Re: Vaccination against pandemic A/H1N1 2009 influenza in pregnancy and risk of fetal death: cohort study in Denmark
In his rapid response to our paper on H1N1 vaccination in pregnancy and the risk of fetal death, Dr Goldman requests data on the number of vaccinated women by trimester and the number of fetal deaths by trimester and exposure group. He states that this would allow the calculation of rates of fetal death per million vaccinated women by trimester and incidence rate ratios.
Our study was designed to take into account the fact that the risk of fetal death varies considerably according to gestational age. Therefore, the underlying time scale in the Cox proportional hazards regression models was gestational age in days and vaccination was defined as a time-varying exposure variable. Because, on the whole, the rate of fetal death decreases with increasing gestational age, while the number of women who receive the vaccine increases, calculating crude rates of fetal death by number of women without taking the gestational age into account will invariably result in lower rates among the vaccinated group. In fact, as evident from the number of person-years and events in Table 2 in our paper, the crude rate of fetal death is 27 per 1758 person-years in the exposed group and 1785 per 19578 in the unexposed group. This corresponds to crude rates of 15 per 1000 person-years in the exposed group and 91 per 1000 person-years in the unexposed group, spuriously suggesting a strongly protective association in favor of the vaccine. However, results from the actual Cox model, which takes the gestational age into account, show that there is no difference between the vaccinated and unvaccinated group. This clearly demonstrates that the gestational age at which fetal death occurs always needs to be taken into account in this type of study and that reporting crude rates would be misleading.
Competing interests: No competing interests