The authors acknowledge several of the limitations of their preliminary analysis. The timing of exposure to AVA (i.e., whether or not it occurred in the first trimester) was not precisely known for each infant but rather was estimated based on traditional gestational age cut points for term, preterm, and very preterm infants. Thus, time of exposure was subject to mis-classification because of the manner in which exposure periods were estimated. Inexact vaccination dates could also contribute to misclassification of time of exposure. In addition, it appears that all “major” birth defects were combined, which may not be biologically appropriate. The accuracy of identification of birth defects is uncertain, and analyses were not adjusted for differences between groups in other factors that might influence risk of birth defects such as maternal alcohol use, exposure to medications, or use of folic acid supplements. The number of infants exposed during the first trimester is relatively small, making estimates of risk derived from such analyses highly uncertain. These limitations again emphasize the need to distinguish possible “signals” generated by exploration of large databases, which require further and more definitive studies, from findings of causal associations.

These study results remain preliminary and therefore may change with further analysis. Because of the importance of this issue, the study investigators are working rapidly to validate both exposures and outcomes using primary data sources, which is highly appropriate. In the meantime, the standing DoD policy to avoid immunization of women during pregnancy has been reiterated, which is also appropriate. Further conclusions about the safety of AVA during pregnancy must await the results of this and other studies.

The committee has reviewed information from a variety of sources, including VAERS and DMSS, on the association between vaccination with AVA and adverse events. For AVA, as with any vaccine, it is essential in assessing questions regarding the safety of the vaccine to distinguish between immediate-onset health events that are observable within hours or days following vaccination and later-onset events that would be observable only months or years following vaccination.

On the question of immediate-onset health events, substantial amounts of data are now available from VAERS, DMSS, and epidemiologic studies. The committee concluded that vaccination with AVA is associated with certain acute local and systemic effects. Epidemiologic studies have consistently found, using either active surveillance (Brachman et al., 1962; Pittman, 2001b,c; Pittman et al., 1997, 2002, in press) or passive surveil-