The topic of central line bloodstream infection (CLBSI) is interesting to review because of its inclusion as one of the so-called “pay-for-performance” indicators and there is a large amount of research to look at. I want to focus on one aspect of the issue, which is the difference between information obtained from an administrative database and what is clinical valid.
Published ahead of print in the Journal of the American College of Surgeons is an article from the data mines of the Veterans Administration showing that some 63% of CLBSIs seen at VA hospitals over a 5 year period were actually coded incorrectly. They were falsely positive; the patients did not have CLBSIs.
You might wonder, “How can this be?”
Here’s how. When scrutinized carefully by clinician chart review, 27% had thrombophlebitis at a peripheral IV site, 7% had infected surgically placed arterial graft infections and 6% had cellulitis at an IV site that was not a central line. The rest of the false positives were as follows: chart coded as having a CLBSI when the work-up was actually negative for infection; the patient had no evidence of ever having had a central line; the patient had a CLBSI that was present on admission to the hospital and therefore not preventable during the admission being reviewed.
The problem is that the ICD-9 codes used during the study period (2003-2007) were not specific enough. New codes have been developed but are still not suitable as justification to withhold money from hospitals.
The paper also points out that the CDC has published three different sets of criteria for determining whether a central line is infected.
ICD-9 codes were created for billing purposes, not for studying clinical problems. They are also subject to coding errors as the paper clearly depicts. This is one of a number of papers that illustrate the problems associated with the use of administrative data in analyzing clinical problems.
I’m not necessarily against the pay-for-performance concept, but it must be based on accurate data.
What do you think about "pay-for-performance"?
This post appeared on Sermo yesterday and 84% of those physicians voting did not favor the use of pay for performance.