Sunday 12 April 2015

Another take on S&M - congenital malformations

It did not take a long time after the Alaska hoax, Janette Sherman and Joseph Mangano (S&M) are back again, this time with a peer reviewed article, published on 19 March 2015. As we shall see this is not such an impressive feat.

In summary, here are a few interesting details that will be discussed in this text:

  1. They cherry-pick data in order to claim a 13% increase in congenital malformations after Fukushima, when looking at the five US states that border to the Pacific Ocean (Alaska, Hawaii, Washington, Oregon, and California). A look at the available data for the time period 2007-2013 clearly shows a different story.
  2. They use measured data for gross beta activity, trying to claim that the increased levels after Fukushima would lead to the increase in congenital malformations. A look at their data shows that the levels of radioactivity are ridiculously low, even when taking into account the increase after Fukushima.
  3. It is the third time they get something published in Open Journal of Pediatrics (OJPed). After an inquiry it turns out that this journal has no functioning peer review process and is neither fulfilling its own code of ethics, nor the code of conduct set up by COPE (Committee for publication ethics).

If this is enough for you, then you can happily stop reading. If you want the long and boring details you can continue.
Source: The Keep Calm-O-Matic (http://www.keepcalm-o-matic.co.uk)


1. The claims of S&M

Here is a quote from the abstract of the article, published in Open Journal of Pediatrics, Vol. 5, No. 1, pp 76-89. They write:
We compare rates of five congenital anomalies for 2010 and 2011 births from April-November. The increase of 13.00% in the five western states is significantly greater than the 3.77% decrease for all other U.S. states combined (CI 0.030 - 0.205, p < 0.008). Consistent patterns of elevated increases are observed in the west (20 of 21 comparisons, 6 of which are statistically significant/borderline significant), by state, type of birth defect, month of birth, and month of conception. While these five anomalies are relatively uncommon (about 7500 cases per year in the U.S.), sometimes making statistical significance difficult to achieve, the consistency of the results lend strength to the analysis, and suggest fetal harm from Fukushima may have occurred in western U.S. states.
The article is at times quite entertaining. The elaborate exercise of dividing the data into month of conception (which will not be addressed here, it is not necessary as the very first claim can be disqualified and thus making the rest of the reasoning meaningless), the use of terms like "borderline significance" (some concerns are raised here, here and here), the extensive list of references, starting off with Herman Muller's classical fruit fly experiments in 1927, ending with a pompous quote from Rachel Carson's Silent Spring. It all makes for a Sunday afternoon reading whenever the sudoku gets too challenging, or if the cultural pages do not cover any interesting topic.

Ignoring all that we should focus on their main message. A 13% increase in congenital malformations for the children born in April-November 2011, compared with the children born in April-November 2010. The five types of congenital malformations referred to are (each with a link to the corresponding wikipedia page explaining the illness):


For each of the malformations S&M mention that ionizing radiation can be a cause, but considering the long list of references in the article, this particular section is lacking a comprehensive list, save for a few random shots.
The reason that S&M looked at these five congenital malformations is that they are the ones available in the Wonder data base at Centers for Disease Control and Prevention (CDC). So it is real data that S&M have used, but the way they use them is nothing to be proud of.

Checking the numbers

From the Wonder data base at CDC the five types of congenital malformation can be extracted separately since 2007. Selections have to be made made for time periods and US states, there are also other constraints that can be used, such as maternal age, educational level, tobacco use, etc.

Technical side notes

  1. In the following plots I have extracted data from the Wonder data base. For each searched option (for instance, number of cases of Down syndrome in Oregon in August 2010) the resulting data is divided into different categories:
    • Total number of births
    • Yes (i.e. number of born children with Down syndrome)
    • No
    • Not stated
    • Not reported
  2. Due to privacy issues some of these categories may be suppressed if the number of cases within that category is less than 10. As the number of congenital malformations is relatively low it is quite frequent that the search returns suppressed categories, but there are some tricks in order to obtain the correct numbers.
  3. S&M have handled this in a somewhat different way, and they have excluded the cases in the category "Not stated". This leads to a higher rate, especially for Alaska, but in most cases my data agree with the data of S&M to the second digit, though with a few exceptions. None of the deviations change any of the conclusions below.
  4. Unlike S&M I give all values rounded to integers, two digits accuracy is just silly in this case.


If we assume that the increased levels of radioactivity from Fukushima in 2011 could cause congenital malformations, then it would make sense to look at the five states that border the Pacific Ocean, i.e. Alaska, California, Hawaii, Oregon and Washington for the months following the exposure of pregnant mothers. S&M have done that by extracting values for the months April to November in 2011, comparing the rates of congenital malformation with the same time period in 2010. In table 3 of the article they show the numbers for the time period April-November, and then for the subsets April-July and August-November. They compare the five Pacific states (13% increase) with the other US states (4% decrease). If we plot the data for April-November it will look like in Figure 1 below. Added is also the total values for all of the US states (i.e. the combined effect of "5 states" and "Other US").

