Replies

I posted an analysis in 2002 about apparent differences between the anthrax in Florida as compared with the anthrax in the Northeast. [This doesn't address the issue of whether there were differences between the NY anthrax attack and the DC anthrax attacks.]

I think this analysis might still be of interest today. Below are copies of a couple of posts about this. (I think there were earlier postings, but these are probably the clearest and most complete.)

A summary of the differences in symptoms in Florida vs. the Northeast. Originally posted on 5/12/2002, http://www.freerepublic.com/focus/news/680790/posts?page=73#73

There do appear to be differences in symptomatology among the various anthrax mailings. It's too early to say whether these differences are caused by genetic differences in the anthrax, different physical preparation of the powder, or some other differences in the patients or the environment.

There are three differences in symptoms, of which two appear to require further explanation:

Compare the long-term symptoms of the people who contracted inhalation anthrax and survived. The postal workers in the Northeast all continue to suffer from symptoms similar to Gulf War Syndrome, while the patient in Florida is back to his old self.
Compare the people who contracted anthrax after being exposed directly via letters opened in their place of business (unrecognized by the recipient at the time). In Florida, the only cases were of inhalation anthrax. In New York, the only cases were of cutaneous anthrax. (The Washington letters can't be compared in this way. The Daschle letter was caught upon opening, and everyone was given immediate treatment. The Leahy letter was caught by the authorities before delivery.)

[The third difference appears to be easily explainable: the number of cross-contamination cases and postal worker infections in the Northeast (many inhalation anthrax) vs. none in Florida. But this can probably be explained by the different machines in use in the Trenton post office, which squeezed powder out of the envelopes there and sent it flying.]

Differences 1 and 2 above may not be statistically significant, due to very small sample sizes (especially in Florida); but they are suggestive of a difference.

If I had to guess, I'd say that difference 1 would appear to be due to a genetic difference between the FL anthrax and the anthrax distributed in the Northeast. It could also be due to some other difference (a chemical agent added to the NY and/or DC anthrax, or some other aspect of the physical preparation). (Or maybe it's just due to chance. Perhaps Blanco in FL was unusually hardy. But he's quite old, which makes me doubt that it's just chance in this fashion.)

Difference 2 is strange. The same bacterium causes both inhalation and cutaneous anthrax; the difference is just the site of infection. My first inclination was to say that this difference indicates a difference in physical preparation (after all, the whole point of "weaponization" is to make the particles small enough to lodge in the lungs, as well as to make them free of electrical charge so they'll move around easily) or in delivery method. But the delivery methods were apparently the same. And what kind of physical preparation could prevent cutaneous anthrax cases from arising at AMI? The building was heavily contaminated, after all.

So I'm not sure what to make of difference 2. Maybe it indicates a genetic difference as well? There could be different genetic propensities for the bacteria to do differentially better or worse at different infection sites.

A preliminary statistical analysis indicating that the symptomatic differences probably aren't due to chance, but this isn't definitive due to the small sample size. Originally posted on 4/24/2002, http://www.freerepublic.com/focus/news/671281/posts?page=13#13 and http://www.freerepublic.com/focus/news/671281/posts?page=14#14

Let's analyze this statistically. Take as the null hypothesis the statement that there is no difference between the FL and NJ anthrax. Of the 6 inhalation anthrax survivors, 1 refuses to be interviewed. So, of the 5 we know about, 1 (Blanco) has fully recovered, and 4 are experiencing the syndrome of symptoms described in the article (memory loss, fatigue, joint pain). Assuming the null hypothesis now, all patients were exposed to the same anthrax; this anthrax would then appear to cause this syndrome among about 80% of the survivors (due to genetic differences among patients or other factors). The fully-recovered patient could equally likely have been any one of the five (by the null hypothesis). The probability that the fully-recovered patient (we would expect to have one) would be the patient from FL is 1 out of 5, or 20%.

So I conclude that the probability that this is due to chance (rather than to some difference between the FL and NJ anthrax) is 20%. In statistical terms, we can say that there is a difference between the FL and NJ anthrax at the p=0.8 confidence level. This is not statistically significant (because of the small sample size), but it's high enough to be suggestive. It does add to the weight of other evidence that there is a difference.

In terms of other measures, notice that there doesn't appear to be a difference in mortality rate between the FL anthrax and the NJ anthrax, among the people who contracted the inhalation form of the disease. Fatalities numbered 1 out of 2 cases in FL, and 4 out of 9 cases in the Northeast.

On the other hand, there does appear to be a difference in the site of infection. There were many cases of cutaneous anthrax in the Northeast, but none in FL. This suggests a difference in physical preparation or delivery method, but it's conceivable that a subtle genetic difference could make one type more virulent at a particular site than another.

There's one minor correction to make here. I wrote: ...at the p=0.8 confidence level.

I meant, of course, the 80% confidence level, or, equivalently, p=0.2. (Typically, people would like to see a 95% confidence level, or p <= 0.05, to call something statistically significant.)

This p value of 0.2 is based just on the differential rate of full recovery among survivors in the two populations (FL vs. the Northeast). I suspect that if you also include the difference in cutaneous anthrax incidence rate, the difference may prove to be statistically significant.