Errors and Omissions in Experimental Trials - 2d
ADDITIONAL OBSERVATIONS ON THE EVANSTON, GRAND RAPIDS AND NEWBURGH TRIALS
1. Gross numerical errors in statements of the number of children examined.
The Evanston Trial
These comments on the Evanston trial were published in 1980 in the present author's book Fluoridation Scientific Criticisms and Fluoride Dangers. It was stated:
"Additional Errors in the Evanston Trial Data.
In January, 1967, which was the twentieth anniversary of the commencement of the Evanston Trial, an entire special issue of the Journal of the American Dental Association was devoted to a report on that study (Blayney and Hill, 1967). In this, the original tables, complete with their gross numerical errors, were reproduced, despite the fact that these [errors] had been pointed out eight years earlier (Sutton, 1959) and some of them had been acknowledged by the authors (Sutton, 1960). In addition, several faulty tables were published for the first time.
The tables [in this issue of the J. Amer. Dent. Ass.] then showed three different statements regarding the number of children aged 6-8 years who were examined in Evanston during the 1946 examination:
(i) 1991 children - see Tables 10, 11,30,40 and 47.
(ii) 1985 children - see Tables 7,8,16,18,21 and 32.
(iii) 1754 children - see Tables 24 and 25.
There were also no fewer than six different statements in that article of the number of children aged 12-14 years examined in Evanston in 1946:(i) 1703 children - see Tables 15 and 32.
(ii) 1702 children - see Table 47.
(iii) 1701 children - see Tables 11,30,41,44 and 45.
(iv) 1697 children - see Tables 7,9,12,13,17,19,22 and 31.
(v) 1556 children - see Table 26.
(vi) 1146 children - see Table 46.
Between the sum of the two highest statements of the number of children examined in Evanston in 1946, and the sum of the two lowest statements of children examined in the same year in the same study in the same city, there is a difference of 794 children (1991 + 1703 - 1754 - 1146 = 794).
The number of children stated to have been examined in Evanston is even more divergent in the original papers than in this special article. Blayney and Tucker (1948) and Hill et al. (1950) both gave a figure of 4375 children, compared with the number of 3310 in Hill et al. (1957b), a difference of 1065 children.
It was these differences which the medical journalist Anne-Lise Gotzsche, in a letter to the Lancet in 1975, said that she had showed to workers in other fields, and that they had "simply laughed" at the statistics (see Fig. 5, p. 167).
In that book (Sutton, 1980) - prepared as a submission to the Committee of Inquiry into the Fluoridation of Victorian Water Supplies (1980) - it was stated (p. 203):
"These errors were mentioned [by the present author] 12 years ago to the Tasmanian Royal Commission on Fluoridation. Since that time I have not heard of any mention of them or of a criticism having been made of the numerical data published in that report."
It appears that, in the manner common in fluoridation trials, those erroneous tables have been accepted at their face value, without investigation.
More than thirty years ago it was pointed out (Sutton and Amies, 1958b) that:
'This uncritical attitude to these studies is rife." "Also it has been assumed that associations and individuals that ... accepted the responsibility of publicly advocating fluoridation, have undertaken independent examinations of the data, and not merely repeated the opinions of others."
This situation was referred to by Professor John Polya (1964) in his book Are We Safe? He wrote:
"It is immaterial that other evidence in favour of fluoridation is not always false; the point is that persons, bodies and arguments that knowingly or in simplicity acquiesce in one blatant falsehood are unreliable witnesses before a jury either of scientists or of lay common sense."
"The scandal created by the exposure of this absurdity resulted in the admission that the first figure (4,375) was correct. In defence of the other claims it was explained that "out of range" children were eventually excluded from the survey, but then further critical check revealed more numerical inaccuracies, not to speak of the magnitude of a correction exceeding 1,000. In better examples of scientific work the author sticks to his experimental group; discarding on the scale quoted strongly suggests that the experiment had to be altered to fit pre-conceived results. This is one of the common consequences of working without control of observer bias."
