary resistant isolates does not mean VRE was the most common urinary pathogen overall; it predominated within the monitored urinary resistance phenotypes. Possible contributors include endemic VRE circulation, repeated healthcare exposure, urinary catheterization, and selective pressure from broad-spectrum antimicrobials. Because the retrospective design lacked ward-level temporal clustering, institutional antibiotic consumption data, and molecular typing, the researchers could not determine whether the urinary VRE pattern reflected a localized outbreak, clonal spread, or endemic institutional circulation.
On glycemic control, the association with resistant cultures was robust across sensitivity analyses. HbA1c was analyzed as a continuous variable, and each 1 percent increase was associated with increased odds of a resistant culture after adjustment for age, sex, and specimen type (adjusted odds ratio 1.22; 95 percent confidence interval 1.10 to 1.36; p < 0.001). When HbA1c was divided into strata of 7.0 percent or less, 7.1 to 8.0 percent, 8.1 to 9.0 percent, and above 9.0 percent, the proportion of resistant cultures rose progressively across worsening glycemic categories, a dose-response pattern with p for trend below 0.001. A model further adjusted for inpatient status at the time of culture collection still yielded a significant association (adjusted odds ratio 2.55; 95 percent confidence interval 1.63 to 3.98). Because resistant cultures were not rare, the researchers also ran modified Poisson regression with robust standard errors; the adjusted risk ratio for HbA1c above 8.0 percent was 2.18 (95 percent confidence interval 1.57 to 3.03), and 2.05 (95 percent confidence interval 1.46 to 2.88) after additional adjustment for inpatient status. The risk ratios were smaller in magnitude than the odds ratios, but the direction and significance were consistent.
The researchers took several steps to address potential bias. Comparing isolates linked to patients with and without available HbA1c data, the multidrug-resistant proportion was similar (20.2 percent versus 18.1 percent; p = 0.471), and age, sex, specimen type, and bacterial species distributions were broadly comparable. However, isolates with available HbA1c data were more often obtained during inpatient encounters (53.7 percent versus 43.6 percent; p = 0.006), suggesting that HbA1c availability partly reflects clinical testing patterns and healthcare utilization. Specimen type and bacterial species distributions were also broadly comparable between the two HbA1c categories (p = 0.848 and p = 0.983), while the resistant-culture proportion differed significantly between them (p < 0.001). Routine lipid markers and uric acid showed no significant association with resistance status.
The findings carry practical weight for clinical microbiology interpretation and antimicrobial stewardship. The authors suggest that antimicrobial decision-making should incorporate specimen type, prior culture history, local resistance patterns, glycemic-control information, and final susceptibility results. Urine cultures from patients with diabetes may warrant attention to VRE, abscess and blood cultures may require awareness of local MRSA epidemiology, and sputum cultures may call for attention to carbapenem-resistant non-fermenting Gram-negative bacilli. They emphasize that these observations are not prescriptive empirical treatment recommendations; rather, they show how routine culture and susceptibility data can support specimen-specific interpretation and communication between laboratories, infection-control teams, and frontline clinicians, particularly in regional hospitals where resources for molecular surveillance may be limited.
Important limitations frame the conclusions. This was a single-center retrospective study, so the results may not generalize to other hospitals, regions, or healthcare systems with different patient populations and resistance patterns. The analysis was based on culture-positive isolates rather than prospectively adjudicated clinical infections, so some isolates, particularly from sputum and urine, may represent colonization rather than true infection, which is why the authors prefer the term multidrug-resistant cultures over multidrug-resistant infections. The study also lacked detailed clinical variables such as prior antimicrobial exposure, hospitalization history, ICU admission, catheter use, wound severity, duration of diabetes, and comorbidity burden, so residual confounding by healthcare exposure, disease severity, device use, and antimicrobial selective pressure may remain despite adjustment. Molecular testing for resistance determinants such as mecA or vanA/vanB was not performed. The authors interpret the HbA1c-related findings as associations rather than proof of causality or fully independent predictive effects, and they call for multicenter and prospective studies to distinguish colonization from infection and to test whether integrating glycemic-control information into microbiology interpretation improves empirical antimicrobial selection.
Subject of Research: Specimen-specific distribution of multidrug-resistant organisms and their association with glycemic control in adults with diabetes
Article Title: Specimen-specific distribution of multidrug-resistant organisms and its association with glycemic control in adults with diabetes: a retrospective clinical microbiology study in southern Taiwan
Article References: Chen, J.-S., Chou, M.-Y., Huang, S.-R., & Liao, C.-S. (2026). Specimen-specific distribution of multidrug-resistant organisms and its association with glycemic control in adults with diabetes: a retrospective clinical microbiology study in southern Taiwan. Annals of Clinical Microbiology and Antimicrobials, 25(1), Article 38. https://doi.org/10.1186/s12941-026-00877-0
Image Credits: AI Generated
DOI: 10.1186/s12941-026-00877-0
Keywords: multidrug-resistant organisms, diabetes mellitus, HbA1c, glycemic control, MRSA, VRE, carbapenem resistance, clinical microbiology, antimicrobial stewardship, urine specimens, Taiwan, retrospective study
News Source: Ophelia Keating. (November 5, 2026). Poor blood sugar control linked to resistant bacteria in Taiwanese diabetes study. Scienmag.
