Last week, researchers at the Centers for Disease Control and Prevention (CDC) warned that an influenza vaccine given in the first trimester of pregnancy might have caused miscarriages. News outlets all over America picked up the story.
The CDC’s claim wasn’t trivial. About 50 percent of all pregnant women in the U.S. receive an annual influenza vaccine. Now, some expectant mothers were wondering whether they had done the right thing by getting their flu shots.
These women had no need to worry. For several reasons, this CDC study should have never been published. Why?
• The CDC’s observation was inconsistent. Researchers had studied two influenza-vaccine seasons: 2010-2011 and 2011-2012. The problem of first-trimester spontaneous abortions occurred during the first season but not the second.
• Six previous studies had failed to find any evidence that an influenza vaccine given during the first trimester had increased the risk of spontaneous abortions. In other words, the CDC study was an outlier.
• CDC investigators found the increased risk of spontaneous abortions occurred only in those who had also received an influenza vaccine the previous year. If women hadn’t received an influenza vaccine the previous year, the risk disappeared. The CDC’s question prior to this study was “Does influenza vaccine cause spontaneous abortions?” The answer to that question was no. It was only after investigators sub-stratified their data to include those who had or hadn’t received a vaccine the previous year that they could find statistical significance. Epidemiologists refer to this as cherry-picking, or fine tuning data—and it’s not meant as a compliment.
• After the CDC researchers had finished sub-stratifying their data, the numbers were small: 14 cases of spontaneous abortions in the vaccinated group and four in the unvaccinated group. The problem with small numbers is that they’re often misleading. For example, in a large study of children who had received thimerosal-free or thimerosal-containing vaccines, a statistically significant number of boys had tics. In the same study, a statistically significant number of girls displayed better cognitive function. In both groups, the numbers were small. These findings left researchers with the impossible task of explaining why thimerosal—an ethylmercury-containing preservative in vaccines—had caused tics in boys while at the same time making girls smarter. Said another way, if you flip a coin five times, the chance that it comes up heads each time is 32 to 1. Events occurring at odds greater than 20 to 1 are considered to be statistically significant (i.e., p≤0.05). If you flip the coin thousands of times, however, it’s likely that the coin will come up heads five times in a row somewhere in that run, thus proving that it’s a two-headed coin. Again: the curse of small numbers gleaned from a large database.
• The finding that the influenza vaccine might cause spontaneous abortions also didn’t make biological sense. For example, influenza infections exacerbate the symptoms of multiple sclerosis in people who are afflicted with the disease; on the other hand, influenza vaccination of these patients doesn’t exacerbate symptoms. These findings are consistent with the fact that influenza vaccination is far less capable of driving the immune system than natural infection. To their credit, the CDC researchers stated that, “the biological basis for our observations has not been established.”
Although reporters and editors did their best to soften the blow of this supposed finding with phrases like “possible association” or “possible safety signal,” anyone who saw these reports couldn’t help but be worried, especially if they only read the headlines. And if some women choose not to get an influenza vaccine during pregnancy, then the CDC researchers, in their compelling desire to be “transparent” with weak, biologically inexplicable data, will have done harm. Pregnant women are much more likely to be hospitalized and die from influenza than women of the same age who aren’t pregnant. Also, natural influenza infection during pregnancy increases the risk of spontaneous abortions, premature births, and very low birth weights. Further, more than in any other age group, infants under 6 months of age suffer severe complications or die from influenza. The best way to protect young infants from influenza is by maternal immunization, following which the mother passively transfers potentially life-saving, virus-specific antibodies to her baby before birth.
The authors of this paper concluded by stating, “This study does not and cannot establish a causal relationship between repeated influenza vaccination and spontaneous abortions.” Which is why it should never have been published. The authors also concluded that “further research is warranted.” One could argue that the research spurred by this paper should focus on how many women will now choose not to receive an influenza vaccine during pregnancy, putting both themselves and their unborn children at needless risk.
Paul A. Offit is a professor of pediatrics and director of the Vaccine Education Center at the Children’s Hospital of Philadelphia. His most recent book is Pandora’s Lab: Seven Stories of Science Gone Wrong (National Geographic Press, April 2017).