Thursday, 26 February 2015

S&M: Fifty shades of scaremongering

Janette Sherman and Joseph Mangano (S&M) are at it again, trying to scare people about radioactivity from Fukushima. It is worth pointing out immediately that:
  • Their claims are published in the magazine Counterpunch, not in a scientific journal.
  • Once again, their claims are based on embarrassingly low quality handling of health statistics that is publicly available.
  • S&M know very well that many people will get scared about their claims, and that very few of them will be able to check the numbers themselves.
  • As usual they are not really claiming anything, they just show the numbers and ask an open question if there can be any correlation with radioactivity. This is very convenient, they can portray themselves as heroes by showing some data, and if proven wrong they can always defend themselves by saying that they are just "asking questions".
  • There is no reason to believe anything they say, as this again turns out to be a lousy cherry-picking exercise.

Figure 1. The article by S&M in Counterpunch, February 19, 2015.

So what do they say this time?

There was an article in the newspaper Alaska News Dispatch (February 14, 2015) regarding a recent increase in numbers of infant deaths. Every such case is a tragedy, and many of the deaths can be avoided if the parents know some basic information. Unfortunately, many of the cases occur in low income families where in some cases the use of drugs, tobacco and alcohol play a role. Other risk factors are letting the baby sleep on a sofa or together with the parent, suffocation may occur.

The number of infant deaths has slowly decreased in Alaska, as in the rest of the U.S., but recently there has been a reversal of the trend. Health officials in Alaska are investigating the individual cases and are trying to reach out better with information.

S&M were quickly on to this story, and only five days later, on February 19, they had an article published in the political newsletter Counterpunch where they come up with a different explanation: Fukushima did it!

They write:
"Infant mortality in Alaska has been falling for years, however 122 infants died in 2012-2013, compared to 85 deaths two years before. 
Research of causes of this highly unexpected increase is needed, and consideration should be given to the arrival of radioactive fallout from Fukushima after the 2011 meltdown. Radiation levels were highest in Alaska, Hawaii and the Pacific west coast . 
Since we know the un-born and young are at greater risk from exposure to nuclear radiation, effects that have been documented since the Marshall Islands nuclear tests, x-rays of pregnant women, and the Chernobyl catastrophe of 1986.
According to the CDC, infant (<1 year) deaths in Alaska have been falling steadily, but increaseds (sic) after 2011: 
       2010-2011    390.82 per 100,000 births(86 deaths) 
       2012-2013    533.66 per 100,000 births (122 deaths) 
 This is a 37% increase in the rate per 100,000"

We ignore for the moment the rest of the article, where they make some confused sentences about radiation measurements in Alaska for 2010 and 2011, followed by a mix of references to Chernobyl in order to support their "hypothesis". Let us instead check the numbers they use on infant deaths.

Checking the numbers

S&M refer to data from CDC, i.e. Centers for Disease Control and Prevention, that is an entity of the U.S. Department of Health. From their WONDER data base we can extract the data used by S&M in order to check them.

Figure 2. The data used by S&M. Inserted is also an extract from their article in Counterpunch.

In Figure 2 there are data extracted from CDC for the years 2010-2013 for the number of infant deaths and converted to infant death rate, i.e. the number of infant deaths per 100,000 born children. The error bars reflect the statistical uncertainty, any variation from one year to the next that is within the limits of the error bars can usually be explained as being due to random effects. But as seen there is indeed an increasing trend over the four years, just as S&M says, and when taking the difference between the average of 2012-2013 compared with 2010-2011 we do see the claimed 37 percent increase. So, are there any reasons to be concerned that this drastic increase could be due to radioactivity from Fukushima? No. First we need to look at a broader picture, something that S&M are keen to not do as it tends to ruin their story.

Figure 3. The data used by S&M for 2010-2013 (red), together with the data for the time period 1990-2013 (blue).

Figure 3 shows again the data used by S&M for 2010-2013 (red), but added are data for the time period 1990-2013 (blue). What we see is:
  • That there is a long term decrease in the infant death rate.
  • That there can be relatively large differences between two years, although the data tend to decrease along the general trend, indicated by the dashed blue line.
  • That the drastic increase in 2012-2013 is relatively large, but more remarkable is the drastic decrease in 2010-2011. In other words, in 2010-2011 there was a dip in the data compared with 2009.
So the infant deaths in 2013 is in some sense back to the "normal" trend. Of course everybody would like it to stay at the low level reached in 2010, but there is no reason to accept what S&M want us to believe, that there is an abnormal increase in the number of infant deaths and that radioactivity from Fukushima could be behind it.

"But wait", you may say, "Alaska is closer to Japan, so there is probably an increase compared to the rest of the U.S". CDC will give us the answer.

Figure 4. Data for Alaska,compared with the entire U.S.

Figure 4 shows the same data as before, but added is the data for all of U.S. (grey line). The general trend line is also shown (dashed grey line). Now we see that the infant death rate in Alaska is actually the same, or lower, than in the rest of the U.S. Any attempt to make us believe that the increase in 2012-2013 is abnormally high and could be due to Fukushima is simply not true.

What about the radiation levels?

