We summarized the adverse event profile of 14,474 pregnant women who received the 2009 H1N1 vaccines in Taiwan. Most adverse events reported were consistent with those described following administration of inactivated seasonal influenza vaccine to pregnant women 
. Review of reports did not find any concerning pattern of adverse pregnant or fetal outcomes.
The inactivated seasonal influenza vaccine without adjuvant has been considered safe when administered during pregnancy 
, and the few postlicensure studies that have been published also supported the safety of unadjuvanted 2009 H1N1 vaccine in pregnant women 
. Compared with the unadjuvanted product, fewer data have been available on exposure to MF59®-adjuvanted influenza vaccine in pregnancy. A review of the Novartis pregnancy database for 43 reported exposures to MF59®-adjuvanted influenza vaccines found that the distribution of pregnant outcomes was similar in subjects exposed to MF59®-adjuvanted and unadjuvanted products at any time of pregnancy 
. However, the inclusion of only 43 pregnancies limited the ability to conclude on risks associated with exposure to MF59®-adjuvanted influenza vaccines 
. Our study provided one of the few assessments on the safety of 2009 H1N1 vaccine adjuvanted with MF59® based on postlicensure data involving 1,275 pregnant recipients. Data are reassuring regarding the safety of MF59®-adjuvanted 2009 H1N1 vaccine in pregnancy.
AEFI reporting rates in our study needed cautious interpretation because data collected through the passive surveillance was underreported 
. Reporting completeness of passive AEFI surveillance can vary but is not routinely available 
. Information on the magnitude of underreporting would be essential to evaluate an association between 2009 H1N1 vaccine and a reported adverse event. In Taiwan, reporting completeness of the passive surveillance for spontaneous abortion after 2009 H1N1 vaccination was estimated to be 5%, but accuracy of this estimate was limited by the equal likelihood of capture assumption of the capture-recapture method 
. Spontaneous abortion with increasing onset interval from vaccination was less likely to be captured by the passive surveillance system 
. In addition, the type of facility where a person received the 2009 H1N1 vaccine could also affect case ascertainment by the LLDB. At provider offices, vaccination details were electronically transmitted to the LLDB through NHI computerized database; however, immunization records at schools, workplaces, or mass vaccination stations were maintained in paper forms and would have to be manually computerized 
. Pregnant women who received the 2009 H1N1 vaccine at provider offices were more likely than others vaccinated elsewhere to be captured by the LLDB.
The estimated risks of spontaneous abortion in pregnant women receiving the unadjuvanted or MF59®-adjuvanted product were lower than background rates for the general pregnant population 
. The comparisons to background rates, however, have some limitations. Most spontaneous abortions occur within the first trimester of pregnancy and the rate declines throughout pregnancy 
. Our study did not allow risk calculations for each trimester because the denominator data was based on number of 2009 H1N1 vaccine doses administered to women at all stages of pregnancy; therefore, we may have underestimated the risk of spontaneous abortion after 2009 H1N1 vaccination. Receipt of 2009 H1N1 vaccine is voluntary and thus may be preferentially sought by motivated and healthier individuals. In practice, clinicians may not have vaccinated patients perceived to be at risk for adverse pregnant outcomes. These would result in a preferential receipt of 2009 H1N1 vaccines by a relatively healthy pregnant population (“healthy vaccinee phenomenon”) 
. Spontaneous abortions have been reported among pregnant women with 2009 H1N1 infections 
. There also have been reports of higher miscarriage rates during previous influenza pandemics 
. Pregnant women who received the 2009 H1N1 vaccine might have been protected from influenza infection and therefore, were less likely to develop spontaneous abortions than the unvaccinated group.
This study shared other inherent limitations of all passive surveillance systems 
. Although clinical reviews of medical records for reports involving pregnancy-specific adverse events could improve data quality and completeness, limitations such as reporting biases and lack of a controlled unvaccinated remained. Therefore, these data cannot be used to determine whether a vaccine causes an adverse event. Our study collected only information about adverse events that occurred after the administration of 2009 H1N1 vaccine and did not have all the information essential for epidemiologic assessments of causality. It is not known if the women who received the 2009 H1N1 vaccine were different in baseline characteristics or at higher risk for adverse pregnant or fetal outcomes compared with the general pregnant population.
Nevertheless, the passive AEFI surveillance provided rapid initial assessment of adverse pregnant and fetal outcomes, but it is only part of the 2009 H1N1 vaccine safety monitoring activities in Taiwan 
. Taiwan has developed an infrastructure to actively follow up with women who received the 2009 H1N1 vaccines at different stages of pregnancy, as well as to follow their fetal and newborn outcomes, using comparison groups 
. Data from these ongoing studies can provide further information on the safety of 2009 H1N1 vaccines in pregnant women.