Chapter 3. Vaccines and Vaccine Safety

Vaccines represent one of the most successful public health interventions of all time. Diseases that once killed thousands of children each year have been virtually eliminated from the United States (Table 3–1). Because vaccines have been so effective, many parents and younger physicians have little firsthand experience with the infectious diseases they prevent. In this context, attention has shifted away from concerns of vaccine-preventable diseases themselves toward concerns of vaccine safety, both perceived and real.

Figure 3–1 graphically depicts what may occur if the public loses faith in the immunization system. Disease incidence begins to decline when a new vaccine is introduced. If there is loss of confidence in the vaccine among a critical proportion of the population, outbreaks may occur. Continued decrease in disease incidence with potential disease eradication can only occur if public confidence in the vaccine program is restored.

There are several historical examples of what may occur when immunization practices suddenly shift. For instance, the incidence of pertussis was found to be 10–100 times lower in countries that maintained high levels of whole-cell diphtheria–tetanus–pertussis (DTP) vaccination in the 1980s as compared to countries with prominent anti-DTP movements.1 In Japan during the mid-1990s, immunizations were made optional after the occurrence of rare case reports of aseptic meningitis associated with measles–mumps–rubella (MMR) receipt. Consequently, measles made a recurrence, resulting in more than 100,000 cases and 50–100 deaths per year.2

Even in the United States, where many parents feel that there is no risk for infectious diseases, outbreaks have occurred when individuals are unvaccinated. These outbreaks are more likely to occur in clusters of vaccine-resistant communities.3 For example, in Colorado there were 14 measles outbreaks between 1987 and 1998.4 Children who were exempted from vaccination were 22 times more likely to acquire measles as children who were vaccinated. Another outbreak occurred in Indiana in the summer of 2005 when 34 members of a church acquired measles after an unvaccinated member returned from a missionary trip to Romania. 32 of the 34 affected individuals were unvaccinated, primarily due to concerns of vaccine safety.5 As the number of unvaccinated individuals increases, the potential for a large-scale outbreak in the United States becomes greater. In 2008, 127 cases of measles in 15 states were reported to the Centers for Disease Control and Prevention (CDC) as of the end of June, mostly in unvaccinated individuals.58 Although these data are for the first half of the year alone, this already represents the greatest number of cases reported since 1997, when 138 people were diagnosed with measles.

In this chapter, we review the process of assessing vaccine safety. We then review the biologic and epidemiologic data surrounding popular myths about vaccines.

Vaccines are unique among pharmaceutical agents in that they are given to healthy children to prevent disease in the future. This poses several challenges. First, if parents have even the slightest concern about the safety of a vaccine, the perceived risk of vaccination may outweigh the perceived benefit of protection against a disease that is no longer prevalent. Second, because nearly all children are vaccinated, it is difficult to determine whether adverse events that occur after vaccination are truly as a result of vaccine receipt or mere coincidence. While vaccines are highly efficacious, they can prevent only a small percentage of infectious diseases, and do not prevent other childhood adverse outcomes.6 This may be confusing when diseases with uncertain etiology such as autism are diagnosed at the same age as most vaccines are given. Finally, most vaccines are given simultaneously with other vaccines and it can be difficult to determine which specific vaccine, if any, is responsible for a given adverse event.

Randomized controlled trials (RCTs) provide the strongest epidemiologic evidence of a causal relationship. However, postlicensure RCTs are not ethically feasible once a vaccine becomes recommended for routine pediatric use. Therefore, vaccine safety has traditionally been studied as part of prelicensure RCTs and further monitored using postlicensure surveillance.

Prelicensure RCTs are usually designed to assess vaccine efficacy and may be underpowered to detect uncommon vaccine adverse effects. A notable exception is the prelicensure trials surrounding the newly approved pentavalent bovine rotavirus vaccine (RotaTeq), which enrolled nearly 70,000 children.7 This large study was designed because of safety issues with an earlier rotavirus vaccine (Rotashield), which was found to be associated with intussusception in an estimated 1 in 10,000 vaccine recipients.8 Because the Rotashield prelicensure study included only slightly more than 10,000 children, it was not large enough to detect this adverse effect. Ideally, one would simply increase the size of vaccine prelicensure RCTs. However, some vaccine side effects occur too rarely to be detected in an RCT. For instance, a trial designed to detect a threefold increase in relative risk for an outcome with a background incidence of 0.01 would require a two-arm trial with 175,000 children.9 Even if it were possible to design a study of this size, it might not be financially feasible.

