Wednesday, July 14, 2010

Catheter Related Central Line Infections: The Numbers (and the Truth) Behind the Story

A "preventable" "plague" that needlessly results in "30,000" deaths because the health system refuses to implement a "stunningly basic" set of preventive measures. According to impressive research, implementing them could reduce their occurrence by "two thirds."

The Disease Management Care Blog was also alarmed when it read this Washington Post article. so it decided to look into things.

The topic is "catheter related bloodstream infections." While most intravenous fluids are delivered via an arm vein, sometimes the volume or the make-up of the fluid requires a larger diameter tube (or "catheter") to be tunneled or placed in a larger diameter vein. The good news is that many times, these veins are relatively close to the surface of the neck and shoulders, a fact that hasn't been lost on the producer's of HBO's True Blood. The bad news is that large plastic catheters are also large deep foreign bodies that can enable skin bacteria to track/grow along them and easily spill into the blood stream.

The unit of measurement for these infections is based on the counterintuitive concept of "catheter days." This combines the number of large vein catheters in use with the number of days they are being used. So, if the DMCB needed a large vein catheter for ten days, that would represent "ten catheter days." If the 4th floor of Our Mother of Holy Deficit Hospital had three patients that each needed a large vein catheter, say .... one for four days, another for five days and another for six days, that would add up to 15 (4 plus 5 plus 6) "catheter days." If one of those patients develops a blood stream infection involving bacteria found on the skin, that works out to "one infection" per "15 catheter days." For ease of discussion, this can be normalized (like percents) to "100 catheter days" - so the infection rate is "6.6 per 100 catheter days." If it were "per thousand" catheter days, it'd be 66. More on this below.

Congratulations, because you can now read this free access article in the New England Journal that showed how well these infections could be reduced. If you don't want to, that's OK because your DMCB blog can expedite things by getting right to the bottom line:

When, in October of 2003, providers in 108 Michigan intensive care units agreed to: 1) hand washing prior to sticking in the catheter, 2) using "barrier precautions" (gloves, gowns and drapes) when placing it, 3) using chlorhexidine to clean the puncture skin site ahead of time, 4) avoiding use of the large veins in the groin and 5) removing the catheter as soon as it's not needed, infection rates, compared to baseline, dropped:

The overall median* rate of catheter-related bloodstream infection decreased from 2.7 (mean,7.7) infections per 1000 catheter-days at baseline to 0 (mean, 2.3) at 0 to 3 months after implementation of the study intervention and was sustained at 0 (mean, 1.4) during 18 months of follow-up. A significant decrease was observed in both teaching and nonteaching hospitals and in small hospitals (less than 200) (Bolding DMCB)

Note that the unit of measurement used was infections per thousand catheter days. That's necessary because the total number of infections at the start was relatively small: at baseline, the mean* (or "average") number of infections was 7.7 per thousand catheter days. In other words, if a hundred patients were treated with these catheters for ten days, only about 8 would get an infection. The other 92 would do fine.

An Unwieldy Unruly Hospital System?

Check out that many of the participating study hospitals had less than 200 beds. Go to this American Hospital Association web site, and you'll see that the average number of beds in U.S. community hospitals is 161. Assuming that, at any time, 20% of the patients filling these beds have large venous catheters (a huge overestimation but for illustration purposes) 32 persons over the course of the year (365 days...) would comprise 11,774 catheter days. The physicians and administrators would have to deal, based on these data, with about 90 infections. That's between one to two a week. It's probably far less.

While any infection is one too many, the DMCB thinks that rate, compared to other hospital safety concerns (they can be seen here) is easy for the average hospital administrator or physician to miss. After all, most patients coming and going from the average hospital with these catheters do just fine. The DMCB doesn't think it's money, laziness or a failure: this is a problem that defies easy detection at the individual hospital and patient level.

And that's why its possible for so many hospitals with far less catheter days to report "no" infections. They're rare to begin with.

Perfect Patient Safety Is In Reach?

Check out that the rate of infections declined from 7.7 to 2.8 (at 3 months) to 1.4 (18 months) per thousand catheter days. That's not zero: infections still happened. Using the 100 patients for 10 days analogy above, there'd still be 1-2 patients developing an infection and, instead of 92 doing fine, 98 would do so. That's better, but not perfect.

It also portrays the pernicious use of "relative" risk reduction statistics. If a drug reduced death rates from 50 per 100 cases to 25 per hundred cases, that would be a "50%" reduction. However, decreasing a death rate from 2 in 10,000 persons to 1 in 10,000 persons would also represent a 50% reduction. The "two thirds"reported in the article above is technically correct but inflates the appearance of the benefit.

Role for Government

Finally, in case readers perceive that the eternally skeptical DMCB believes that Washington DC can't do anything right, it says the government did it right this time. The Federally funded Michigan ICU study pooled the data of over a hundred hospitals because that's what it took to see if the five preventive measures described above could reduce the rate of something that happens relatively rarely at an individual patient/hospital basis. This is an instance where the taxpayers got their money's worth. The DMCB uses the term "impressive" in the opening paragraph. That's the only part that is completely and transparently true.

*This article reported "median" infection rates because the distribution of infection rates was probably skewed and medians are better measures of central tendency. Mean or average is a more familiar concept, however, so the authors are to be commended for reporting both. You can read more about the distinction here.

Image from Wikipedia

No comments: