Monday, April 20, 2009

The 2008 Naval Environmental Health Center Study by Dr. Margaret Ryan et al. on anthrax vaccination during pregnancy: a critique and historical review


Comments on: Ryan, MAK, et al. Birth Defects among Infants Born to Women Who Received Anthrax Vaccine in Pregnancy. American Journal of Epidemiology; July 2008.


Meryl Nass, MD

April 20, 2009


Dr. Margaret AK Ryan reports that of the 95,595 military women who delivered babies between 1998 and 2004, anthrax vaccine (one dose or more) was given to 3,465 military women during their first trimester of pregnancy. Dr. Ryan, a physician epidemiologist and Navy Commander, has studied these women for the past eight years, and published results in 2008.


The Original Study Led to Major Policy Changes


Dr. Ryan presented results of a subset of this group (those vaccinated 1998 through mid 2001) in late 2001 and early 2002. Shortly after her presentation to FDA, in January 2002 FDA changed the pregnancy warning on the anthrax vaccine label to category D from C, indicating that data showed the vaccine was associated with birth defects, and should not be given to women during pregnancy. The Assistant Secretary of Defense for Health Affairs, Dr. William Winkenwerder, also responded to Dr. Ryan's findings. He sent memos to all the military services, directing that special efforts should be made to avoid giving the vaccine during pregnancy, including the use of pregnancy tests if appropriate.


[He was reacting to the increased rate of birth defects, but also to the relatively high rate of first trimester vaccinations. Before 2002, military medical professionals tended to vaccinate women who thought they might be pregnant but had no proof. Fortunately, that changed after Dr. Ryan's 2001 preliminary report. A much smaller number of women with first trimester pregnancies have been vaccinated since.]


Her 1998-2001 study data were shared with the Institute of Medicine Committee to Assess the Safety and Efficacy of Anthrax Vaccine and with the Armed Forces Epidemiology Board, as well as FDA. When initially presented, the study compared first trimester-vaccinated women to women who had been vaccinated at any other time. The data showed a reportedly small increase in birth defects in offspring of first trimester-vaccinated women, yet the increase was statistically significant. For some reason, these data were never published, despite the effect they had on both vaccination policy for military women, and changing the pregnancy warning on the vaccine label.


Expanding the Dataset but Reducing Statistical Significance


Subsequently, Dr. Ryan et al. added to the original data collection, extending the period of births through 2004. In 2008, Dr. Ryan published these data, discussing the original comparison between first trimester-vaccinated mothers and mothers vaccinated at another time, but the published paper also compared these births to births in mothers never vaccinated against anthrax.


This expanded dataset yielded a smaller difference in birth defect rates in offspring between first trimester-vaccinated women and those vaccinated at any other time, which was only statistically significant when offspring of first trimester-vaccinated women were compared with offspring of those never vaccinated.


In February 2002 it was reported (in the CDC’s Morbidity and Mortality Weekly Report: MMWR 127 vol 51 No. 6: Notice to Readers: Status of US Department of Defense preliminary evaluation of the association of anthrax vaccination and congenital anomalies) that investigators were conducting a systematic evaluation of original (paper) medical records of women who received anthrax vaccine to determine how well the electronic medical data agreed with the (paper) medical records.


In 2008, Dr. Ryan described significant problems with the electronic data that she used. Her published paper states that only “a subset of women whose data were archived after they left military service” had their hard-copy data compared with the electronic database. Hardcopy records were reviewed for 11,271 of the 95,595 women who gave birth during the study period, and “were more likely from women who gave birth earlier in the observation period and left military service soon after” and “did not represent a random sample.”


Of the 11,271 women whose paper records were reviewed, 1,318 had anthrax vaccinations recorded on paper. Of these 1,318, only 1,158 had anthrax vaccination recorded electronically. “When compared with those of [paper] medical records, the specificity of electronic data was 97.5%, and the sensitivity was 61.5% for correctly identifying [anthrax] vaccine recipients.” This means very few women whose paper records showed they did not get the vaccine were identified electronically as having received it, but that many who did receive it, according to their paper medical records, failed to have the vaccinations recorded electronically.


Could it be that the poorer sensitivity (ability to correctly identify those who received the vaccine) of the electronic database contributed to the drop in birth defect rates in offspring of those vaccinated women added to the database later? Given the demonstrated low sensitivity of the electronic record for identifying vaccinations, it is very likely that some of the women who had infants with birth defects failed to have their vaccinations accurately recorded in the electronic database. Since electronic records supplied more of the study data for women vaccinated later compared to earlier, relatively more late-vaccinated women would be expected to be incorrectly placed in the unvaccinated group. This could dilute the measured adverse vaccine effect in infants born later, compared with the data used in Ryan’s earlier assessment.


In other words, the greater use of the more accurate paper records, for those vaccinated earlier, would be expected to result in more accurate vaccination ascertainment for mothers of infants with birth defects who were born earlier.


