You may only read this site if you've purchased Our Kampf from Amazon or Powell's or me
• • •
"Mike and Jon, Jon and Mike—I've known them both for years, and, clearly, one of them is very funny. As for the other: truly one of the great hangers-on of our time."—Steve Bodow, head writer, The Daily Show

"Who can really judge what's funny? If humor is a subjective medium, then can there be something that is really and truly hilarious? Me. This book."—Daniel Handler, author, Adverbs, and personal representative of Lemony Snicket

"The good news: I thought Our Kampf was consistently hilarious. The bad news: I’m the guy who wrote Monkeybone."—Sam Hamm, screenwriter, Batman, Batman Returns, and Homecoming

March 22, 2011

Infants Died at TMI

By: Aaron Datesman

I have stolen the title of this post from Harvey Wasserman and Norman Solomon’s excellent book, Killing Our Own. I am about to steal their data, too.

Before I engage in this theft, however, I would like to explain again why this historical topic engages my interest at a time when an incredible disaster is unfolding in Japan. Because the historical consensus is that nobody died from the TMI accident, and that as a consequence it wasn’t really a big deal, in my opinion we are failing to understand what the median health effects will be for the population of Japan due to the much larger disaster unfolding there.

TMI killed a lot of people, some miles or some years down the road, and caused increased illness and poor health for many tens or hundreds of thousands more. But in the natural variability of living, health, and time, that signal is difficult to discern. (It scarcely helps that we generally refuse to look for it.) Perhaps this fact indicates that the harm from nuclear power, partial meltdowns and all, is preferable to the harms we also suffer due to other technological choices we make, such as burning coal. George Monbiot makes that argument here. I may have a comment on that idea once I finish what I have to say about the health effects of the TMI accident. I think it’s interesting; but for now, it doesn’t engage me.

This is the data on the infant death rate in the area surrounding Three Mile Island compiled by Wasserman and Solomon from the records of the Pennsylvania Department of Health. The PA State Capitol Building in Harrisburg is a bit more than ten miles north (and a bit west) of Three Mile Island, right along the Susquehanna River. The accident at TMI-2 took place at the end of March, 1979, at the very end of the Winter 1979 time period displayed in the chart. Therefore the Spring 1979 row covers the first months after the accident.

Infants Died at TMI 1.jpg

It is difficult to make heads or tails of this data. The sample size is not very large, for one thing. In 1980, the population of Harrisburg was only 53,000, so there are probably in the neighborhood of 300-500 births represented in each quarter. (It is possibly 2-5x more. I don’t know how many people lived inside of the ten-mile radius but not in Harrisburg.) Therefore, it is difficult to view the statistics as robust. Additionally, there is a strong seasonal variation to the death rate in Harrisburg, which confounds interpretation of the data.

However, there is at least one clear signal in the data presented. In Spring 1979, immediately following the TMI accident, the infant death rate in Harrisburg was 29.7 per 1000 live births. In the comparable period from 1978, it was 11.5 per 1000; and 8.1 per 1000 in 1977.

The infant mortality statistics refer to deaths in the first year of life. Neonatal mortality is a subset of infant mortality, referring to deaths in the first month of life. This chart provides the neonatal mortality statistics.

Infants Died at TMI 2.jpg

You will note the number in the Spring, 1979 line in Harrisburg: 29.7 per 1000. It is identical to the infant death rate. Wasserman and Solomon explain why this is:

And in those tragic three months after the TMI accident every Harrisburg baby listed as an infant-mortality statistic had in fact died in the first twenty-eight days of life.

What would the results of fallout from a severe nuclear accident descending upon a moderate-sized American city actually be? Dead babies seem a likely one. If you can detach yourself from the conventional wisdom that there were no ill effects due to the accident at Three Mile Island, I think the data show rather clearly that there was a radiological disaster in Harrisburg in the spring of 1979.

If you have ever been to Harrisburg - especially if you’ve driven through - you might note that the Susquehanna bridges are pretty high. With a topographic map, I think one might predict that the likeliest place for TMI fallout to land is also the regional population center just upriver. But to make this connection one must first embrace the basic reality that emissions from nuclear power plants are not magically and immediately diluted in the full volume of the atmosphere once they leave the vent stack at the plant. Unfortunately, this is a reality that many nuclear experts have yet to embrace.

Perhaps no one died as a result of the accident at Three Mile Island. For myself, I think we would come to another conclusion altogether if we really opened our eyes and took the time to look.

— Aaron Datesman

Posted at March 22, 2011 09:27 PM

Very, very interesting stuff, Aaron. While the spike in infant death can't be pegged directly to TMI on its own...we must ask ourselves, what else would account for it? And the cancer in farm animals? And a farmer who saw blue? And the presence of fine, gray ash nearby? And, and, and...

Posted by: Bill Coffin at March 22, 2011 10:03 PM

Thanks for presenting this data.

I should very much like to see a rigorous analysis of the statistical significance of the spring and summer 1979 infant mortality numbers.

Posted by: joel hanes at March 22, 2011 10:18 PM

As to implausibly obtuse scientists, the words of Upton Sinclaire commonly quoted at ATR apply: "It is difficult to get a man to understand something when his salary depends upon his not understanding it!"

