2025/07/17 by Po-An Su, Pei-Chun Lai, Yen‐Ta Huang · 1 voice
Biochemistry, Genetics and Molecular Biology · Medicine · #Bacterial Identification and Susceptibility Testing #Sepsis Diagnosis and Treatment
paper · pdf · doi:10.1093/cid/ciaf394
openalex publication_date 2025/07/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/30
To the editor—We read with interest the systematic review by Koroki et al [1] comparing the contamination rates of blood cultures drawn from arterial catheters versus 3 types of venous access in intensive care unit patients. The authors performed pairwise meta-analyses using risk differences for sparse events, which are methodologically appropriate. We commend their approach and complement their findings with a bayesian network meta-analysis (NMA) using the multinma package in R, which allows simultaneous comparison of all 4 blood collection methods: arterial catheters, central venous catheters, venous catheters, and venipuncture. The advantage of NMA is its ability to incorporate both direct and indirect comparison effects [2]. Peripheral venipuncture is regarded as the reference standard for obtaining blood cultures. Using venipuncture as the reference, we observed median risk differences in contamination rates of 0.7% (95% credible interval, −1.4% to 1.6%) for arterial, 3.6% (0.9%–10.2%) for venous, and 5.5% (1.8%–14.7%) for central venous catheters. These results align with the article's conclusions based pm pairwise direct comparison meta-analysis, confirming that blood cultures drawn from arterial catheters may have contamination rates similar to those drawn from venipuncture. Bayesian analysis further revealed an 18.2% probability that blood culture contamination rates from arterial catheters were lower than those from venipuncture catheters. An additional advantage of bayesian NMA is the ability to rank treatments using surface under the cumulative ranking curve (SUCRA) values [3]. Higher SUCRA values indicate a lower contamination risk. The rankings were venipuncture (0.937), arterial catheter (0.717), venous catheter (0.282), and central venous catheter (0.064). These results provide quantitative evidence supporting the traditional view that venipuncture is the reference standard. The bayesian approach yields deeper insight by generating absolute-risk predictions [4]. The predicted median contamination rates were 1.6% (95% credible interval, .7%–3.7%) for venipuncture, 2.3% (1.8%–2.7%) for arterial catheters, 5.3% (2.2%–12.8%) for venous catheters, and 7.2% (3.0%–17.6%) for central venous catheters. International benchmarks classify contamination rates <1% as low and those >3% as high [5]. Our half-eye plot visualization (Figure 1) illustrates the posterior distributions of the predicted contamination rates for each blood collection method. The figure clearly shows that contamination rates for the arterial catheter group totally overlap with those for the venipuncture group, and blood cultures from arterial catheters almost never exceed the 3% high-risk contamination threshold. By contrast, both peripheral and central venous catheters exhibit higher contamination rates and wider uncertainty. Half-eye plot showing posterior distributions of predicted blood culture contamination rates by collection method. Density curves represent the full posterior distribution from bayesian network meta-analysis. Points indicate median estimates (with 95% credible intervals). Vertical dashed lines mark internationally recognized benchmarks of 1% (acceptable) and 3% (high contamination). These findings support the conclusion of Koroki et al [1] that arterial catheters may be a reasonable alternative to venipuncture in critically ill patients when venipuncture is challenging. NMA using a bayesian approach is increasingly being applied to generate clinical evidence on the issue of infection in critical care [6, 7]. The bayesian NMA framework offers the advantage of borrowing strength across all available comparisons, potentially increasing the precision of the estimates. Acknowledgments. The authors thank Yu-Kang Tu, DDS, PhD of the Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei; his workshop equipped them with the know-how to carry out bayesian meta-analyses using the multinma package. Author contributions. Writing—original draft: P. A. S. Formal analysis: P. C. L. Writing—review and editing and project administration: Y. T. H. All authors read and approved the final manuscript. Data availability. The data used in this study are publicly available in Clinical Infectious Diseases, May 2025 (DOI: 10.1093/cid/ciaf260). Financial support. The authors received no financial support for the research, authorship, and/or publication of this article.