Biased Low Incidence of Central Venous Catheter-Related Bloodstream Infections in Controlled Clinical Trials?

To the Editor—In the 3SITES trial, ¹ the incidences of central venous catheter (CVC)-related bloodstream infections (CRBSIs) after 1:1:1 randomized catheterization of the internal jugular, subclavian, or femoral vein, respectively, were compared among adult intensive care unit (ICU) patients. The low total CRBSI incidence in this trial was stated to be consistent with data from other ICUs. ² Unfortunately, however, the total CRBSI incidence was not clearly indicated in the 3SITES trial. Using the data provided, ¹ we calculated a CRBSI rate of 1.0% (26 per 2,532 CVCs); for all 3 CVC insertion sites, the mean CRBSI incidence was 1.7 per 1,000 CVC days.

We believe that the reported incidence was rather low and was biased by the diagnostic criteria used. ³ For CRBSI diagnosis in the 3SITES trial, only CVC-tip colonization combined with detection of the same pathogen in peripheral blood was taken into account. Differential time to positivity of central and peripheral blood cultures was not used in the 3SITES trial. ¹ However, this is a well-established parameter for CRBSI diagnosis ⁴ in addition to the aforementioned criterion; it is also a criterion of definite CRBSI. ^{3 , 5} Furthermore, the cutoff of colony-forming units (CFU) for CVC-tip colonization was 1,000 in quantitative culture, ¹ which is higher than is usually recommended. ^{4 , 5} Instead of quantitative cultures, semiquantitative cultures with a CFU cutoff of 15 should be used for diagnosis of CVC-tip colonization. ⁴ However, a CFU cutoff of 15 in semiquantitative culture is the recommended equivalent of a cutoff of 100 in quantitative culture. ^{4 , 5} In addition, the median duration of CVC catheterization in the 3SITES trial was only 5 days; ¹ however, CRBSIs are known to be associated with the length of CVC use in situ. ^{3 , 6}

Due to the diagnostic criteria used and the low number of CVC catheterization days, we believe the CRBSI rate and incidence, respectively, in the 3SITES trial were underestimated and do not represent a real-life setting. Therefore, we compared CRBSI data used in the 3SITE trial with CRBSI data from our SECRECY registry. SECRECY (German Clinical Trial Register No. DRKS00006551) is a real-life registry of CRBSIs in patients with hematological and oncological malignancies; ^{3 , 7} this is a high-risk population for CRBSI, comparable to the ICU patients in the 3SITES trial.

Using SECRECY data up to January 25, 2016, we report on 447 CVCs (internal jugular vein, 419 CVCs [93.7%]), which accounts for a total of 6,700 CVC days. A comparison of diagnostic criteria and characteristics of CRBSI in the 3SITES trial vs the SECRECY registry is provided in Table 1. In this real-life setting, CVCs were in situ longer and the CRBSI rate and incidence were higher than in the 3SITES trial. In the SECRECY registry, median CRBSI onset occurred on day 15.5 after CVC insertion, which is much longer than the median time the CVCs were used in situ in the 3SITES trial (ie, 5 days). Thus, the points discussed here may explain the lower CRBSI incidence in the 3SITES trial, and they should be considered when interpreting data on CRBSI in clinical trials.

TABLE 1 Comparison of Diagnostic Criteria and Characteristics of CRBSI

NOTE. CRBSI, central venous catheter-related bloodstream infection; CVC, central venous catheter; CFU, colony-forming units; n/a, not available.

^a See Mermel et al ⁴ and Hentrich et al. ⁵

^b A CFU cutoff of 15 in semiquantitative culture is the recommended equivalent of a cutoff of 100 in quantitative culture (see Mermel et al ⁴ and Hentrich et al ⁵ ).