S&M mention some radiation levels:
Few data exist, but CDC did collect gross beta in air (picocuries per cubic meter).  The period March 15 to April 30 in 2011 was the peak period when Fukushima fallout entered the environment.
For Anchorage AK, the levels are:
       March 15 to April 30, 2010 (14 measurements) .0029 pCi/m3 
       March 15 to April 30, 2011 (13 measurements) .0113 pCi/m3 
Dividing .0113 / .0029 and you get a ratio 3.86 times higher in 2011. 
The 2011/2010 ratio for the rest of the year was 0.79 (2010 was actually higher than 2011).

Not very coherent, but a few things are worth noting:
  • CDC do not perform any radiation measurements themselves. They have very nice information pages about radiation for different situations, and a dedicated page for the events in Japan. The data referred to by S&M comes from RadNet, which is handled by the U.S. Environmental Protection Agency (EPA). S&M should know this and be able to write that, at least if they want to portray themselves as being serious in this business.
  • S&M give no numbers for 2012, which would be of relevance in order to assess any correlation with infant deaths in 2013. This is based on the reasoning that any radioactivity could a affect the health of an infant from the moment of fertilization until 1 year of age.
  • To give the data for the time period 15 March - 30 April is quite meaningless if they are meant to indicate any effect when the number of infant deaths are counted over the whole year. 
  • The radiation levels mentioned are ridiculously low. The average value for 15 March - 30 April 2011 of 0.0113 picoCurie means that there is one decay in a cubic meter of air once every 40 minutes. 
    • The probability of that decay occurring in an adult with good lung capacity (about 5 liters) is once every 5 days. Meanwhile the adult has had about 3 billion beta decays within the body from the natural decays of Carbon-14 and Potassium-40.
    • For an infant below 1 year we can assume (with some exaggeration) a lung capacity of 0.5 liter, thus there is almost two months between each decay occurring within the body. If we assume a low body weight of 2 kg then there will be about a billion of beta decays within the body from natural decays during that time. One extra decay per billion normal decays, not very likely that it is the culprit.
    • These two examples for adults and infants are somewhat simplified, we can reason about the lungs acting as filters where particles inhaled will stop there, thus enhancing the probability of decays occurring within the body, but the numbers above indicate that the extra dose should be negligible.
Let us in any case assume that it would be worth looking closer at the radiation data from RadNet, then some clear patterns should occur. Figure 5 below shows the recorded data for gross beta decay since 2009 in Anchorage, Alaska (blue dots). Is there any support in data for S&M's reasoning?
  • If the values for the time period 15 March - 30 April (red bars) were of any relevance, then it is indeed feasible to expect an increase of infant deaths in 2012 due to the peak seen in 2011 (this is the peak directly correlated to the releases in Fukushima). YES
  • But then we would also expect a large peak in 2012 since the number of infant deaths in 2013 were even higher than in 2012. NO
  • Furthermore, the recorded data for the same time period in 2009 are on average almost as high as in 2011, contrary to what would be expected since the number of infant deaths in 2010 were at an all time low. NO

Figure 5. Gross beta decay (blue dots) from RadNet since 2009. Yearly averages from the recorded data are given as green lines. The averages for the time period 15 March - 30 April are given for each year as red bars.

How about giving S&M a final chance, by looking at the average values for each year (green lines)?
  • We have the largest average values in 2009. Admittedly the data set is only for part of the year, but it would be way lower if one should be able to correlate it to the low number of infant deaths in 2010. NO
  • The 2010 data are low, supporting a hypothesis of few infant deaths in 2011. YES
  • The yearly average value is almost constant from 2010 to 2013, thus no support for the increased numbers of infant deaths in 2012 and 2013. NO
  • S&M seem to have calculated the 2011/2010 ratio backwards, 2011 had on average higher beta activity than 2010, while they claim the opposite. The mathematical operation of dividing one number with another seems to have been challenge. So by doing the math right one could actually give them some support here, but it is indeed a very weak one considering the lack of other correlations. Therefore: NO
So two out of seven is not very impressive. But who am I to complain, they are just asking questions, right?


Once again S&M have shown that they are not worth listening to. Based on earlier experiences it would be naive to expect them to admit the errors and withdraw their reasoning. Instead there will probably be an extended study, being accepted in some less than serious journal, and their press releases will scare even more people. For no reason at all, except for that they make a living out of it, and that they say what some people want to hear. Janette Sherman's description in the bottom of the Counterpunch article "Her primary interest is in the prevention of illness through public education" is an insult to the word "education". And how Joseph Mangano can hold a MPH (Master of Public Health) is also a mystery. Masters of scaremongering they can claim in any case, they have explored almost every possible shade of it.

In Sweden we have at least one member of the present government who have referred to some of the earlier studies by S&M when arguing in Parliament against nuclear power. That person has not repeated it for a couple of years, being a minister in government maybe forces him to act in a more responsible way than before. Or was his earlier (and enthusiastic) referral to S&M based on ignorance and he is better informed now? Let's hope for that, running a country can be done at least partly with a given opinion as compass, but it is good if that opinion is based on proper numbers, not fake ones.

/Mattias Lantz

Other scrutinies of S&M

Here we go...

Time to start a blog of my own (10 000 years after everybody else...) in order to make things available that do not fit into other places. Focus will be on numbers, in particular about when people misuse numbers in order to give a certain impression, or in order to affect people's opinions.
But I will probably also share some opinions of my own, not necessarily backed by numbers...