The following letter was sent to the editor of Open Journal of Pediatrics (OJPed) regarding the congenital hypothyroidism study by Joseph Mangano, Janette Sherman and Christopher Busby.
The letter indicated that the actual count of confirmed cases of congenital hypothyroidism (CH) from the California Public Health Department does not match the authors' count because they disregarded the actual count given to them. Instead, the authors (1) invented their own definition of confirmed cases of CH, (2) misrepresented the real definition of CH, and (3) invented a fictitious diagnostic category of CH which they call "borderline cases.”
OJPed's response was, "Thank you for your mails. However, the letter cannot be published. And the paper published in OJPED will not be withdrawn."
The letter was the subject of this post by Jeffrey Beall who keeps track of predatory journals.
Also, the video by Ian Goddard clearly explains issues regarding this study as well as another study by two of the authors.
*****
April
14, 2014
A letter to the editor of Open Journal of Pediatrics
In the article, “Changes in confirmed plus borderline cases of congenital hypothyroidism in California as a function of environmental fallout from the Fukushima nuclear meltdown,” Mangano, et al. base their conclusion on erroneous and selective data interpretation regarding the number of confirmed congenital hypothyroidism (CH) cases in the study population.
The authors obtained the newborn screening data for congenital hypothyroidism (CH) from the Genetic Disease Screening Program at the California Department of Public Health (CDPH), including the number of confirmed cases and children screened, grouped by their thyroid stimulating hormone (TSH) levels from 2009 to 2012. The CDPH office was contacted to obtain a copy of the original data used by the authors in an attempt to replicate their findings.
In the article, “Changes in confirmed plus borderline cases of congenital hypothyroidism in California as a function of environmental fallout from the Fukushima nuclear meltdown,” Mangano, et al. base their conclusion on erroneous and selective data interpretation regarding the number of confirmed congenital hypothyroidism (CH) cases in the study population.
The authors obtained the newborn screening data for congenital hypothyroidism (CH) from the Genetic Disease Screening Program at the California Department of Public Health (CDPH), including the number of confirmed cases and children screened, grouped by their thyroid stimulating hormone (TSH) levels from 2009 to 2012. The CDPH office was contacted to obtain a copy of the original data used by the authors in an attempt to replicate their findings.
In California, the cutoff for a preliminary indication of CH is a TSH value >=29 mIU/L, as seen in the Newborn Screening cutoff/reference ranges dated 09/11/2013.
However, the authors erroneously state,
“The program confirms CH cases by using only TSH scores greater than 29.0 micro international units per milliliter (μIU/ml). Any child meeting this criterion is prescribed replacement thyroid hormone, to promote normal physical and mental development. On January 1, 2011, the state changed the assay method used to calculate TSH. Scores increased in most newborns, and thus the number of CH cases also increased.” (Note: μIU/mL is equivalent to mIU/L)
“The program confirms CH cases by using only TSH scores greater than 29.0 micro international units per milliliter (μIU/ml). Any child meeting this criterion is prescribed replacement thyroid hormone, to promote normal physical and mental development. On January 1, 2011, the state changed the assay method used to calculate TSH. Scores increased in most newborns, and thus the number of CH cases also increased.” (Note: μIU/mL is equivalent to mIU/L)
Authors make several errors in the study, beginning with the passage above. First, the only thing being measured by the Newborn Screening program is an elevated whole-blood TSH level which is not a diagnosis of CH as further clinical confirmation is required for a diagnosis of CH. TSH values over the cutoff of 29 μIU/ml only means “positive” screening, and it is not true that any child meeting this criterion is prescribed replacement hormone.
In e-mail correspondence with Robert J. Currier, Ph.D., Acting Chief, Program and Policy, from the CDPH Genetic Disease Screening Program, Currier stated,
“A presumptive positive baby (for CH) takes a serum TSH testing and other optional tests (clinicians’ choices) such as free T4. The baby is also referred to a state-approved endocrinology center and seen by an endocrinology specialist. Repeating follow-up if needed. If the baby is a confirmed CH case, then an annual follow-up check with the endocrine center will be established.”
“A presumptive positive baby (for CH) takes a serum TSH testing and other optional tests (clinicians’ choices) such as free T4. The baby is also referred to a state-approved endocrinology center and seen by an endocrinology specialist. Repeating follow-up if needed. If the baby is a confirmed CH case, then an annual follow-up check with the endocrine center will be established.”
