Shepherd EG, et al. Significant Reduction of Central-Line Associated Bloodstream Infections in a Network of Diverse Neonatal Nurseries. The Journal of pediatrics. 2015. This is an observational study from a group which includes a large number of neonatal beds. They report the results of their quality control efforts specifically for central line associated blood stream infections (CLABSI, as we are now calling them). They show a dramatic reduction over the period from 2003 to 2007, and a progressive continued improvement since then, down to very low current levels of  less than 1 infection per 1000 catheter days. The patients include small preterm infants as well as a large number of surgical patients. There has also overall been a reduction in the number of central line days per 1000 admissions.

One issue I have with this study is the question of definitions of CLABSI; they state that they used the definitions which were in place at each time, approved by the national surveillance programs. But it is very hard to find out exactly what those definitions were. The current definition for CLABSI caused by “skin commensals” (including coagulase negative staph) requires 2 blood cultures with the same organism, taken at separate times, less than 48 hours apart. Was that always the case? If not it could have a major impact on the implications of these numbers. They also don’t report the data in the highest risk subgroups, such as the very preterm baby, nor the overall rate of BSI that are not CLABSI. If the infections among extremely immature babies have just shifted from CLABSI to BSI without CL (which I think is probably not the case, but you can’t tell from these numbers) then that has not necessarily benefited the babies.

These issues have to be taken into account when we try and compare rates and approaches over time, and between hospital systems. For example, the Canadian Neonatal Network reports CLABSI as any positive blood culture, including CoNS, in association with a CL; in order to acknowledge that many of the time babies have a single culture, followed by the start of antibiotics. They won’t necessarily have another positive culture, as the antibiotics are already working, so even though they may be clearly infected, they would not be counted as a CLABSI according to the definition used in this paper. The large majority of CLABSI occur in the very preterm infant, so differences in patient populations and in catheter use can have a major effect on CLABSI rates. The CNN data also show somewhat more variability in CLABSI than in overall rates of late-onset infection, reflecting different patterns of catheter use I think, and that some of the variability of CLABSI is offset by occurrence of non-CLA BSI in the very preterm infant.

I think we really need to have data on CLABSI, non-CLA BSI, those numbers among infants of less than 29 weeks, and the numbers if you accept a single CoNS culture (maybe with some other evidence that it is a real infection); in order to be able to really compare results.

Nevertheless these are impressive results, the methods they followed are all available in the appendix on-line.