pediatrics
June 2014, VOLUME133 /ISSUE 6

Timely Versus Delayed Early Childhood Vaccination and Seizures

  1. Simon J. Hambidge , MD , PhD a , b , c , d ,
  2. Sophia R. Newcomer , MPH a ,
  3. Komal J. Narwaney , MD , PhD a ,
  4. Jason M. Glanz , PhD a , d ,
  5. Matthew F. Daley , MD a , c ,
  6. Stan Xu , PhD a ,
  7. Jo Ann Shoup a ,
  8. Ali Rowhani-Rahbar , MD , PhD e ,
  9. Nicola P. Klein , MD , PhD f ,
  10. Grace M. Lee , MD , MPH g , h ,
  11. Jennifer C. Nelson , MPH i ,
  12. Marlene Lugg , DrPH j ,
  13. Allison L. Naleway , PhD k ,
  14. James D. Nordin , MD , MPH l ,
  15. Eric Weintraub , MPH m , and
  16. Frank DeStefano , MD , MPH m
  1. aInstitute for Health Research, Kaiser Permanente Colorado, Denver, Colorado;
  2. bDepartment of Community Health Services, Denver Health, Denver, Colorado;
  3. cDepartment of Pediatrics, University of Colorado School of Medicine, Aurora, Colorado;
  4. dDepartment of Epidemiology, University of Colorado School of Public Health, Aurora, Colorado;
  5. eDepartment of Epidemiology, School of Public Health, University of Washington, Seattle, Washington;
  6. fKaiser Permanente Vaccine Study Center, Oakland, California;
  7. gDepartment of Population Medicine, Harvard Pilgrim Health Care Institute and Harvard Medical School, Boston, Massachusetts;
  8. hDivision of Infectious Diseases and Department of Laboratory Medicine, Boston Children's Hospital, Boston, Massachusetts;
  9. iGroup Health Research Institute, Seattle, Washington;
  10. jDepartment of Research and Evaluation, Southern California Kaiser Permanente, Pasadena, California;
  11. kKaiser Foundation Hospital Center for Health Research, Kaiser Northwest, Portland, Oregon;
  12. lHealth Partners Research Foundation, Minneapolis, Minnesota; and
  13. mImmunization Safety Office, Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia

Abstract

BACKGROUND: Little is known regarding the timing of childhood vaccination and postvaccination seizures.

METHODS: In a cohort of 323 247 US children from the Vaccine Safety Datalink born from 2004 to 2008, we analyzed the association between the timing of childhood vaccination and the first occurrence of seizure with a self-controlled case series analysis of the first doses of individual vaccines received in the first 2 years of life.

RESULTS: In infants, there was no association between the timing of infant vaccination and postvaccination seizures. In the second year of life, the incident rate ratio (IRR) for seizures after receipt of the first measles-mumps-rubella vaccine (MMR) dose at 12 to 15 months was 2.65 (95% confidence interval [CI] 1.99–3.55); the IRR after an MMR dose at 16 to 23 months was 6.53 (95% CI 3.15–13.53). The IRR for seizures after receipt of the first measles-mumps-rubella-varicella vaccine (MMRV) dose at 12 to 15 months was 4.95 (95% CI 3.68–6.66); the IRR after an MMRV dose at 16 to 23 months was 9.80 (95% CI 4.35 –22.06).

CONCLUSIONS: There is no increased risk of postvaccination seizure in infants regardless of timing of vaccination. In year 2, delaying MMR vaccine past 15 months of age results in a higher risk of seizures. The strength of the association is doubled with MMRV vaccine. These findings suggest that on-time vaccination is as safe with regard to seizures as delayed vaccination in the first year of life, and that delayed vaccination in the second year of life is associated with more postvaccination seizures than on-time vaccination.

  • vaccine safety
  • immunization
  • vaccine
  • seizures
  • vaccine delay
  • Abbreviations:
    ACIP
    Advisory Committee on Immunization Practices
    CL
    confidence limit
    DTaP
    diphtheria, tetanus, and acellular pertussis vaccine
    DTaP-IPV-HIB
    diphtheria, tetanus, acellular pertussis, inactivated poliovirus, and Haemophilus influenza type B combined vaccine
    ED
    emergency department
    IRR
    incident rate ratio
    MCOs
    managed care organizations
    MMR
    measles-mumps-rubella vaccine
    MMRV
    measles-mumps-rubella-varicella vaccine
    PPV
    positive predictive value
    SCCS
    self-control case series
    VSD
    Vaccine Safety Datalink
  • What’s Known on This Subject:

    Reasons for childhood immunization delay include parental intent and barriers such as transportation. To date there has been 1 study of the association of delayed vaccination and seizures, which found measles-mumps-rubella and measles-mumps-rubella-varicella vaccines are both associated with a higher rate of seizures if received after 15 months of age.