It is pertinent to point out that, in the Foreword to that article in the special edition of the Journal of the American Dental Association, in January 1967, Dr F.A. Arnold, Jr., the Assistant Surgeon General, Chief Dental Officer, U.S. Public Health Service (and formerly the chief experimenter in the study in Grand Rapids) stated:
"Here, in a single report, are data on the effect of water fluoridation on dental caries so completely documented that the article is virtually a text book for use in further research. It is an important scientific contribution towards the betterment of the dental health of our nation. It is a classic in this field."
It is indeed a classic - a first-class example of the errors, omissions and misstatements which abound in the reports of these fluoridation trials.
2. False information in the Abstracts of papers
2. False information in the Abstracts of papers
The abstracts of reports on fluoridation trials are unusually important, for it is likely that lay people, and politicians in particular, will confine their reading of the report to the Abstract, assuming that it accurately reflects the findings, and will base their opinions and actions on its statements.
The Grand Rapids Trial. The final report of the Grand Rapids study was published in 1962. Reading the Abstract which preceded the body of the article it would seem that, at last, the authors (Arnold et al., 1962) had come to realize the necessity for comparing the results from the test city with those from the control one for they stated that the results had been "...compared with the caries attack rates in the control group of children in Muskegon, Mich." This claim was not made in the body of the article, which included the statement that: "... fluorides were introduced to this [Muskegon] water supply in July, 1951" Therefore at that time Muskegon ceased to be a control city, some eleven years before this final report (Arnold et al., 1962) from Grand Rapids.
How then, in 1962, could the final result from the test city be compared with data from a non-existent control one?
The claim of Arnold et al. (1962) that they compared the Grand Rapids caries rates with those in the "control group of children in Muskegon, Mich." is shown to be false by their statement that: "...in subsequent [after 1954] analyses of Grand Rapids data, comparison has been made with the original Grand Rapids findings and with those for Aurora."
This is confirmed by the statement in the Abstract that:
"Caries attack rates were lowered by 57 per cent in children 12 to 14 years old in 1959." This figure of 57 per cent is obtained by averaging the figures of 57.0, 63.2 and 50.8 per cent for the ages of 12, 13 and 14 years shown in their Table 2 to be the "per cent reduction in DMF teeth (19441959)" in Grand Rapids (not between Grand Rapids and its control city of Muskegon).
The Newburgh Study. Similar mis-information regarding comparisons being made between test and control cities was published in the same year (1962) by Dr David Ast, the senior author of the Newburgh study. In the Abstract of that paper (Ast and Fitzgerald, 1962) he wrote:
"Among children 12 to 14 years old in the four study areas, reductions in the DMF rates as compared to the rates in control cities ranged from 48 to 71 per cent."
Table 2 is the only one in that paper showing DMF rates for children aged 12-14 years (in one case 13-14 years). In the first two studies listed, Grand Rapids and Evanston, no reference is made to a control, the "reduction" in Evanston, shown as 48.4 per cent, is obviously the 48 per cent mentioned in the Abstract. This "difference" is between the rates in Evanston in 1946 and 1959, not between Evanston and a control, as stated in the Abstract. The Grand Rapids rates are also shown between that city in 1944-45 and 1959, no control data being used. Indeed Ast and Fitzgerald stated in the main text:
"In the Grand Rapids and Evanston studies the control cities were lost before the study was completed, so that the current data have been compared with the base line data."
Not with control cities, as they stated in their Abstract.
There should not have been any confusion regarding the use of the term "control", for the co-author of that paper, Bernadette Fitzgerald, was described as the "senior biostatistician, division of special health services, New York State Department of Health." Therefore the authors' incorrect statement that they compared the caries rates "in the four study areas" with rates in control cities is unlikely to have been made inadvertently.
