Is the CLABSI metric now doing more harm than good?

It’s a good news, bad news situation. 

The good news is that we’ve got an effective bundle of practices that can prevent most central line associated bloodstream infections (CLABSIs). Public reporting and pay-for-performance have added to the incentive for hospitals to implement these practices, leading to a dramatic nationwide decline in CLABSI rates. 

The bad news for most hospitals is that the remaining events that meet the NHSN CLABSI definition represent a motley assortment of “one-offs” that aren’t clearly preventable (see this great post from Scott Fridkin for more detail, and to share your “one-offs”). Given how low the CLABSI baseline now is, just a slight “blip” in these one-offs can push a hospital’s SIR above 1 and contribute to financial penalties (for some perspective, the post-intervention CLABSI rate in the widely cited Keystone ICU project (1.4 per 1K line days) now translates to an SIR of >1 at many hospitals). 

These blips result in a tremendous amount of effort by unit leadership and infection prevention programs, as they examine each event for preventability or for common themes—often concluding that the events are either secondary or mucosal-barrier injury (MBI) attributable, but don’t meet the requisite NHSN definition (so must be reported as primary CLABSIs).

HAI definitions, like lab diagnostics, perform best when applied to individuals with moderate-to-high pretest likelihood of disease (or, populations in which the disease is prevalent). This leads to a paradox for HAI surveillance—as prevention approaches improve and HAI rates fall, the positive predictive value of surveillance definitions also declines (if one defines a “true positive” as an event that meets an intuitive clinical definition of the HAI in question—e.g. an old-fashioned primary CLABSI likely to be due to breaches in the process of line insertion and care).

Aside from the time and energy spent chasing one-offs, the continued pressure on the CLABSI metric is going to result in some counterproductive approaches to reach “zero”. After all, there are really only two ways to eliminate CLABSI as it is currently defined: (1) stop using central lines, and/or (2) stop obtaining blood cultures in patients who have central lines.

Show me the data!

The following post is from Scott Fridkin, MD, about the new public availability of NHSN antibiotic resistance data:

Finally, some NHSN antibiotic resistance (AR) data for easy access to all! 

Use the HAI Antibiotic Resistance Patient-Safety Atlas to get metrics of AR for the U.S., your region, or your state. It’s currently limited to NHSN defined HAIs, and aggregate measures; but it is dynamic and will grow in size and functionality. Hopefully this will help public health, the public, providers, and researchers to improve patient care. 

Given the public health priority of preventing antibiotic resistance in healthcare, even before the National Action Plan to Combat Antibiotic Resistance was in place, there was a recognition by CDC that it was imperative to make HAI data reported to CDC more accessible to the public, including the public health community, consumers, the press, and industry partners. In addition, academic researchers could benefit from easier access to generate specific hypotheses to test with more definitive research.

Toward this end, CDC finally has expedited the availability of antibiotic resistance data to allow for more time-sensitive evaluations independent of publishing timelines, to allow diverse approaches to ecologic assessments such as geographic comparisons, and to allow evaluation of subsets of data not previously explored in-depth. This month CDC launched the first version of the HAI Antibiotic Resistance Patient Safety Atlas:

As CDC’s National Healthcare Safety Network has migrated from a sentinel surveillance program to a national performance measurement system, the number of facilities reporting has surpassed 4,000 for acute care hospitals, and 15,000 when including dialysis facilities, long term care, inpatient rehabilitation, and long term acute care. The “events” reported into the system have skyrocketed as well. When you consider that the antibiotic resistance data can include up to four pathogens per infection, there is a huge amount of antibiotic resistance data that rarely sees the light of day. In fact, historically all of the antibiotic resistance data reported to NHSN have been released mostly as peer reviewed papers, with very two-dimensional views of the data. This model provided very limited access to the data, diminished relevance when publication lagged several years behind the reporting year, and limited amounts of data presented given the constraints of the paper-based publication model. Although this first version of the Atlas is fairly limited in one’s ability to create customizable queries, it does allow for temporal and geographic evaluation of trends at a superficial level. For now, the identities of facilities are protected and the data are presented at only the national or state level. However, in future iterations more national customized queries will be possible, and perhaps more granular geographic divisions. For now, I urge anyone to access the maps and query functions and let CDC know how to make the site more useful to your professional endeavors.

Exactly how useful these data will be to the public, press, public health, and most importantly patients – is still uncertain. It is a starting point for improved access, transparency, and innovation to advance antibiotic resistance infection prevention – I hope the users of this Atlas can help us make it better over time.