Cribsheet

by Emily Oster

  A surprising aspect of vaccine resistance is that it tends to be stronger in areas with more educated parents. For most health outcomes—heart disease, obesity, diabetes—more educated people tend to be healthier. But in the case of vaccines, the correlation often goes the other way. Areas with more educated parents actually have, on average, lower vaccination rates.2 This suggests it is not necessarily lack of information getting in the way of choosing vaccination.

  The scientific consensus on vaccinations is extremely clear: vaccines are safe and effective. This conclusion is supported by a very wide range of doctors and medical organizations, and by government and non-government entities. But despite this, there are parents who choose not to vaccinate, and many of them are well educated and have thought about the decision. It is worth, therefore, at least visiting the evidence.

  BACKGROUND

  There have always been people who distrusted vaccines. A colleague of mine at Brown, Prerna Singh, studies resistance to vaccination—in this case, for smallpox—in China and India when the vaccine was first introduced. In that context, the concerns were focused on the harm the vaccine might cause, and the feeling that they might not prevent the disease anyway.

  The most well-known concerns about vaccines at the moment relate to a possible link with autism, but there is an earlier round of vaccine-danger concerns dating from the 1970s. During this period, a series of case reports suggested that the pertussis vaccine—which prevents whooping cough and is given as part of the DTaP vaccine—might be linked to infant brain injury. It was subsequently revealed that this link was not supported in the data, but in the wake of the initial suggestion, there was a round of lawsuits filed against vaccine manufacturers.

  The threat of these lawsuits was sufficient to almost completely shut down production of this vaccine. Vaccine prices rose and availability tanked. Lack of access to the vaccine presented a significant public health risk. In 1986, in response to this, Congress passed the National Childhood Vaccine Injury Act, which protected companies from being sued over mandated vaccinations. People who claim to be injured by vaccines can appeal to the federal government for compensation, but they cannot seek damages from the vaccine manufacturer.

  A somewhat unfortunate side effect of this (sensible) policy solution is that it seems to imply that vaccine injuries are a real and substantial risk. (It might have been better to name the policy something else.) In practice, lawsuits brought by people reacting to flawed research were the motivation for the passage of the act, not any actual risks posed by vaccines. This policy is still in effect, and it unfortunately gives some background support to contemporary claims that vaccines are risky.

  The latest round of vaccine resistance was tipped off by a former doctor (“former,” since he subsequently lost his license) named Andrew Wakefield.3 In 1998, Wakefield published a paper in the Lancet—a highly regarded medical journal—that suggested a link between autism and vaccines.4 The paper is a summary of twelve case studies. The twelve children studied all had autism, and the paper claimed that in at least eight—and possibly more—of the twelve cases, the symptoms of autism began more or less immediately after the child received the measles, mumps, and rubella (MMR) vaccine.

  Wakefield provided a hypothesized mechanism linking the two, related to digestive health.

  First point: The conclusion of this paper is wrong. Other evidence, better evidence, from both before and after this article was published refutes this link. I review some of this in the following pages. Indeed, a vague case study of twelve children is hardly strong evidence in the first place, so it’s not surprising that it didn’t hold up.

  But it turns out that the paper was also fraudulent. The children included in the sample were not—as Wakefield stated—all the kids who could have been included. Wakefield specifically chose children who supported his conclusion. In addition, many of the facts of the particular cases were falsified. Details were changed to make the onset of autism symptoms seem closer in time to the vaccinations. When, in reality, the onset of symptoms was six months or more after the vaccine, the reported case details suggested it was within a week or two.

  Why would Wakefield do this? It turns out he was planning a lawsuit against vaccine manufacturers, and this would be part of the evidence. His motivation was the oldest reason in the book: money.

  In 2010 the Lancet retracted the article, and Wakefield was stripped of his medical license. But the damage was done and Wakefield has never admitted the article was fraudulent or apologized. He continues to travel the world, hawking his discredited theories. The Somali immigrant community with the measles outbreak? They had had two visits from Wakefield over the preceding years.

  Among the most insidious aspects of this episode is that it revived general concerns that vaccines are unsafe. Some people do not believe the link with autism, but still feel that vaccines may cause some other kind of injury. Antivaccination websites cite concerns about, for example, aluminum in vaccines, and also the general feeling that activating the immune system can cause brain injury.

  These antivaccination websites seem evidence based; they cite papers and studies to support their position. On the other side, organizations like the Centers for Disease Control and the American Academy of Pediatrics assure people that vaccination is safe. A downside of their approach, however, is that they rarely confront the antivaccination literature head-on. There is little effort to explain why the papers cited on antivaccination websites are problematic (if they are). It can end up seeming like the antivaccination side is serious and evidence based, and the pro-vaccine side is just dismissively insisting you trust them.

  This is not the case. The recommendations of the AAP, among others, are based on careful and complete evaluation of all the possible risks of vaccination.