3. Continuing publication of false statements.
It has just been shown that Dr Ast (the senior author of the Newburgh study) and Dr Arnold (the senior author of the Grand Rapids study) continued to disseminate false statements regarding their studies many years after those ten-year studies were concluded, Also, the arrogance of Drs Blayney and Hill (the authors of the Evanston study) in publishing an article in 1967, which repeated, in a special issue of the Journal of the American Dental Association, figures which they had acknowledged seven years earlier were faulty (Sutton, 1960), indicates the reckless disdain of all those authors for the truth, and for the members of the scientific community (which normally trusts statements made in established journals by senior scientists, for it is not used to being misled by such readily-verified deceptions).
Their false statements do not engender confidence in the reliability of the data published and the statements made by those senior scientists in their original reports of what are still regarded by fluoridation advocates as three of the four main fluoridation studies on which the case for fluoridation mainly relies - those in Newburgh, Grand Rapids and Evanston in U.S.A.
Commenting on the Grand Rapids study, Ziegelbecker (1983) pointed out that the experimenters had examined "all" children from 79 schools in Grand Rapids at the commencement of the trial, but that:
"After 5 years in 1949 they selected children at only 25 schools in Grand Rapids for their investigation and observed children at the same time at all schools in Muskegon (the control city)."
For instance, the number of children aged 12 to 16 years who were examined in Grand Rapids at the commencement of the trial was 7,661, but only 1,031 were examined in 1959 (Arnold et al., 1962).
In 1988, Colquhoun stated:
"In the control city of Muskegon all children were examined throughout the period. From the year-by-year figures for six-year-olds which were published three years later in 1953, it is revealed that an impossible 70.75% reduction was recorded in the first year of the trial (Arnold et aL,1953) and that there was then an increase and no overall reduction in the following years. Examination of similar data for other age groups shows that the sample of 25 schools could not have been representative of the population being studied."
He pointed out that:
"The reported DMF of several of the age groups in this sample, approximately one year after the initial examinations, was lower than that of the same children when they were a year younger."
"Fluoridated water cannot turn decayed, missing or filled teeth into sound ones. It follows that the caries experience of the children had not been reduced as claimed. The large recorded reductions, which were mostly in the first year only, were a result of selection of data."
4. Fictional results?
In 1954 De Stefano reported the findings of professional statisticians regarding the Grand Rapids study. They raised the question whether "... the reported results are merely the fiction of a biased sample."
Ziegelbecker (1983) also, studied this situation. He stated:
"We must conclude from this result that the sample in Grand Rapids was not representative for all children and with respect to the basic examination. In the following years from 1946 to 1949 (and later to 1954) the 25 schools in the sample were the same each year and we see that the caries experience in the sample was not reduced by fluoride in 1946-1949.
If we accept that the sample was representative for the children, aged 6, in the 25 schools in those years then we must conclude that fluoride in the drinking water had not reduced the dental caries experience of children, aged 6, in Grand Rapids in the years before the US Public Health Service released the policy statement [endorsing fluoridation] to the American Dental Association."
"We must conclude from these results that a fluoride content of 1 ppm in the public water supply does not reduce dental caries experience."
Colquhoun stated in 1988:
"In their final study in Grand Rapids, published in 1962 after 15 years of fluoridation, American health officials [including the director of the U.S. National Institute of Dental Research, Dr F.A. Arnold, Jr.] wrote: "... no such dramatic and persistent inhibition of caries in large population groups had ever been demonstrated by any other means than fluoridation of a domestic water supply."
'That statement, which could be described as the dogma of fluoridation, is now considered by an increasing number of critics to be unscientific and untrue."
In view of the disclosure of the types of error which have just been mentioned, such a grandiose claim, although it was widely accepted at the time, can no longer be considered to be true.
More than thirty years ago Sutton and Amies (1958a) commented on this sudden initial decrease in caries reported from Grand Rapids (and from other studies considered). It was stated that the results reported were not those which would be expected if the hypothesis was correct that fluoride "strengthens" developing teeth and makes them more resistant to attack by caries. Despite the fact that the results published from fluoridation studies do not support this hypothesis, it is still mentioned. For instance, the ten members of the task group which in 1984 wrote the latest WHO book on this subject: Environmental Health Criteria 36. Fluorine and Fluorides, referred to the importance of "lifelong consumption" of fluoridated water.