Postlicensure monitoring of vaccine safety uses the Vaccine Adverse Event Reporting System (VAERS). VAERS is a passive postlicensure reporting system, maintained jointly by the Food and Drug Administration (FDA) and the Center for Disease Control (CDC). It allows any individual who believes that a vaccine-associated adverse effect has occurred to report this to the system. The strength of VAERS is that it incorporates data from the entire country and is therefore widely generalizable.6 However, VAERS also has several limitations. First, simply reporting an adverse effect does not imply causation. In fact, common, acute symptoms such as fever occur in up to 10% of healthy children regardless of vaccination status and so the observed association may be attributed to chance alone.10 Second, there is no denominator that can be used to calculate the incidence of adverse effects. While the total number of delivered doses can be obtained from pharmaceutical companies, it is difficult to determine how many doses have actually been administered. Furthermore, the distribution of vaccine across age groups, which would be helpful for calculating age-specific incidence rates, is unknown.6 Finally, there is potential for significant bias in reporting. In one recent analysis it was found that a large percentage of VAERS reports were filed by attorneys involved in lawsuits against vaccine companies.11 Similarly, parental and physician reports may be influenced by public perception. In another study, parents who reported autism as a vaccine-associated adverse effect to VAERS were much more likely to do so after the publication of a paper that claimed a link between MMR and autism, even though their children were vaccinated several years before.12 On the other hand, there may be significant underreporting to VAERS. While there are data to suggest that serious adverse events such as intussusception and death are reported,13 more common and benign events that occur after vaccination may be overlooked by parents and physicians. Despite these limitations, VAERS is able to detect potential vaccine-associated adverse events that may be further evaluated using other epidemiologic methods.

One such method uses the Vaccine Safety Datalink (VSD), which was created in 1990 and is modeled after other pharmacoepidemiologic-linking studies.14 It is partnered with eight large managed care organizations (MCOs), and prospectively collects data on 5.5 million individuals each year.15 Vaccinations are entered into the medical record as part of routine medical care, and any subsequent medical history, including adverse events, can be detected. This allows for a true denominator that may be used to calculate the incidence of adverse reactions after vaccination. While this system offers a higher standard of epidemiologic data than VAERS, it is still unable to account for the fact that most children in the United States are vaccinated.16 Another limitation is that it only includes data on children enrolled in managed care organizations, who may not be representative of the entire population. Despite the lack of unvaccinated children, the VSD can be used to determine the risk of acute vaccine adverse effects using the case-crossover study design, in which cases serve as their own controls.6 Several of the studies included in this chapter are based on VSD data. In the sections below, we will focus on specific concerns about vaccine safety.

Measles–Mumps–Rubella Vaccine

In February 1998, researchers in Britain suggested that receipt of the MMR vaccine may cause autism.17 This was based on a small series of 12 children, all of whom had inflammatory bowel disease and 8 of whom also had autism. This study was significantly flawed. First, case reports do not offer strong proof of causal association. Second, the exposure—MMR vaccination—relied on parental recall. Because these parents believed that MMR was responsible for their children's autism, it is not surprising that they reported a temporal association between MMR vaccination and the development of autistic symptoms. More importantly, the lead author was funded by a group of lawyers who were representing the families of the eight autistic children in a lawsuit against the manufacturer of MMR.18 Ten of the original 13 authors eventually retracted their statement of causality.19

Despite these limitations, this story was widely published in the popular media and on the Internet. In Britain, national MMR immunization rates fell from 92% to 75%, which resulted in several measles outbreaks and the first death caused by measles in a decade.20 In the United States, immunization rates have remained fairly high because of school-entry requirements, yet pediatricians do report that parents are concerned about the possible association between MMR and autism. These concerns led the Institute of Medicine (IOM) to investigate the potential relationship between MMR and autism.21,22 In 2001, and again in 2004, the IOM concluded that there was sufficient data to reject the hypothesis that MMR caused autism.