Thus, the definition used by the authors of “confirmed CH cases” is inaccurate, yet the authors use "positive" screening to erroneously count confirmed cases.
Second, as for the new assay method reportedly instituted on January 1, 2011, what increased was not “the number of CH cases” alluded by the authors, but the TSH cutoff value. As Currier also states,
“Before 2011, the cutoff value for CH presumptive positives was 25 μIU/ml. In early 2011 our labs adopted a brand new technology for TSH testing. As a result, there had been a change in TSH values at populations basis and we adjusted the cutoff to 29 μIU/ml in April 2011. In other words, before April 2011, TSH screening >=25 μIU/ml were considered as CH positives. This information (of changed testing method) was supplied to Joe Mangano as well.”
“Before 2011, the cutoff value for CH presumptive positives was 25 μIU/ml. In early 2011 our labs adopted a brand new technology for TSH testing. As a result, there had been a change in TSH values at populations basis and we adjusted the cutoff to 29 μIU/ml in April 2011. In other words, before April 2011, TSH screening >=25 μIU/ml were considered as CH positives. This information (of changed testing method) was supplied to Joe Mangano as well.”
Third, the authors further define “borderline cases” as “a cautious range of 19.0 - 28.9 μIU/ml,” and add the "borderline cases" to their estimate of total CH cases. But this definition is meaningless, as CH cases are only confirmed after follow-up clinical observation and further testing.
In fact, Currier states further,
“We do not have any definition for 'borderline cases.' TSH testing is a screening tool and only clinically diagnosed CH cases are considered 'CH cases.'” Currier added, “It is possible, although very rare, that a confirmed case would have a TSH value below cutoff and was diagnosed later clinically.”
While the authors declare that California has "a consistent definition of the disorder," conjuring the ill-defined "borderline" category to be combined with mere "positive" screening cases represents an egregious introduction of bias.
This step appears to increase the 2011-2012 CH total from 658 to 4670, inflating the p value (p < 0.00000001 for “combined confirmed and borderline cases”) in the statistical analysis. In their Discussion the authors then write "...despite having to define borderline cases arbitrarily. In California, adding borderline cases (TSH between 19.0 and 28.9 μIU/ml) to those confirmed cases (TSH over 29.0) increases the number of cases by more than seven fold." This statement confirms how data selection was biased and that the authors consider "borderline" and "positive" screening to be equal to "confirmed cases" of CH.
Currier also provided the same two sets of data that were given to the authors (see Tables 1 and 2) and additional details about the CH screening program.
Currier indicated that the data on the confirmed case count table had been updated since it was given to authors, because “in rare cases, a baby who was misdiagnosed or diagnosed with a degree of uncertainty could be removed from our registry at a later date.” However, Currier assured that it should be very close to data initially provided to the authors. Currier also mentioned that “Mangano did not use the data from this table (Table 2) in his article.”
Table 1 TSH Value Breakdown from Initial Newborn Screening in California, 2009-2012
Table 2: Confirmed case count of Primary Congenital Hypothyroidism in California, 2009-2012
Alfred Körblein, a retired physicist and independent consultant in epidemiology in Germany, plotted the number of “confirmed” cases of congenital hypothyroidism per 100,000 (Figure 1). As can be seen, an increase in the confirmed cases in the period from March 17th to December 31st shows variation similar to other time periods.
Figure 1: Incidence of Confirmed Cases of CH 2009-2013
Fourth, the authors state, at the end of the Results section, “With much larger samples than just confirmed cases, a better understanding of the true change can be approached.” The desire for larger sample size by the authors of this paper suggests a limited understanding of the CH screening program in California.
In conclusion, the study by Mangano, Sherman and Busby has critical flaws: 1) incorrectly taking raw positive screenings (TSH >=29 μIU/ml) to be confirmed CH cases, thus disregarding the correct number of actually confirmed CH cases they received from the CDPH; 2) defining a meaningless diagnostic category of “borderline” cases that has no basis in the screening program or in medical practice; and 3) claiming that the faux rate they conjured (positive screening results plus "borderline" screening results) is a valid construct; and 4) claiming that the CH increase in 2011 was statistically significant, whereas the plotting of the number of actual clinically confirmed cases from 2009 to 2012 clearly shows no significant increase.
Sincerely,
Yuri Hiranuma, D.O.
Member, Radiation and Health Committee, Physicians for Social Responsibility