And if that seems not enough, in Professor Sisela Bok's book Lying: Moral Choice in Public and Private Life, at the front of the introduction she quotes Francis Bacon's "Of Truth":

"Doth any man doubt, that if there were taken out of men's minds vain opinions, flattering hopes, false valuations, imaginations as one would, and the like, but it would leave the minds of a number of men poor shrunken things, full of melancholy and indisposition, and unpleasing to themselves?"

Facing facts is easier said than done, and not simply because, as everyone knows, the truth hurts. To steal the physicists' jargon, since that's the coin of the realm today, the truth is definitely not a noble gas, and its unusual buoyancy makes it dangerously unpredictable. Not always odorless but often tasteless, the truth commonly dissipates without being recognized, and when diffusion, convection, and blasts of hot wind disperse it, no one easily can discern what has become of it. What's more, the truth is combustible, perhaps even radioactive, and when it doesn’t disperse it either explodes or consumes itself, leaving uncounted casualties behind. Even among the nerdy and scientifically inclined, the instinct for self-preservation is strong, so those who tend to come in closest proximity to the truth most often sense its dangers far more than they will ever admit, especially to themselves, lest they learn firsthand what Francis Bacon meant.

Posted by: N E at March 22, 2011 11:50 PM

60 minutes did a program with Noam Chomsky entitled Fear in the Newsroom and they discussed the Three Mile Island incident. They interviewed the reporter who had reported for the News Hour and the reporter said he was shocked when he watched the broadcast, the News Hour had deleted half the story in order to make the people protesting the TMI look like nuts when in fact they really did have something to worry about.

Posted by: Rob Payne at March 23, 2011 02:31 AM

should we be angry at Monbiot about this? I called him a cunt in my rage against the Libya invasion earlier.

Posted by: Jenny at March 23, 2011 03:21 AM

You say,

"Additionally, there is a strong seasonal variation to the death rate in Harrisburg, which confounds interpretation of the data."

I agree that it confounds interpretation. Having said that, however, you attempt to interpret it anyway.

It's as if you're convinced that an effect of this sort occurred, and you're determined to interpret data in a way which conforms to your prejudices.

Perhaps there was such an effect, but I'd like to see some evidence which does not "confound interpretation".

Posted by: James Cranch at March 23, 2011 05:01 AM

James Cranch's point demonstrates part of a 'damned if you do, damneded if you don't" problem that isn't his fault.

If someone pursues the acquisition of evidence about radiation ambivalently, he'll fail, because there will be significant obstacles along the way--like those encountered by the News Hour reporter 60 minutes interviewed referred to in the comment above. There is a reason the data is insufficient--very powerful interests do not want such research to undermine either nuclear energy or nuclear weaponry.

On the other hand, if a journalist or scientist says that the paucity of evidence is alarming, and says it with a real commitment to its importance, he ends up looking like he has no objectivity--i.e., that he is "determined to interpret the data in a way that conforms to [his] prejudices." After all, it's not objective to assert that underdeveloped data supports an alarming theory.

Even putting aside corruption, among those with responsibility for running things like the government and military, theories with significant implications are only accepted reluctantly, if at all, after careful examination of substantial evidence, because far-reaching theories upset the apple cart with their implications and damage or destroy whole industries and important national interests. If a theory is wrong, then to permit that damage would be wrong--and foolish and costly. So people who propound far-reaching theories without substantial evidence are usually mistrusted if not totally ignored.

If proponents of these sorts of disfavored theories persist in their lack of objectivity, the words used to dismiss them turn into euphemisms. Implying that a scientist lacks objectivity already does plenty of damage, but it's only halfway to crazy.

Posted by: N E at March 23, 2011 06:51 AM

NE - I appreciate that sentiment, but you're overthinking it.

James Cranch, have you ever listened to a staticky radio? Huh?

You can evaluate a data set in the presence of confounding influences. In fact, there's generally no other way to evaluate data in the medical and human health fields. You identify the confounding influence, develop a means to account for it or remove it, and then look for the signal.

In this case, there appears to be a strong seasonal influence on the infant mortality rate in Harrisburg. Perhaps this occurs because infants born in fall and winter generally suffer higher mortality (though the state-level data doesn't really support that hypothesis). Perhaps it's because many of the residents of Harrisburg are poor (which is true), they can't afford to heat their homes very well, and there's a chain onward which results in higher infant mortality. Maybe somebody knows this answer. I do not, although I have a separate hypothesis.

There is clearly a way to remove the confounding influence for this data: COMPARE YEAR-ON-YEAR IN THE SAME SEASON. Corporations benchmark their profits this way. Why does it not suffice for infants? The infant mortality rate in Harrisburg about tripled y-o-y. The rates for spring 1977 and 1978 were around steady, statistically speaking.

This is a strong signal. It's not proof that TMI killed hundreds of people (or anybody), but it's absolutely an indication of a problem.