  VACCINE SAFETY

  In 2011 the Institute of Medicine (IOM) published a nine-hundred-page tome entitled Adverse Effects of Vaccines: Evidence and Causality.5 (I know what you’re thinking: beach reading!)

  The book is the product of years of work from a large number of researchers and practitioners. They were tasked with a daunting job: to evaluate the evidence for linkages between common vaccinations and a very large set of possible “adverse events.”

  They evaluated the evidence—from more than twelve thousand papers— on 158 vaccine-adverse event combinations. What does this mean? For each vaccine, the authors looked for evidence on a possible link between that particular vaccine and any claimed risk. The risks here are referred to as adverse events. So the authors looked for, say, evidence on a link between the MMR vaccine and seizures.6

  What kind of evidence did they look for?

  First, there are adverse-event reports: the CDC collates all reports of adverse events that people (parents, doctors, etc.) attribute to vaccination. You can explore this yourself online: searching for reports of links between MMR vaccine and autism yields a large number of reports from parents who claim their child developed autistic symptoms shortly after receiving the vaccine. You might have the instinct that these reports are enough to at least prove some link between vaccination and the outcome—but evidence of this type is tenuous at best.

  Consider the following: Imagine that people believed that cutting an infant’s fingernails was medically dangerous—that it led to illness or other complications. And imagine we set up an adverse-event reporting system for fingernail cutting.

  In all likelihood, you’d get all kinds of reports. There would be parents saying that the day after they cut their infant’s fingernails, the baby came down with a terrible fever. Others would say they had a very liquid-looking poop. You’d get reports of children who didn’t sleep well for days after the fingernail cutting, and others about babies crying uncontrollably for hours.

  These would all be true things that happened. But they would not be causally linked to the fingernail cutting! Sometimes infants get a fever; sometimes the
y have weird poops. Most babies do not sleep, and others cry a lot. In order to figure out whether there was any real link, you’d need to know the general base rate of these events—how likely people are to report them when there was no fingernail cutting. But that isn’t something we have a reporting system for. There is no website where you can report every time your kid has an unusual poop.

  You’d have to try to piece together whether these adverse events really seem more common among babies whose nails are cut than those whose aren’t. This is especially hard for things that happen all the time, like “baby cried.”

  In your fingernail-reporting system, you probably would also learn something. You’d get a lot of reports of finger injury—cuts in need of Band-Aids. This is not something that happens all the time, and there is an obvious mechanism for the connection with nail cutting. So you would probably conclude that fingernail cutting is linked to accidental finger cutting, which is true (Penelope is at least one case report).

  But how do we know the finger cutting is a real effect and the fever is not? How can we use evidence like this?

  In the IOM report, the authors used reports like this in combination with evidence on mechanisms. Is there a biological reason to think this relationship would exist? In some cases, the biological link was so plausible that researchers drew conclusions based only on these adverse-event reports. In others, without a mechanism, they required more evidence to draw conclusions.

  The second major piece of evidence comes from “epidemiological studies,” which, in this case, compare children who are vaccinated with those who are not. These are typically not randomized, but they can be very large. If the adverse events reported are backed up by relationships in the population overall, this may support a link, even if the mechanism is not obvious.

  The authors of this report classified each of their 158 possible links into one of four categories: convincingly supports (there is a convincing causal relationship between the vaccine and the event), favors acceptance (there is probably a causal relationship), favors rejection (unlikely, based on the available evidence), or insufficient evidence.

  For the vast majority of these links, the evidence is insufficient. This includes things like the link between the MMR vaccine and multiple sclerosis onset, or between DTaP vaccines and SIDS. In these cases, the authors could find no good evidence to support the link, but also no evidence to firmly refute it. This doesn’t always mean there is no evidence. In most cases, there is some report linking the events from the adverse-event reporting system. But when the authors looked into it, it seemed unlikely that the two were related.

  This is a somewhat frustrating conclusion. Basically, whatever you thought before (in statistics speak, your “prior beliefs”) is what you’ll think after seeing the evidence. If people come in thinking vaccines are safe, then there is nothing here to argue against that. Conversely, if they come in thinking vaccines are unsafe, there is nothing here to help refute that. For people who really want to believe that vaccines are damaging, this nonevidence may be seen to support their beliefs—as in, “We cannot rule out a link between MMR and multiple sclerosis.” Based on this standard, you cannot rule out a link between fingernail cutting and multiple sclerosis. The only difference is that no one believes the latter link exists in the first place.

  In general, it is very difficult to prove there is no relationship between two events. If we are worried about a very small relationship, we’d need huge sample sizes to statistically reject it. We don’t often have these. It would be great to have more evidence, but the IOM can only work with what they have.

  Of the seventeen cases where the IOM thought they could draw conclusions, fourteen were judged to either convincingly support a relationship or favor acceptance. This may seem scary, but it is important to look carefully at what the risks are.