Figure 1. Data showing the rate of the five congenital malformations for the
five Pacific states (Alaska, California, Hawaii, Oregon, Washington), when
looking at births during April-November for the years 2010 and 2011. Data are
also shown for the other US states, and all US states.
But as the Wonder data base has data for these five congenital malformations since 2007, it would be helpful of S&M, if they had an honest intent, to plot the entire time period 2007-2013. As seen in Figure 2, there is a good reason for them to not do it: it effectively disqualifies the claim, and thus the entire article.

Figure 2. The same data as in Figure 1, but for the years 2007-2013.
As seen the drastic increase from 2010 to 2011 is not so remarkable. The rate for 2011 is at a quite normal level. Instead it could be of interest to discuss the reasons for the drastic decrease in 2010, but such a discussion can not involve any radioactivity from Fukushima. Once again S&M have clearly showed that they are not to be trusted. They are cherry-picking data again, and celebrities like Alec Baldwin and Christie Brinkley should spend their money elsewhere than filling Joseph Mangano's pockets.

Any further claims in the article are pretty much irrelevant after seeing the time trend as displayed in Figure 2, but let us anyhow look at some of their other results. In table 4 they display the data for 2010-2011 separately for each state, and for all five Pacific states there is some increase. Figure 3 shows this, together with the data for the other US states. Figure 4 shows the same thing but for the time period 2007-2013.

Figure 3. Same data as in Figure 1, but divided into data for each state.


Figure 4. Same as Figure 3, but for the time period 2007-2013.
For states with relatively small populations, such as Alaska and Hawaii, the data scatters a lot for each year. The reasons for that is that a small change in absolute numbers will give a relatively large change in the ratio. Again, S&M has nothing to prove here. More noteworthy is the difference in average levels for these five states. California has a rate of about 100 per 100 000 births, while the other four states tend to have yearly levels above 200 per 100 000 births.

In table 5, S&M have separated the data into each form of the five congenital malformations. It could be of general interest to also compare the levels for the different malformations, so let us plot them. First in S&M cherry-picking mode, Figure 5, followed by a more honest display in Figure 6.

Figure 5. The rate of birth defects for the five Pacific states, divided for each
form of congenital malformation.
Figure 6. Same as Figure 5, but for the time period 2007-2013.
Once again we see that the discovered increase from 2010-2011 is not very remarkable. S&M have no case to show. Again.


2. What about the radiation levels?

The data S&M show in table 1 of the article do indeed indicate an increase of radioactivity, they claim a factor of seven increase for the time period 15 March-30 April when comparing 2011 with 2010. It should be noted that they complain about lack of reliable data (only "gross beta" activity is used), but still they make some claims based on the official data that they have extracted. They give the average values with 5 digit accuracy, but no indication about how the data vary over time or between different measurement sites. The table with measurement sites that they supply in an appendix of the article is therefore not of much value. Ignoring all that, let us look at what this increase of a factor seven does mean. This part follows the reasoning in an earlier blog post about S&M (see Figure 5 in that post).
  • In 2010 the average activity for the time period was 0.005 picocuries per cubic meter of air. Converting this to number of decays it means one decay every 5287 seconds in this cubic meter of air. Put in a more convenient time frame it means one decay every 1 hour and 45 minutes.
  • In 2011 the average activity for the same time period was 0.033 picocuries per cubic meter of air. This means one decay every 13 minutes in this cubic meter of air.
  • An adult has about 5 liters of lung capacity. 
    • This means that in 2010 there was an extra beta decay occurring within that person's lungs once every 12 days.
    • In 2011 there was instead one extra beta decay occurring within that person once every 2 days.
    • In 2 days an adult person has about 1 billion beta decays within his or her body, from natural sources (Carbon-14 and Potassium-40). S&M want us to believe that one extra decay per 1 billion normal decays within the body of an expecting mother would cause a 13% increase in congenital malformations in the children to be born. Nah!