    What This Study Adds:

    Our study found no association between the timing of vaccination and occurrence of seizures in the first year of life. By using different methods, our results support the observation that delaying vaccination with measles-containing vaccines past 15 months of age increases the incidence of postvaccination seizures.

    Despite the evidence for the safety of childhood vaccines, 1 3 an increasing number of families are requesting delayed immunization schedules for their young children, 4 often out of concern that the schedule recommended by the Advisory Committee on Immunization Practices (ACIP) 5 may confer risks for their children. To date, although there are multiple studies detailing the risk of a variety of vaccine-preventable diseases in children who are undervaccinated, 6 8 there are few studies directly comparing vaccine safety in children on delayed versus recommended immunization schedules. 9 The Institute of Medicine has recently called for an assessment of studies related to the safety of the recommended versus nonstandard schedules. 2

    Children may be on delayed schedules because of parental intent, or because of barriers to immunization, such as lack of health insurance and transportation. 10 14 Regardless, there is no reason to think a priori that vaccine adverse events will differ based on the underlying reason for a child being on a delayed schedule. In fact, a recent large cohort study demonstrated that emergency department (ED) use was roughly equivalent in undervaccinated children compared with those vaccinated on time. 15 We used a previously defined large national cohort of children on both recommended and delayed immunization schedules 15 to examine risk for seizures after vaccination in young children. Specifically, we asked the following questions: Is there an association between seizures and receipt of the first dose of each vaccine administered in the first 2 years of life? Does the magnitude of any association differ in children who received vaccinations on time versus on a delayed schedule? These questions are particularly relevant for vaccines, such as measles-containing vaccines, that have known associations with postvaccination febrile seizures. 9 , 16 21

    Methods

    Setting and Population

    We used a previously described cohort 15 from the pediatric population of the Vaccine Safety Datalink (VSD), 22 a collaborative project between the Centers for Disease Control and Prevention and several managed care organizations (MCOs) from across the United States that cover >3% of the US population. The MCOs offer similar preventive service packages and age-specific delivery of childhood vaccines. The study period was 2004 through 2010. The initial cohort (Fig 1) consisted of any child born between 2004 and 2008, continuously enrolled in 1 of 8 VSD MCOs from 2 to 12 months of age and up to 24 months of age, and who had at least 1 outpatient visit within the MCO. This study was approved by the institutional review boards at all participating sites and at the Centers for Disease Control and Prevention. Children older than 24 months were excluded, as very few vaccines are administered in the VSD cohort in the third and fourth years of life, resulting in very few vaccinated cases to analyze in this age group.

    FIGURE 1

    Cohort for self-controlled case series analysis. aVaccination status was assessed until the time of the first seizure by using the average number of days undervaccinated metric. 15 Children vaccinated on time had average number of days undervaccinated = 0 and children on a delayed schedule had average number of days undervaccinated >0. bExcluded children (n = 171) who had a diagnosis of seizure before 93 days of life, making them ineligible for a delay.

    For this study, we first identified any child with an International Classification of Diseases, Ninth Revision, Clinical Modification code for seizure (345.x and 780.3x, based on previous published work 16 ) in the ED or hospital between 38 days and 730 days (2 years) of life. We excluded time before 38 days, as we did not want to identify neonatal seizures that would have occurred before the earliest age that an infant should receive the recommended 2-month immunizations. 5 We next excluded any child who had a diagnosis for newborn convulsions or myoclonus, so as to exclude children with chronic seizure disorders. The final analytic cohort consisted of 5667 children, 1659 of whom had a first seizure in the first year of life, whereas 4008 had a first seizure in the second year of life. Of note, ∼2% of children in the VSD population have a seizure in the first 2 years of life.

