L
AUTOIMMUNE DISEASES 193not. Obviously, this at first seems to demonstrate considerable uncer-
tainty, going a long way to discredit the hypothesis.
However, the five studies that were counted as "negative" did not
show that cow's milk decreased Type 1 diabetes. These five studies
showed no statistically significant effect either way. In contrast, there
are a total of five statistically significant studies and all five showed the
same result: early cow's milk consumption is associated with increased
risk of Type 1 diabetes. There is only one chance in sixty-four that this
was a random or chance result.
There are many, many reasons, some seen and some unseen, why an
experiment would find no statistically significant relationship between
two factors, even when a relationship really exists. Perhaps the study
didn't include enough people, and statistical certainty was unattainable.
Perhaps most of the subjects had very similar feeding practices, limiting
detection of the relationship you might otherwise see. Maybe trying to
measure infant feeding practices from years ago was inaccurate enough
that it obscured the relationship that does exist. Perhaps the researchers
were studying the wrong period of time in an infant's life.
The point is, if five of the ten studies did find a statistically significant
relationship, and all five showed that cow's milk consumption is linked
to increasing Type 1 diabetes, and none show that cow's milk consump-
tion is linked to decreasing Type 1 diabetes, I could hardly justify say-
ing, as the authors of this review did, that the hypothesis "has become
quite murky with inconsistencies in the literature. "38
In this same review,38 the authors summarized additional studies that
indirectly compared breast-feeding practices associated with cow's milk
consumption and Type 1 diabetes. This compilation involved fifty-two
possible comparisons, twenty of which were statistically significant. Of
these twenty significant findings, nineteen favored an association of cows
milk with disease, and only one did not. Again the odds heavily favored the
hypotheSized association, something that the authors failed to note.
I cite this example not only to support the evidence showing a cow's
milk effect on Type 1 diabetes, but also to illustrate one tactic that is
often used to make something controversial when it is not. This practice
is more common than it should be and is a source of unnecessary con-
fusion. When researchers do this-even if they do it unintentionally-
they often have a serious prejudice against the hypothesis in the first
place. Indeed, shortly after I wrote this, I heard a brief National Public
Radio interview on the Type 1 diabetes problem with the senior author