In a sample of 19 iiveborn infants with anencephaiy, Meinick and Myrianthopoulos [ 19871 found markediy shorter survivai compared with our previousiy published data on survival in a series of 181 cases [Baird and Sadovnick, 19841. They suggest that our data represent either biased case ascertainment or better perinatai care. We would like to address these points.
Our data were obtained through the population-based Health Surveillance Registry [ Baird, 19871 which ascertains anencephalic births through severai sources inciuding mandatory provincial documents, namely Physician's Notice of Birth and Vital Registrations of Death. As registration from these sources is not selective, but inciudes al1 identified cases, it is uniikely that we wouid identify only those infants with better survival. It is also uniikeiy that many anencephaiic fetuses or infants would have been missed. Therefore, bias of ascertainment does not seem to be a probable explanation.
The analyses used by Melnick and Myrianthopoulos are appropriate and we agree that there is a significant difference between the two samples.
However, the two studies differ markedly with respect to sample size (19 subjects compared with 181). Whiie we recognize the validity of results from both studies, it is important to remember that the smailer the sample, the greater the effect of adding (or deieting) very few additional cases [Rao, 19651. Our cases were manualiy reviewed, especialiy those with reiatively long survival, to ensure that the diagnosis of anencephaly was accurate. As noted in our paper, the infant with the longest survival was given iife support. Perhaps better perinatal care to these cases in British Columbia is indeed an explanation for any difference. British Columbia has a universal average medicai insurance program, and it also may be that more support is given to these cases after birth.
In conciusion, although we recognize the difference in survival between our cohort and that presented by Melnick and Myrianthopouios [1987], we do not feel that biased ascertainment can account for this.