    Defining Exposure Status: Immunization On Time Versus Delayed

    We used a modification 15 of the method first described by Luman et al 23 to define the days underimmunized for each child in the cohort. For each vaccine received in the first 2 years of life, with the exception of influenza and hepatitis A vaccines, we defined on time versus delayed per the recommended ACIP schedule. 5 Any vaccine recommended at 2 months of life was considered on time if received before 93 days of life. Any vaccine recommended at 12 to 15 months of life was considered on time if received before 489 days of life (16 months of age). Hepatitis A vaccine was excluded because it was not universally recommended until 2007, and influenza vaccine was excluded because of the changing make-up of the vaccine on an annual basis and the seasonality of vaccine administration. We did not include the first dose of hepatitis B vaccine (recommended shortly after birth) in the analysis. We examined only the first dose of each vaccine because others have observed that the first dose of certain vaccines (diphtheria, tetanus, and acellular pertussis vaccine [DTaP] 24 and measles-containing vaccines 16 ) may be the most reactogenic. We did not analyze specific “catch-up” vaccination schedules.

    Defining Outcome Status: Seizure

    As described previously, a seizure was defined by International Classification of Diseases, Ninth Revision, Clinical Modification codes 345.x and 780.3x. In previous VSD work, 16 these codes have been shown to have a positive predictive value (PPV) of 94% for seizures in the ED or hospital settings in children age 12 to 23 months; 92% of these were febrile seizures. Additional work in the VSD has shown that similar codes have a 92% PPV for seizures in the ED setting for infants from 6 weeks to 12 months of age, and 99% for children older than 1 year. 25 Based on these high PPVs, we did not conduct chart review on seizure cases. We limited our analysis to evaluation of first-ever seizure diagnosis for each child.

    Study Design and Analysis

    We used a self-controlled case series (SCCS) design 26 to examine the relationship between vaccination and incidence of seizures. In this case-only method, the incidence rate of events in a postvaccination risk window is compared with the incidence rate of events in an unexposed window composed of time periods before vaccination and after the risk window. Each case serves as its own control, thus implicitly controlling for confounders that do not change over time, such as gender or racial/ethnic background. We conducted the SCCS analysis for the first dose of each vaccine recommended at 2 months and 12 months of age, stratified by timing of vaccination (on time versus delayed). Each vaccine was evaluated separately, without regard to receipt of other concomitant vaccines. The risk window for each vaccine was based on biologic plausibility and evidence from the literature. 27 We used a 0- to 2-day risk window for all vaccines recommended at 2 months of age, except rotavirus, for which we used a 0- to 7-day risk window, as this is a time period of possible risk for intussusception 28 and a time when the live attenuated or reassortant virus vaccines would be expected to replicate; there are no data on seizures after rotavirus vaccination. For measles-mumps-rubella (MMR), varicella, and measles-mumps-rubella-varicella (MMRV) vaccines, we used a risk window of 7 to 10 days 16 after vaccination. The time period from 1 to 14 days before vaccination was excluded to reduce the potential “healthy vaccinee effect.” 29 The control period was defined as the 14-day period directly after the postvaccination risk window, and the 14-day period directly before the healthy vaccinee window; this earlier control period was truncated if any days included age 37 days or younger.

    For each vaccine and exposure group (exposures were receipt of vaccine on time or late), we calculated the incidence rate ratio (IRR) of first-time seizures in a postvaccination window using conditional Poisson regression. 30 The IRR represents, among children with a first diagnosis of seizure, the incidence rate of seizure in an exposed time period (risk window) after vaccination versus the incidence rate of seizure in unexposed time periods (control window). Because we used an SCCS study design, in which cases serve as their own controls, we used conditional Poisson models to analyze the discrete outcome and account for the dependence of observations within a case. All analyses were conducted by using SAS version 9.2 (SAS Institute, Inc, Cary, NC).

    Results

    The initial cohort consisted of 323 247 children. After exclusions and limiting the analysis to vaccinated cases with seizures, the final analytic cohort contained 5667 children (Fig 1). Of these children, 49.7% were vaccinated on time in the first 2 years of life for all vaccines. Assessing vaccination status at the time of first seizure, 71.2% of children with a first seizure at age 38 to 364 days were vaccinated on time; for children with a first seizure at age 365 to 730 days, 62.0% were vaccinated on time.

    For children who received their first infant vaccines at the ACIP-recommended age of 38 to 92 days, there was no association of vaccination with seizures (Table 1). Seizures were less common in this age group in general, but were no more likely to occur in a risk window after vaccination than in the control periods. For example, the IRR for seizure occurring within 2 days of DTaP vaccination compared with control periods was 1.26 (95% confidence interval [CI] of 0.65–2.45). For children who received their first vaccination on a delayed schedule between 93 and 730 days of life, the IRRs for seizures were generally elevated but not significant. For example, for DTaP, the IRR was 1.56 (95% CI 0.19–12.92).

    TABLE 1

    Timing of First Vaccinationa and Occurrence of Seizure, Stratified by Vaccine: Vaccines Recommended at 2 Months of Age

    We next examined vaccines first recommended for administration after 1 year of age (Table 2). When the MMR vaccine was administered according to ACIP recommendations at 12 to 15 months of age (361–488 days), it was associated with an increased risk of seizures in the 7 to 10 days after vaccination: IRR 2.65, 95% CI 1.99–3.55. This association was greater if administration of the vaccine was delayed past 15 months: IRR 6.53, 95% CI 3.15–13.53. When we conducted a subgroup analysis to examine the timing of vaccination in more detail, we found the association of MMR vaccination with seizure at 16 to 18 months had an IRR of 5.09 (95% CI 2.05–12.66) and was most pronounced at 19 to 21 months of age, with an IRR of 8.75 (95% CI 2.35–32.58). Data were too sparse in children ages 22 to 23 months to permit analyses (n = 1 exposed case, no unexposed cases) because of the low number of children vaccinated at this age.

    TABLE 2

    Timing of First Vaccination and Occurrence of Seizure, Stratified by Vaccine: Vaccines Recommended After 12 Months of Age

    Varicella vaccine was associated with an increased risk for seizures 7 to 10 days postvaccination. When administered at 12 to 15 months, the IRR was 2.75 (95% CI 2.05–3.70); the IRR increased to 3.64 when administered at 16 to 23 months of age.

    The association of vaccination with seizure was approximately twice as strong among recipients of MMRV vaccine than recipients of MMR vaccine, both among those who received vaccination on time and among those whose vaccinations were delayed (Table 2). Specifically, for on-time MMRV vaccine receipt, the IRR for seizure in the 7 to 10 days after vaccination was 4.95 (95% CI 3.68–6.66). For delayed receipt of MMRV, the IRR was 9.80 (95% CI 4.35–22.06). The vaccine-seizure association was most pronounced if MMRV vaccine was administered between 16 and 18 months of age (IRR 11.00, 95% CI 4.26–28.38).

    Discussion

    We found no significant association between vaccination in the first year of life and acute seizure events regardless of vaccine type and regardless of whether the vaccine was received on time or delayed. However, in the second year of life, delay of the first MMR vaccine until 16 months of age or older resulted in an IRR for seizures in the 7 to 10 days after vaccination that was 3 times greater than if administration of MMR vaccine occurred on time. Receipt of MMRV compared with MMR doubled the IRR for postvaccination seizures, both at 12 to 15 months and at 16 to 23 months of age, as described recently. 9

    Historically, the whole-cell diphtheria-tetanus-pertussis vaccine was associated with an increased risk of postvaccination febrile seizures in infants. 17 , 18 There is no evidence that the acellular DTaP vaccines in use since the late 1990s are associated with seizures in the United States. 31 Other infant vaccines currently in use, for instance the DTaP, inactivated poliovirus, and Haemophilus influenza type B combined vaccine (DTaP-IPV-HIB), have not been associated with seizures in the United States, 32 although DTaP-IPV-HIB has been linked with increased febrile seizures in Denmark. 24 Other early childhood vaccines that have been associated with febrile seizures in the United States include inactivated influenza vaccine, but only in some influenza seasons, such as 2010–2011, and the 13-valent pneumococcal conjugate vaccine. 33 The risk for seizures after inactivated influenza vaccine and 13-valent pneumococcal conjugate vaccine was greatest if the vaccines were given the same day and in the second year of life. 33 It should be noted that early-childhood vaccines in the first year of life are given at a time of relatively low background rate of febrile seizures. 34 , 35

    The relationship between timing of vaccination and febrile seizures changes in the second year of life, when receipt of MMR and MMRV vaccines between 16 and 23 months is associated with a higher relative incidence of seizures than between 12 and 15 months. Regardless of vaccination, young children are at their greatest risk for febrile seizures at ∼16 to 18 months of age. 34 , 35 In the VSD cohort, the incidence of febrile seizures increases from just >1 per 100 000 person-days at 7 months of age to a maximum of almost 5 per 100 000 person-days at 17 months of age before decreasing to 3 per 100 000 days by 24 months and to 1 per 100 000 days by age 45 months (data not shown). The stronger association of seizures with both MMR and MMRV vaccines administered after 15 months of age, compared with 12 to 15 months, is likely due to a complex interplay between the immunogenicity of the vaccines, the genetic and physiologic susceptibility of the child, and the age-based maturation of the child’s immune system; as the immune system matures in the second year of life 36 it also becomes capable of greater febrile response to immune stimulants, such as vaccines. The relationship between the reactogenicity and the immunogenicity of vaccines was suggested in a recent study that demonstrated a greater risk of measles disease among school-aged children who had received 2 doses of MMR vaccine with the first dose at 12 to 13 months versus at least 15 months of age. 37 Thus, lower reactogenicity of vaccines earlier in the second year of life may also result in lower clinical effectiveness.