  First, for many of the vaccines (all but the DTaP vaccines), there is a risk of allergic reaction. This is extremely rare (about 0.22 in 100,000 vaccines) and can be treated with Benadryl or, in an extreme case, an EpiPen. Allergic reactions account for half the documented risks in the report.

  Second, fainting sometimes occurs after vaccination, mostly among adolescents. It is unclear what the mechanism is, but fainting does not have long-term consequences. This accounts for another two of the convincingly supported risks.

  There are then several cases in which vaccines are linked to more serious risks. However, in these cases, the risks are generally extremely tiny. An example is the link between the MMR vaccine and “measles inclusions body encephalitis.” This condition is a very serious long-term complication of measles infection that occurs in people who are immune-compromised. It is very rare, nearly always fatal, and is a well-known complication of actual measles infection. The question for the IOM report was whether someone could also get this after measles vaccination. In the report, the authors examine three cases in which subsequent testing of children diagnosed with this disease showed that they were very likely exposed to measles through vaccination, not through an actual case of measles.

  Given this evidence—that we know this to be a risk of the measles virus, and that the children in these three cases weren’t exposed to actual measles—the report concluded that in these cases, it is likely the vaccine caused the disease.

  This relationship is categorized as “convincingly supports.” It is very important to be clear, though, that this doesn’t mean this is a risk everyone should be concerned about. It arises only for children who are immune-compromised, and even then it is vanishingly rare. There are just three case reports in the history of vaccination. If your child has an immune issue, you’ll know, and you’ll talk through vaccination with your doctor. For healthy children, this simply isn’t a risk you should consider in your vaccine calculus.

  Similar issues arise for immune-compromised children who get the chicken pox vaccine. Again, these complications are extremely rare. There is a vaccine link here, but this is far from saying these are scenarios you should be actively worried about it. They are not.

  There is, finally, one vaccine risk that is more common and, while not serious, can be scary. Specifically, the MMR vaccine is linked with febrile seizures—seizures that occur in infants or young children in association with a high fever. They typically do not have long-term consequences, but are very scary in the moment.

  These are common enough that we can study their relationships to vaccines using large datasets of children. About 2 to 3 percent of children in the US will have a febrile seizure before they are five years old (most of these are not vaccine associated).7 A number of studies find that these seizures are about twice as likely in the period ten days or so after the MMR vaccine.8 They are actually more likely for children who get their first MMR dose later (i.e., older than one year); this is a reason to vaccinate on time, rather than to delay.

  One thing the IOM report does not cover is infant crankiness, which, as your doctor will probably tell you, is a result of vaccines for many babies. I learned about this link the hard way. We inexplicably scheduled a large student brunch at our house for a few hours after Penelope’s first vaccinations. We also failed to have any infant Tylenol stocked. Jesse ended up serving pastries to the students in our dining room on his own, while I wrestled a hysterically screaming baby into the Baby Bjorn for a walk to CVS. Not our finest afternoon. Still, by the next morning, the storm had passed.

  This crankiness—often accompanied by a fever—may be annoying, but it is not something to worry about. Your baby is working to create antibodies to a virus, and this work has some side effects. But not ones to be concerned about. Just make sure you have infant Tylenol around.

  This covers the data-supported risks of vaccines. What about the relationships that are not supported in the data? The IOM report explicitly rejects several links. One of them is the link between the MMR vaccine and autism, the link suggested by Andrew Wakefield in his Lancet pap
er.

  There are a number of big studies of this relationship. The largest of them includes 537,000 children—all the children born in Denmark from 1991 to 1998. In the Danish data, the authors were able to link vaccination information to later diagnosis of autism or autism-spectrum disorders. They found no evidence that vaccinated children are more likely to be autistic; if anything, the results suggest vaccinated children are less likely to be diagnosed with autism.9

  There are many similar studies; some are included in the IOM report, others postdate it. One study focuses on children who have an older sibling with autism and who are therefore more likely to have it themselves. Again, researchers found no link with the MMR vaccine.10

  There is no mechanism by which this would occur, and controlled studies in monkeys also show no plausible relationship.11 At the end of the day, there is simply no reason to think autism and vaccinations are linked.12

  It is not fair to say there are no risks associated with vaccination at all. Your child may well get a fever. It is also possible (although really quite unlikely) that this fever would lead to a seizure. It is also possible (although, again, very, very unlikely) that they could have an allergic reaction.

  But it is reasonable to say there is no evidence of significant long-term consequences of vaccines for healthy children.

  VACCINE EFFICACY

  Those of us in the US are lucky to live in a place where most people do get vaccinated, and cases of vaccine-preventable disease are rare. Few children get measles or mumps, and a few more get pertussis, but not many. If people stopped vaccinating, this would not be true anymore. All these diseases exist around us, and in the absence of vaccination, infection would be common.