We can't predict HAI with ICD-9 codes and it's only going to get worse

I'm getting ready for a chat with a reporter concerned with issues surrounding HAI surveillance. During my preparation, I thought again through the issues of code-based algorithms (e.g. ICD-9) and I've come to the conclusion that they are useless for assessing the burden of HAIs and HAI trends and it's only going to get worse.

One area we (and many others) have looked at is the utility of ICD-9 code-based algorithms (ie administrative codes) for detecting HAIs efficiently. A key metric frequently reported by researchers is the sensitivity of a specific code or code algorithm, which is great if the purpose of the algorithm is to improve the efficiency of detection by manual methods. Thus, if the sensitivity is high-enough, you could use the code-based algorithm to reduce the number of charts that require an IP's review. If you are using codes in this way, great!  I have no problems with that.

However, many are now using code-based algorithms to track trends in specific HAIs and measure the burden of disease. My general feeling on these is that they should be completely avoided for several reasons:

1) No matter how sensitive the algorithm is, all we care about here (since we are not validating with manual review) is the positive predictive value (i.e. the proportion of all code-positive patients that actually have the HAI of interest)

2) The PPV is very low for almost all HAI algorithms

3) If we are doing our job and lowering the incidence of HAI per admission in our hospitals the PPV by definition will only get worse (given a fixed sensitivity and specificity)

To show you why I have these concerns I have constructed two 2x2 tables evaluating an excellent hypothetical code-based algorithm for UTI with a sensitivity and specificity set at 95%.  In this first 2x2, I have evaluated the performance of the algorithm when the HAI has a 5% incidence per admission (i.e., 5% of the admissions had a UTI). You can see that such a great algorithm with a high-prevalance of disease, has a poor PPV of 50% - like flipping a coin.

Now, assume we have done an amazing job and cut our HAI rate down to 1%.  Given the same hypothetical algorithm, our PPV is now a horrible 16%. Thus, as we get better at preventing HAIs, we get worse at detecting them using code-based algorithms. Are you comfortable saying UTIs are increasing or decreasing or are associated with a certain level of excess costs, when only 16% of the UTIs in your estimation are actual (true positive) UTIs?  Me neither.

Didn't we say not to use ICD-9 codes to track MRSA?

When you're trying to improve epidemiological methods, I guess you have to be patient.  For example, when Yehuda Carmeli and Anthony Harris told us which control group to use when assessing risk factors for antibacterial resistant organisms, it took years for people to regularly follow their advice. However, I'm still a bit surprised when authors and journals keep publishing studies that track MRSA infections using ICD-9 codes. 

A couple years ago, ICHE published our multi-center validation study showing that ICD-9 codes for MRSA have a very poor positive-predictive value: 31%. To quote our conclusion: "In its current state, the ICD-9-CM code V09 is not an accurate predictor of MRSA infection and should not be used to measure rates of MRSA infection." ICHE also published Marin Schweizer and Mike Rubin's excellent editorial summarizing the issues with ICD-9 code based surveillance for MRSA.

So, knowing what we know about recent MRSA trends and considering my feelings about ICD-9 codes and MRSA, I was a bit surprised when the August ICHE included a study suggesting that MRSA was increasing in academic medical centers between 2003 and 2008 and it used ICD-9 codes! Sure, they attempted to adjust for the billing code's limited sensitivity.  Unfortunately, sensitivity isn't the key issue. If you say something is increasing, it's more important to know if what you're counting as an MRSA infection is actually an incident MRSA infection from the index admission, and not an MRSA infection from a prior admission, or even newly detected MRSA colonization. Thus positive predictive value and specificity are more important measures. 

Just remember this when you read stuff on the interweb that says "contrary to data from the CDC" or posts saying MRSA has doubled in 5 years.  Stick with the CDC. Fortunately, most of the best evidence suggests MRSA is in decline...at least for now.

Important note: Our validation study included three hospitals and compared ICD-9 to actual culture data, the gold standard.

-----
Ridiculous Example: What if we wanted to estimate the proportion of 30th Olympiad attendees who were born in England by randomly sampling people walking around the Olympic stadium. We could use a somewhat sensitive test, "is the person wearing a Union Jack t-shirt", and catch many people born in England, but this would be useless (ie very low positive predictive value) since anyone can buy a shirt. We could also use a more specific test like checking their passport. Thus, ICD-9 codes are a bit like t-shirts, while microbiological culture results as used in accurate MRSA estimates are more like a passport.