These decisions were based in part on several large epidemiologic studies. The most rigorous study was performed in Denmark in 2002.23 The investigators incorporated data from 537,303 children born in Denmark from 1991 to 1998, for a total of 2,129,864 person-years. Of these, there were 1,647,504 person-years of follow-up for children who received MMR and 482,360 person-years of follow-up for children who did not receive MMR. After adjusting for age, calendar year, sex, birth weight, gestational age, maternal education, and socioeconomic status, there was no significant difference in rates of autism (relative risk [RR] 0.92; 95% confidence interval [CI], 0.68–1.24) or other autistic-spectrum disorders (RR 0.83; 95% CI, 0.65–1.07) between children who had received MMR and those who had not.

Other research has relied upon ecologic analyses.24 Two studies compared rates of autism before and after the introduction of national MMR immunization programs, and found that the prevalence25 and incidence26 of autism decreased after the introduction of MMR. Other studies employing time-series approaches found that trends in the number of children diagnosed with autism did not parallel trends in MMR coverage.27–30 If MMR did cause autism, autistic children who received MMR might have developed symptoms at an earlier age than autistic children who were unvaccinated. This does not appear to be the case.23,26,29

Shortly after the MMR–autism controversy began, a new concern about vaccines and autism emerged. In 1999, the FDA released a report suggesting that the levels of ethyl mercury, a metabolite of the preservative thimerosal that is used in some childhood vaccines, exceeded acceptable levels as determined by the Environmental Protection Agency (EPA). Based on preliminary data that suggested an increasing trend in autism at the same time as the emergence of new thimerosal-containing vaccines, the Advisory Committee for Immunization Practices and American Academy of Pediatrics (AAP) issued a statement recommending that thimerosal be removed from all childhood immunizations.31 At that time, it was felt that the risk of a potential association outweighed the benefit of using thimerosal-containing vaccines. However, the Environmental Protection Agency data are based on data for methyl mercury, a common environmental toxin. In contrast, thimerosal is metabolized to ethyl mercury, which has different pharmacokinetics and is excreted from the body much more quickly.32 Despite the removal of thimerosal from all childhood vaccines in 2001 (except for the injectable influenza vaccine), rates of autism in the United States continue to increase, making it implausible that thimerosal causes autism. Finally, there have been several large epidemiologic studies that suggest that thimerosal exposure is not associated with autism.

One study based on the VSD-incorporated data from 124,170 children vaccinated at three health maintenance organizations (HMOs).33 In one HMO, researchers discovered a statistically significant association between cumulative thimerosal exposure and tics at 3 months, but not at 1 or 7 months. This association was not seen at the other HMOs. At the second HMO, there was a statistically significant association with language delay at 3 and 7 months. There were no associations observed between thimerosal exposure and any other developmental conditions, including autism and attention-deficit disorder. In a third HMO, consisting on 16,717 children, no statistically significant associations were noted between thimerosal and any developmental diagnoses. Because there was no consistent relationship between thimerosal exposure and developmental disorders, it was concluded that the two observed statistical associations were attributed to chance alone.

A study performed in Denmark found no association between autism and ethyl mercury exposure.34 This study took advantage of the fact that Danish children had received thimerosal in vaccines only until 1992. Therefore, the authors were able to compare rates of autism among children who were exposed to significant levels of thimerosal, some thimerosal, and no thimerosal. The authors studied thimerosal exposure in three separate analyses. Comparing any thimerosal receipt to no thimerosal receipt, there was no association with autism. When these results were stratified by dose of thimerosal, there was still no association. Finally, when thimerosal was treated as a continuous variable there was also no association.

Another study performed in Denmark found that the discontinuation of thimerosal-containing vaccines was actually associated with an increase in autism.35 Finally, a study combining data from California, Sweden, and Denmark found that rates of autism continued to increase after discontinuation of thimerosal-containing vaccines.36 Based on these data, and others, the IOM concluded in 2004 that there was sufficient evidence to reject the hypothesis that thimerosal is causally associated with autism.22