Merging data from both paper and electronic records, despite low sensitivity for identifying anthrax vaccinations in the electronic records, is a major problem with this study. It may have led to loss of statistical significance for some comparisons despite a larger number of subjects in the later study.


Biologic Plausibility of Vaccine-Induced Birth Defects


Dr. Ryan stated that, “there is little information available to support biologic plausibility of anthrax vaccine as being teratogenic.” The CDC’s Morbidity and Mortality Weekly report, cited earlier, reported that, “although the Food and Drug Administration-licensed vaccine has not been suspected to be a hazard to reproductive health, no studies of animals or pregnant women have been conducted.” Shouldn’t studies in pregnant animals have been required prior to mandatory use of the vaccine in the young female military population? Then again, if no animal studies had been conducted, it was easy to say there was no information linking the vaccine to birth defects.


The anthrax vaccine contains formaldehyde, aluminum, benzethonium, 3 anthrax toxins that are known to exert profound biologic effects, and an array of uncharacterized substances derived from anthrax fermentation. Why is it implausible to think that injecting this mix might contribute to birth defects?


Dose-Response Relationship


Is there a dose-response effect from anthrax vaccine on the rate of birth defects, as might be expected if the vaccine were truly causing such problems? Ryan et al. state that “infants exposed to two or more maternal anthrax vaccine doses in the first trimester did not have a significantly increased risk of birth defects,” implying that there was no dose-response relationship.


But a positive dose-response relationship did exist. Infants exposed to two or more vaccine doses in the first trimester had an adjusted rate of birth defects 1.19 times higher than infants of mothers vaccinated at another time. But a smaller number of infants received 2 or more vaccinations prior to detecting a pregnancy, rather than just one, so this increased rate did not achieve statistical significance. More infants of multiply-vaccinated first trimester mothers would have been needed to achieve statistical significance.


Birth Defect Rates


In the US overall, 3.0% of infants are born with a significant birth defect. However, in military women vaccinated outside the first trimester, 4.2% of offspring had at least one major birth defect. In never-vaccinated women, 4.03% of offspring had a birth defect. In those vaccinated during the first trimester, 4.7% had at least one major birth defect.


When 4.7% is compared to 4.2%, the increase in the birth defect rate is only 12%. But consider the vaccine's makeup. Anthrax vaccine uses the aluminum adjuvant Alhydrogel to create a depot effect, slowly releasing vaccine antigen over some months following an injection. So women vaccinated prior to the first trimester could still be exposing their fetus to the vaccine. That is why they should not have been included in the control group: at least, those vaccinated in the year prior to conception should not have been used as controls.


Ryan’s data support the hypothesis of pre-pregnancy vaccination contributing to birth defects. In women vaccinated during the first trimester, the birth defect rate in offspring was 4.68%. In offspring of women vaccinated prior to pregnancy, the birth defect rate in offspring was 4.56%: nearly as high. In those vaccinated after pregnancy, the birth defect rate in offspring was only 3.85%.


By including pre-pregnancy vaccinations in the original control group, the effect of the vaccine on birth defect rates was diluted.


Now compare birth defect rates of infants born to women given anthrax vaccine during their first trimester with all infants born in the United States. When you compare 4.7% to 3.0% (the rate of major birth defects diagnosed in the first year for all infants) the unadjusted increase in the birth defect rate is 57%. This suggests that the increase in birth defects attributable to anthrax vaccine could be quite considerable. (However, these numbers have not been adjusted to take account of other possible differences between the two populations. On the other hand, military populations are considered healthier, on average, than their civilian counterparts, so the comparison is probably fair.)


Conclusions


In the 2008 paper’s conclusions, Dr. Ryan et al. note that additional research should be done, and make the excellent suggestion to perform active follow-up, such as that being performed by the Smallpox Vaccine in Pregnancy Registry. Ryan et al. also point out that potential adverse effects of vaccinations given pre-conception should be assessed.

But the sentence immediately following this recommendation states incorrectly that, “these analyses found no evidence that prepregnancy maternal anthrax vaccination is associated with an increased risk for birth defects.” Maybe the relationship was not significant, but the rate was considerably higher than for offspring of military women never vaccinated, of military women vaccinated postpregnancy, and for offspring of American women overall.


Ryan’s final sentence makes this reasonable point: “… women with no known exposure to inhalation anthrax should continue to avoid anthrax vaccination during pregnancy.”Overall, the data presented by Ryan et al. give cause for concern, and should provide the impetus to further study the issue of anthrax vaccine and birth defects. Most importantly, these data tell us that the military’s electronic database has serious flaws, and requires enhanced data entry and ongoing testing to assess the validity of the information it contains.


For future study of the effects of anthrax vaccine in pregnancy, paper records should be used. Furthermore, women receiving pre-pregnancy vaccinations, who were shown to have high rates of infants with birth defects, cannot be used as the control group in a scientifically valid study. Finally, an understanding of how and why birth defects rates in the offspring of military women exceed civilian rates must be sought.


No comments: