Can We Find A Trend In The Fungal Meningitis Outbreak?

Epidemiologists have some of the most intriguing jobs in the world. No other branch of science can make an immediate impact akin to that of the study of diseases. The fruit of diligent research can be life- no, world-changing. And when an epidemic strikes, their fast actions can save thousands of lives.

Kudos to the epidemiologists on the case of the recent outbreak of fungal meningitis. I heard about the outbreak, and then one day later they had already located the cause and were doing work to minimize the damage done by the tainted steroids. Unfortunately, they can’t save every life, and nearly 300 cases have been reported with a fatality rate of about 8%. However, their quick work surely saved many more lives.

For my latest post, I’ve decided to play epidemiologist to try an isolate a trend among the data for the meningitis outbreak. Is there a reason that certain states have been hit harder than others? (Besides, of course, the states that haven’t received the infected drugs, and obvious comparisons like population)

_______________________________________________________________________________________________

The CDC website doesn’t have information on former outbreaks of fungal meningitis (or if they do, they’re hiding it very well), and this made my first idea- does this outbreak parallel previous outbreaks?- very short-lived. However, the CDC did have interesting statistics on another topic, which redirected my plan for this post.

The fungal meningitis outbreak has apparently been classified as a “Healthcare-Associated Infection” or “HAI”. The CDC tracks HAIs in a variety of ways; one of them is an SIR value: Standardized Infection Ratio. This value is found by taking the number of actual HAIs and dividing it by a predicted number of HAIs. Lower numbers are better, and values under 0.5 are very good. Similarly, values over 1.0 are very bad- this means that the included facilities are actually causing more infections than they’re projected to.

Here is a chart of the SIR values (in 2010, the most recent I could find) for all of the states where the tainted steroids have been sent:

 These values are mostly good; Indiana is the only state with a value above 1, and Michigan and West Virginia both have values under 0.5 (Remember that!)

You may have noticed that three states- Idaho, Minnesota, and Rhode Island- don’t have SIR values on the graph. This is because SIR values are independently submitted by health care centers, and some states don’t have enough centers submitting information to the CDC for effective calculation of SIR. These three states are some examples- less than five centers submitted information over 2010, whereas most states have several dozen.

Since the outbreak is a HAI, it would be reasonable to assume that most of the infections occurred in states with high SIR values. But that isn’t the case:

STATE	CASES
ILLINOIS	1
NEW YORK	1
IDAHO	1
PENNSYLVANIA	1
TEXAS	1
NORTH CAROLINA	2
MINNESOTA	7
FLORIDA	17
MARYLAND	16
NEW JERSEY	16
OHIO	11
NEW HAMPSHIRE	10
INDIANA	38
MICHIGAN	53
VIRGINIA	41
TENNESSEE	69
CALIFORNIA	0
CONNECTICUT	0
GEORGIA	0
NEVADA	0
RHODE ISLAND	0
SOUTH CAROLINA	0
WEST VIRGINIA	0
TOTAL	285

Indiana, which had the highest SIR value, has quite a few cases compared to other states. Michigan, however, had the lowest SIR value- and has more cases than any state except Tennessee. But West Virginia, which had the second-lowest SIR value, has zero cases.

Why is there variability in the data? One reason is because not every state received the same amount of the infected drug. Only one facility in West Virginia received the drug, compared to six in Indiana. Based on the data, each facility that received a shipment of the steroid had about 3.8 infections. From this average, we can predict how many cases will occur in each state:

This isn’t very good. We can take it our prediction one step further by applying a state’s SIR to the predicted number of cases (for example, Illinois: 11 predicted cases x 0.678= 8 predicted cases with SIR):

 

There’s still no strong correlation here between SIR and the number of cases. We can calculate our own SIR values for these states using our predicted number of cases and the number of actual cases. Unfortunately, when we do this, only four states- the ones in blue on the data table and the following graph- have SIR values that fall within the standard range of scores (that is, the range of scores for all 50 states.):

M-SIR represents my own calculated SIR value based solely on the meningitis statistics

Quite simply, there’s no correlation between HAI SIR values and the recent fungal meningitis outbreak. The only explanation I can come up with for this is that most facilities used the tainted steroid believing it to be safe, whereas with most HAIs the healthcare center should know how to avoid the problem.

My data:

https://docs.google.com/spreadsheet/ccc?key=0AnZrkjWWJajQdDVETDlPQ0dTcUFXVU1TXzcxS2h1TFE

Note: All data as of October 21, 2012