In a nationally representative telephone survey conducted in 1999, 23% of parents reported concern that children receive too many immunizations and 25% of parents believed that their child's immune system could be weakened as a result of too many immunizations.37 Since the publication of this report, even more vaccines have been added to the immunization schedule.38 Despite these new vaccines, most childhood infections are not vaccine-preventable.39 Therefore, vaccines represent only a small proportion of the antigenic burden to the developing immune system. Additionally, while the number of vaccine-preventable childhood diseases has increased from 1 (smallpox) in 1900 to 13 in 2007, the total number of antigens to which children are exposed in current vaccines is significantly less than it had been for most of the twentieth century. This is because of the discontinuation of the whole-cell pertussis vaccine, which contained 3000 proteins; and the smallpox vaccine, which contained 200 proteins.40 In contrast, the combined antigenic burden of all current childhood vaccines is less than 200 proteins or polysaccharides.40 Another concern is that receipt of immunizations may actually weaken the immune system. This does not appear to be the case. In fact, children who are vaccinated may actually be protected against infections that are not included in the childhood vaccines.41,42 Additionally, several studies have shown that vaccine immunogenicity is not altered by giving multiple vaccines at the same time.40

Another common misperception shared by parents and medical providers is that children should not receive vaccinations when they are sick. One study did find decreased serologic evidence of measles immunity in children with upper respiratory illness (URI) at the time of vaccination as compared to asymptomatic children.43 However, the results from this study have not been replicated. Two subsequent studies44,45 found no difference in seroconversion rates between children with afebrile URI and asymptomatic children. Finally, a prospective cohort study including 356 children compared seroconversion after MMR vaccination in asymptomatic children with children who had either URI, otitis media, or diarrhea, excluding children with temperature greater than 37.4°C.46 The authors found no significant difference in seroconversion between these groups. In fact, children with mild illness had slightly higher rates of seroconversion, although this did not reach statistical significance. While it is recommended that immunizations be deferred for children who have severe infections requiring hospitalization, this is not because of concerns about vaccine effectiveness. Instead, vaccine deferral is recommended to decrease the likelihood of an adverse event associated with hospitalization being misattributed to vaccination.40

There is also concern that vaccines may be responsible for allergies and the development of asthma. This theory is rooted in the “hygiene hypothesis,” which suggests that children with better hygiene are more likely to develop allergies.47,48 Vaccines, some argue, prevent natural infection with viruses and bacteria that would normally stimulate the immune system. However, most infections that occur during the first year of life are caused by viruses that are not prevented by vaccines.39,47 Furthermore, all infants are exposed to environmental antigens such as pollen, cigarette smoke, dust mites, and pet dander that also contribute to immune system development.39 Finally, several studies have failed to find an association between vaccination status and development of allergies or asthma.

One such study used the Vaccine Safety Datalink (VCD) to assess risk factors for asthma in 167,240 children.49 In multivariable regression, receipt of MMR, DTP, and OPV were not associated with asthma. Haemophilus influenza type B (RR 1.18; 95% CI, 1.02–1.36) and hepatitis B (RR 1.20; 95% CI, 1.13–1.27) vaccines were associated with a small increase in relative risk for asthma. Infants who were enrolled in but did not receive long-term medical care in the managed care organizations would appear to be unvaccinated and would also have no record of asthma. This might artificially decrease the risk of asthma among unvaccinated children. A subanalysis including only those children with two or more health care visits was performed to correct for this bias, and revealed no association between any vaccines and asthma. Another study of allergic responses within a randomized controlled pertussis vaccine trial found no association between vaccine receipt and allergic diagnoses.50 Finally, a recent ecologic study demonstrated that the increase in asthma seen in the United States during the 1990s occurred before the increase in the number of recommended vaccines that occurred in the late 1990s.51 This further suggests that the increasing numbers of vaccines are not responsible for observed increases in asthma during the last decade.

In this chapter, we have described the mechanisms in place for monitoring vaccine safety in the United States and have addressed some of the more common myths surrounding childhood immunizations. This information may be useful during discussions with parents who have concerns about vaccine safety. Additional strategies for discussing immunizations with vaccine-hesitant parents have recently been published by the American Academy of Pediatrics, 52,53 and are summarized in Box 3–1. Despite increasing coverage of vaccine controversies in the media and on the Internet, physicians remain the most influential source of immunization information for parents,37,54,55 including those who believe that vaccines are unsafe56 and those who request exemptions.57 Physician familiarity with vaccine safety issues and vaccine risk communication is important in order to maintain optimal immunization rates.