3. But wasn't the article peer reviewed?

According to the journal web site, submitted articles undergo a peer review process:
Submitted manuscripts adhering to journal guidelines are reviewed by the Editor-in-Chief or an Editor, who will assign them to reviewers.
But considering the fact that this is the third time that S&M succeeds with getting an obvious cherry-picking exercise through in this journal (here are the first and second articles), and the way that earlier protests have been handled (see here) it is very doubtful that a proper scrutiny has been performed. Open Journal of Pediatrics belong to the open access publishing house Scientific Research Publishing (SCIRP), which is registered in Delaware, U.S., but operates from China. According to their own publication ethics statement they claim:
SCIRP is committed to maintaining high standards through a rigorous peer review, together with strict ethical policies. Any infringements of professional ethical codes such as plagiarism, fraudulent use of data, or bogus claims of authorship should be taken very seriously by the editors, with zero tolerance.  
SCIRP follows the Code of Conduct of the Committee on Publication Ethics (COPE), and follows the COPE Flowcharts for Resolving Cases of Suspected Misconduct.
Similar statements are to be found on the Publication Ethics and OA Statement page of OJPed. Furthermore it is interesting to read the Appointment Letters send to those who accept being on the Editorial Board, and the Editor-in-Chief (pdf). Of the eight duties expected by the Editor-in-Chief, the first three are worth noting:
  • take an active part in the peer review process as Editor-in-Chief (http://www.scirp.org/aboutUs/ForAuthors.aspx) with quick response, 
  • take the ultimate responsibility for the peer review process and the academic quality of the journal, 
  • supervise editing of manuscripts for each issue with respect to scientific writing and layout conventions
With these in mind I contacted the Editor-in-Chief, asking about his involvement with the journal. The response was the following:
"I am listed as the editor-in-chief, but as it has turned out, I have no substantive responsibilities and am not involved in editorial decisions. I therefore plan to resign my "in name only" position."
Quite interesting. If the person having the ultimate responsibility for the peer review process, and the academic quality of the journal, has not performed any duties at all for the journal, then who is in charge? One may imagine that somebody else has performed any sort of peer review, but since the person who is expected to take the ultimate responsibility has not done so, the academic credibility of the journal is close to zero. We know that the scientific relevance of the S&M article is non-existant (if you are still in doubt, please go back to Figures 1 and 2), and the fact that the entire journal is run like a ship without its captain makes it even less relevant.

In hope of finding out something more I wrote to the 23 persons being named as members of the Editorial Board. These are respected medical doctors and researchers, some of them with quite impressive CVs. For Editorial Board members there are seven duties expected from those who accept being editors, again the three first are relevant:
  • take an active part in the peer review process as Editor (http://www.scirp.org/aboutUs/ForAuthors.aspx) with quick response, 
  • review up to five papers during the year if it is necessary for the journal, 
  • support quality control of the journal
Here is my email to the editors:
Dear Dr/Professor ***,

I am scrutinizing an article recently published in Open Journal of Pediatrics ( http://www.scirp.org/journal/Home.aspx?JournalID=609#.VR2-0o8-s_A), where you are named as editor.

It turns out that the article is a cherry-picking exercise, and there are other flaws in the article that should have stopped it from being accepted in the first place.

Therefore I contacted the Editor-in-Chief. He responded that he had accepted the position but that he has never performed any service for the journal. If the named Editor-in-Chief has never done any work for the journal, then the entire peer review process is in doubt for this journal.

Thus I wonder about your involvement with this journal, have you ever performed any service for it?
Best wishes, Mattias Lantz - Uppsala university, Sweden
So far 15 persons have responded. Each response is unique and has led to some further email exchange, but here is a summary of the answers to my main question:
  • 5 of them have performed some sort of editing or peer reviewing of one or a few articles
  • 2 can not remember if they ever did anything for this journal
    • They act as editors and reviewers for many journals and therefore can not remember what they have done for this particular journal
  • 6 have never been asked to perform any of the duties expected
    • A valid reason could be that their area of expertise has not been or relevance yet
    • A more likely reason seems to be that the journal is only interested in showing their names, thus mis-using their reputation. Some of them have asked to have their names removed, or their contracts have ended a long time ago, but their names are still on the list
  • 2 were not aware of that they are on the list, thus their names were put there without their consent
So there have been a few editors that have acted according to the Appointment Letter. They have performed serious peer review in good faith. But none of them seem to have been involved in more than a few articles. If the same trend follows for the 8 editors that have not responded, then only a handful of the 206 articles published in OJPed since 2011 have been scrutinized according to the journal's own guidelines. These scrutinies are still lacking a quality stamp since the Editor-in-Chief never was involved, and thus never took "the ultimate responsibility for the peer review process and the academic quality of the journal." If somebody else has taken that responsibility, that person should be named. This is not the case.

The fact that two of the named editors never have accepted to be on the list is of serious concern. It means that their names were put there to give credibility to the journal. A serious journal attracts real people to their editorial board, and they use them for the editing and peer review process. In such a journal, whether they charge the authors for the published articles or not (this one does), cherry-picking exercises like the ones by S&M would not pass the quality check. Open Journal of Pediatrics is clearly not a serious journal, and no article published there can be taken seriously. The question is: Can any journal from the SCIRP publishing house be taken serious? Jeffrey Beall's blog Scholarly Open Access indicates to me that the answer to that question is No. This is unfortunate for all researchers who have submitted articles in good faith. For S&M we can only say: Congratulations, you managed to scare some people again!

/Mattias Lantz

Acknowledgements: Many thanks to all of the named editors, including the Editor-in-Chief, who have responded to my inquiry regarding their involvement with the journal.

1 comment:

  1. I’ve read some good stuff here. Definitely worth bookmarking for revisiting. I surprise how much effort you put to create such a great informative website.
    Zofran birth defects

    ReplyDelete