"None of the studies measured paternal smoking objectively, for instance by measuring nicotine metabolites in body fluids, relying instead on self-reported questionnaire data
, gathered usually from the mother
of the affected child, although attempts to interview the father were also made. The diagnosis of cases was based mainly on hospital records, although information from physicians, paediatricians and pathologists, labour wards, death certificates and autopsy reports were also used in some of the studies. The source of diagnosis was not stated in one study. The case-control studies matched for a variety of factors, with all including some index of the child's age. Levels of non-response were not stated at all
in one of the studies, and were only given for the cases (10%) in another. Few differences in non-response were seen in the other two case-control studies.
Although all of the studies were concerned with paternal smoking, only one (Hearey) appeared to have collected smoking data directly from the father of the index child, with the other relying on information gathered from the mother. It is clearly possible that misclassification may occur due to inaccuracies in this information
, as the mother may not have complete knowledge of the father's smoking habits, particularly in couples who are not actually living together
. Additionally, all of the case-control studies suffer from the problem that the smoking data was collected after the outcome of the birth was known
. This may have affected parents' recall, with some over-estimating the amount smoked in an attempt to rationalize their child's illness
The differing methodology used by the studies makes it difficult to ascertain the relative weight which should be given to each study. There may also be a failure to publish studies which do not find a positive result
. As so few studies were found which looked at paternal smoking and birth defects in offspring it is difficult to reach any firm conclusions regarding this point.
At first glance the epidemiological evidence suggests a weak positive association between paternal smoking and birth defects of any type. Of the studies which considered the risks for all types of birth defects found raised relative risks, two of which were significant. Just over half of the relative risks presented were raised, although only two were significantly so. One significantly negative association was also found. However, the risk factors found were nearly all below 2.0, and at this level it would take only a small bias to produce a spuriously positive association, or to mask a true association