Thursday, May 29, 2008

Central Line Infection Update

I have been writing for many months about our efforts to eliminate central line infections, starting with this post in December of 2006, and then giving periodic updates on this blog. Now, there is a regular feature on the BIDMC website where people can check in on this and other clinical indicators.

But we received a report yesterday that I have to share in this forum again. The chart above shows the year-by-year number of central line infections at BIDMC, as measured in cases per 1000 patient days in our intensive care units. Clear and steady progress over the last several years is evident, and I have to admit that I am really proud of our folks -- nurses, doctors, residents, and others -- in our ICUs.

What does this mean in real terms? Well, in a typical month, there might be 1500 patient days in our ICUs. With a drop in our infection rate from over 4.0 to under 1.0, it means that more than 4 people per month (.003 x 1500) did not get a central line infection. With a mortality rate of, say, 20% among those getting such infections, it means we are saving the equivalent of roughly one life per month.

This also saves lots of dollars, mainly for the insurance companies and governments who pay for the extended stays that result from infections. It also frees up scarce ICU capacity for seriously ill patients who show up in our Emergency Department or are otherwise admitted to the hospital.

What's left? What does it take to get to zero? Frankly, we are not sure. There is no distinct pattern of causality among the cases we still see -- and we evaluate each and every one. I'd love to hear from others out there who have also been successful on this front if they want to offer comments on anything we might want to consider to achieve our ultimate goal of a sustained "zero".


Anonymous said...

CONGRATULATIONS. It's been very interesting to follow this issue from the beginning of the effort to this result.

Anonymous said...

Paul, I admire your leadership of this rate of improvement...seems that with committed front line microsystem teams, evidence-based ideas for change, and curious leaders with a never-ending mission to make care better for patients, you show better care week after week. Your transparency on results is energizing! Thanks to you and all at BIDMC for leading the way on improving care.
Maureen Bisognano

Parker said...

While I share the enthusiasm for reducing the numerator of line infections, I wonder if the denominator of patient-days in the ICU is somewhat diluted over the past few years. I don't have the numbers, but our chair of medicine recently told the department that we have the busiest ICUs in Boston, and I think with the Triggers system, there is pressure to put less-sick patients in the ICU "just in case," meaning that the average ICU patient has a lower APACHE-II s score than a few years ago--and is therefore less likely to have a central line at all than a few years back.

I guess what I am asking is, wouldn't the number of patient-days of indwelling central lines be a more appropriate denominator?

Anonymous said...

By indwelling, you mean ICUs + regular med/Surg floors, I take it. We do keep track of those, too, and the numbers outside of the ICUs are very, very small.

Anonymous said...

I think Parker may be asking if your denominator is ALL patient-days in ICU (including patients with no catheter), or only patients-with-indwelling-catheter-days in the ICU. I suspect and hope it's the latter.

Anonymous said...

Oh, thanks for clarifying the question. Yes, of course, it is the latter. Otherwise, it would not be meaningful.

Unknown said...


I am the Administrator of a hospital in the Pacific Northwest and we have been successful in implementatin of the VAP Best Bundles - it's been over 27 months now. We're now moving on to UTIs and BSIs. Any advice on bundles that work that could be shared with our team?


Anonymous said...

We rely heavily on IHI advice on these matters, but I'll also pass along your inquiry to folk here.

Best of luck.

Anonymous said...

I was wrong! The denominator we use is, in fact, all ICU patient days, not catheter days. I learned today that we do not yet have a good method for counting catheter days. We are putting in a new ICU monitor system, which should allow us to do this.

As I note above, this undermines the interpretation I have presented of the number. While it is still useful -- if we assume a relatively constant mix of catheterized vs non-catheterized patients -- it is less so if Parker's hypothesis is correct, i.e., fewer catheterized patients. His point may amount to a rounding error, or it might be more significant. Hard to know.

Meanwhile, though, the state has mandated that hospitals start reporting infections relative to catheter days, which we will have to do by hand until our new IS is in place.

So, apologies if I have overstated the impact of this somewhat. That should in no way take away from the fine work people have done or the fact that every single infection still receives a full root-cause review.

Anonymous said...

A follow-up note from our infection control folks:

I would add that we have recently been looking at this issue as we validate our method of hand gathering catheter days in the ICUs for public reporting (until the IS solution is available in the fall). As we had suspected, in the majority of units, most patients have a central line each day. So, in most ICUs, one patient day equals one catheter day. We do see some exceptions, such as the CCU, where not every patient has a central line each day. Note that the CDC states that if you do not have catheter day data, you should use patient days as the denominator. This is not allowable when you report data to the CDC (as we will start to do for public reporting).

LosAngelesEpidemiologist said...

As an Epidemiologist, I just want to note that the denominator used is 1,000 patient days. NHSN benchmarking traditionally uses central line days as a denominator because it is more reflective of the rate of infection. When patient days is used, it actually is a misrepresentation of the correct population be assessed. It is still a remarkable trend but not comparable to national benchmarking.

Anonymous said...

See comments well above where we discuss this.