One of the best ways to deal with the problem of antibiotic resistance is infection control in hospitals. That is, if patients aren't getting infections, whether those be sensitive or resistant to antibiotics, then there's less of a problem. Massachusetts has released its annual report on hospital infection rates. The report currently looks at two major causes of hospital acquired infections: central venous catheter blood stream infections (CLABSIs) and surgical site infections (SSIs).
The good news is that, compared to the nation as a whole, MA does no worse or even better than the national average. One issue is that many specific types of infections (catheter infections in pediatric ICUs) are very rare, so there's limited power of test. We'll just have to bump up the sample sizes by infecting more people.
The other interesting thing to note is that the report is also broken down by individual hospital. Here too, sample size is an issue, especially for smaller hospitals, although, thankfully, no hospital seems to be doing worse than the national average (which given the sample size would be a disaster). And this has policy implications.
Betsy McCaughey (yes, the 'death panel' Betsy McCaughey) has argued that, to control hospital acquired infections, we need hospital reporting cards, and consumer choice will solve the problem (as opposed to TEH SOCIALISMZ!! which would involve regulatory penalties). Leaving aside the problems of controlling for patient populations, the Massachusetts data show that the report card idea is pretty useless due to statistical power problems.
If a hospital is having repeated problems, 'consumer choice' is not the solution. Forcing the hospital to fix these problems is.
- Log in to post comments
Yep, because I make an informed decision as to which hospital I go to whenever I end up in the emergency room. Health care is just like any other commodity, so applying free market logic to it makes perfect sense.
Not trying to defend the stupid report card idea here, but wouldn't a system of regulatory penalties suffer from exactly the same statistical power problem?
@Phillip
I agree if the hospitals use these somewhat brutish measures of straight up rate of infection they'll suffer the same problem.
Before all of this "nosocomial infection is the only real measure of quality," people used to think that death rate was king. Some body regularly published the "death list" ranking hospitals by death rate.
The problem was that there wasn't any consistancy, hospitals swung wildly from one year to another, precisely because of the small sample sizes. I suspect most of these gross measures are going to have the same problem, and you won't get a realistic picture except over time. And even in that case, only for the very best and very worst hospitals, the ones that stay close to the top and stay close to the bottom year after year.
I tend to think hospital report cards would still be effective by giving hospitals an incentive to lower their infection rates. Even if differences between hospitals are small, a hospital would have an incentive to score better than local rivals because patients are sensitive to rankings even if the differences aren't statistically significant. And while small sample sizes may make much of the year to year variation in ranks random, the incentive to improve will be constant.
WcT, your engineering background is showing.
This is a problem in statistical quality control, and there are means well-known to those of moderate skill in the art for solving the problem.
Quality principles
#1: nonquality always costs more than quality.
#2: If nonquality seems to cost less than quality, check your accounting.
#3: Quality problems are the result of processes, not people.
If your quality system depends on the skill of the craftsmen involved, you don't have a quality system because you don't know what actually results in quality and what doesn't.
In the case of hospitals, let's concede that the monitoring outcomes on a per-hospital (or more practically on a per-unit) level is too noisy to be useful. That's OK, though, because it is possible to monitor the factors that affect the outcomes across the system. Things like "always wear gloves when with a patient and never wear gloves when not with a patient," which ensures that you're protecting the patients and not just yourself.
Once the factors that affect quality are identified and quantified, it becomes possible to rate providers on their practices instead of their outcomes -- and practices are much easier to monitor. Unless, of course, you're allergic to "regulation."
I hope that this addresses Phillip's concern.
Quality assurance practices need to be established with the attitude that the whole point is to prevent problems from arising in the first place.
For example, health departments don't give reports on restaurants based on illnesses per restaurant, but rather cleanliness, food temperatures, and so forth. Things that, if reasonably well done, ought to significantly reduce the number of potential illnesses.
The Massachusetts report cited above has a statistic for a standardized infection ratio, given as the actual number of infections divided by the expected number of infections. It also gives a table of the beds per full time equivalent (FTE) of infection preventionists by bed size.
I'm not convinced that they are getting at the root of the problem.
Excellent comparison.
D. C. Sessions @ #5:
Absolutely, thanks a lot.
Jason Transcriptions--Serves Quality Transcriptions Needs.