人类先天性畸形的流行病学

Epidemiology of Human Congenital Malformation


In 1964, a few years after the thalidomide tragedy, the National Board of Health and Welfare
in Sweden decided to start a register of congenital malformations for surveillance purposes.
Together with Jan Winberg, pediatrician, I was tasked with arranging and running the register
which I continued with until 1997, a third of a century. Since then I have worked with the
National Board of Health on a free-lance basis. In 2014, I will have been working within the
fi eld of reproductive epidemiology for 50 years. For natural reasons I do not expect to be able
to continue this work more than for a relatively short time. I therefore thought it would be a
good idea to collect data and present them in a book with some comments and conclusions
I have reached, based on these experiences. I decided to restrict the period of data analysis to
1998–2010. The 1998 limit was put because that was the fi rst year when the new ICD-10 classifi
cation system was used in Sweden which made malformation registration in some registers
more uniform and more detailed than previously.
Much birth defect epidemiology is based on the identifi cation of outcome (malformations)
and exposures. After having performed a few retrospective case-control studies and experienced
the problems with recall bias and non-responders, I decided to exclusively rely on prospectively
collected exposure information, ascertained in a similar way during pregnancy for
women who had malformed infants and women who had non-malformed infants. This was
possible because in 1973 the National Board of Health and Welfare started a nationwide registration
of medical information on all births, and in 1982, 1992 and 1995 data on some exposures
were entered, ascertained by midwife interviews at the fi rst prenatal care visit. Practically
all pregnant women come to the prenatal care centers and the medical record forms used were
identical throughout the country since 1982. This thus represents a large, ongoing prospective
study even though data are recorded in a national health register. The draw-back is, of course,
that only recorded exposures can be studied and many of the putative risk factors which have
been discussed in the literature are not possible to collect in a system like this. It is always a
balance between what could be of interest and what is possible to achieve. The midwives who
make the interviews around the country and the women who answer them must feel that the
questions are realistic and can be of signifi cance for the clinical situation.
One can, for instance, hardly register information on nutrition (because of the complexity
in description) or personal matters like showering or hot tub baths. Maternal chemical exposure
is important information but unfortunately the only realistic data which can be recorded
are occupation which is too crude – and often recorded in such an unspecifi c way that it is not
useful. Efforts to register important but sensitive matters like alcohol consumption were not
successful. One tried a simple question, “Have you used alcohol during your pregnancy?”
Answer alternatives: No, Seldom, Regularly. I went through 500 randomly selected records
and there was only one woman who said she had used alcohol regularly and a note said it
meant drinking low-alcohol beer at dinner every day. It seems rather unlikely that none had
drunk wine or strong liquor during pregnancy but the general knowledge of the hazards of
alcohol drinking during pregnancy may have mathe women unwilling to disclose their behavior.