Look, I used to be a proponent of nuclear power. I worked for the Nuclear Navy, for Pete's sake! You can believe that I see a signal in this data because I want to believe that the TMI accident killed people, but it's the other way around. I believe that the TMI accident killed people (and harmed, and is harming, a lot of people) because I'm aware of data like this - and because I've worked to understand it.

You have a Ph.D. in mathematics, apparently. Please do the same.

Posted by: Aaron Datesman at March 23, 2011 09:35 AM

I have NOT READ the whole chapter of a book that gives the following data in its 19th chapter ( the whole book is online......"Secret Fallout".... you can go to it when you open the link ). I thought,the data would interest everyone who has been following these posts. If I am repeating something that has already been posted, please ignore it.

"Pennsylvania and United States Infant Mortality (0-1 Year old at death) 1979,
With corrections in July and August* (Data from the U.S. Monthly Vital Statistics)"


Posted by: Rupa Shah at March 23, 2011 11:56 AM

I see that the link is not working. I will try again or please copy and paste the URL above.


Posted by: Rupa Shah at March 23, 2011 12:18 PM

I was thinking about Strontium and that the people around there were talking about it in the milk for Hershey's Chocolate. So I looked it up and see it burns red. I came to the conclusion WE'll never figure out the right story. GE SOLD the old Bechtel Leakers and THAT'S who puts out the "official" story. TRUTH be damned, its the first casualty in anything GE. But thanks to GE's persuasive sales force WE ARE privy to an IDEAL study of these kind of events in, NOT 1, not 2, not 3, not 4, not 5, but 6 reactors in various states of meltdown and the attending nuclear byproducts release "into the wild", so to speak.

Posted by: Mike Meyer at March 23, 2011 02:14 PM

yer baby is perfectly happy
'e won't need a baff anymore
e's muckin' about wif the angels above
not lost, but gone before

Posted by: mistah charley, ph.d. at March 23, 2011 02:47 PM

The rates for spring 1977 and 1978 were around steady, statistically speaking.

But what about winter 1979? I don't find it implausible that the disaster could have increased infant deaths, but the three months before the disaster also seem to show a similar pattern in Harrisburg (neonate death rate higher than previous years, and all infant deaths were neonates), just not as pronounced as spring 1979.

Posted by: t.m. at March 23, 2011 04:39 PM

I have long thought that all nuclear reactors should be built next to State Capital buildings.

Posted by: Bill Jones at March 23, 2011 06:06 PM

tm, that's a pretty sharp observation. It's a smaller signal though, isn't it? 50% vs. 3x. So I would judge it to be less reliable - that is, it's more likely to be statistical noise.

On the other hand, in the past I've seen suggestions that the TMI-2 plant was always troubled, and that there were unauthorized emissions prior to the accident. I can't verify or refute that assertion however.

Posted by: Aaron Datesman at March 23, 2011 06:39 PM

And what's the death rate from coal plants? Wait, that's just global warming, surly THAT won't kill anyone.

Posted by: Joe at March 23, 2011 08:14 PM

Yes - I noticed after commenting that Wasserman and Solomon attribute that elevated winter 1979 death rate to the opening of TMI-2. But that seems to weaken or at least complicate the case that there's a clear signal from the accident itself. Also, I'm not an expert, but wouldn't you expect a longer time lag for most infant deaths?

Posted by: t.m. at March 23, 2011 08:20 PM

Joe: Good time to invest in WIND and all the other sustainables. Use less.

Posted by: Mike Meyer at March 23, 2011 09:57 PM


Right, so let's pretend radiation doesn't kill people. That'll help a lot.

Posted by: N E at March 23, 2011 10:17 PM

@Aaron have you seen this? It's an age pyramid for York County, Penn. from the 2000 census data. True, it doesn't directly address the death rate, but it does indirectly address it. It doesn't look like a pyramid to me. It looks like an arrowhead. A war point, to be exact. The war point is designed to be difficult to remove. A hunting point is designed to be removed easily.,_Pennsylvania_age_pyramid.svg

Posted by: Some guy on the innernet at March 24, 2011 12:11 AM

@Some guy, I agree that's a menacing-looking graph. But I think all it shows is that young people starting their careers have to leave York County to find work elsewhere.

Now if the war point were *glowing*, then, we would have something.....

Posted by: Aaron Datesman at March 24, 2011 08:18 AM

If there were 500 births, then changing infant mortality from 1% to 3% would kill 10 infants, right? So we're talking about numbers that are probably at the edge of statistical significance (although still significant to the people involved). I think to say something more precise, you'll want to look at the cause-of-death data, which exists at least on a county level. Email me if you want the data and have trouble finding it.

Also, I think there's been extensive study of TMI health effects. Wikipedia ( ) gives an overview of some of the larger studies, and there have also been many many epidemiology papers that do what this book does and mine existing data. I don't understand why you say that we haven't "opened our eyes and taken the time to look" at these health effects.

Finally, some of this discussion seems to suggest that people are trying not to pay attention to the harms of nuclear power. While perhaps some people have an incentive to downplay the harms of nuclear power, public health researchers have an incentive to make their papers interesting, and that means finding large, statistically significant effects. If they've failed to do so, it's probably not because they secretly wanted to fail.

Posted by: aram at March 27, 2011 01:43 AM