    A twofold increased risk of febrile seizures in the 7 to 10 days after MMRV vaccine, compared with MMR and varicella vaccines administered as separate vaccines on the same day, was first reported by Klein et al in 2010. 16 They estimated that use of MMRV, compared with separate MMR and varicella vaccines, will result in 1 additional febrile seizure 7 to 10 days after vaccination for every 2300 MMRV doses administered in the second year of life. The more pyrogenic nature of MMRV compared with separate MMR and varicella vaccines may be because of the higher concentration of attenuated varicella virus in the MMRV formulation (>7 times the tissue culture infectious dose compared with varicella vaccine). 20 Alternatively, it may be because MMRV vaccine induces higher antibody titers to measles than does separate MMR plus varicella vaccines, suggesting higher levels of measles vaccine replication. 21

    Rowhani-Rahbar et al 9 recently examined the impact of age in the second year of life on febrile seizures after vaccination. Using a risk-interval cohort study design (as compared with the SCCS design in this study) they found that the risk of a seizure in the 7 to 10 days after any measles-containing vaccine was doubled (from an incident rate ratio of 3.4 to 6.5) if the child was 16 to 23 months rather than 12 to 15 months at the time of vaccine receipt. The risk was doubled in both age groups if MMRV was used instead of separate MMR plus varicella vaccines. 9 Thus, our results, using a partially overlapping patient population and a different analytic approach, confirm these findings. Based on the findings of Klein et al, 16 Rowhani-Rahbar et al, 9 and our team, we estimate that vaccine type and age of the child both independently but additively increase the risk of seizure 7 to 10 days after receipt of measles-containing vaccine. Thus seizures are approximately twofold more likely to occur after MMRV versus MMR plus varicella vaccine, twofold more likely in 16- to 23-month-old children versus 12 to 15 months, and roughly fourfold more likely in older children who receive MMRV versus younger children who receive MMR plus varicella vaccine. Although our data from the second year of life have significant overlap with those published previously, 9 we felt it important to include these results because of the implications for vaccine delivery in the context of parental delay.

    Our findings are subject to several limitations. First, despite the size of our cohort, there were sparse data on seizures in the first year of life (for example, for DTaP vaccine there were 7 exposed cases). Thus, we were not able to directly examine the seizure risk of deferring first vaccination until late in the first year of life, when the incidence of febrile seizures begins to increase. Second, we did not account for simultaneous administration of different vaccines on the same day, but instead conducted analyses on each vaccine individually. There are too few vaccines given in isolation in early childhood to conduct a meaningful analysis on nonsimultaneously administered vaccines. However, our approach results in, for example, an elevated risk for seizures in the 7 to 10 days after varicella vaccine, but this association is due to the elevated risk caused by MMR vaccine that is administered on the same day. 16 Other studies have not found concomitant administration of vaccines to be a risk for increased adverse events in the second year of life compared with nonconcomitant delivery. 38 , 39 Third, the SCCS design does not permit a direct statistical comparison of the IRRs between different vaccines. However, our finding that the IRR for seizures after MMRV vaccine is doubled compared with MMR vaccine mirrors that recently described in the literature. 9 , 16 , 40

    In summary, in our primary analysis, we found no association between vaccination in the first year of life and subsequent seizures, either among infants vaccinated on time or on a delayed schedule. In the second year of life, receipt of MMR and MMRV vaccines was associated with an increased risk of seizure, with stronger associations observed in children receiving vaccinations on a delayed schedule. It is known that the risk of seizure peaks at 16 to 18 months of life regardless of vaccination status; therefore delaying MMR or MMRV vaccine until this age may result in more febrile seizures. Given the overall low absolute risk of seizures after MMR and MMRV vaccines, 9 , 16 the lack of association of simple febrile seizures with long-term adverse consequences, 34 , 35 and the known benefits of on-time vaccination, our findings provide additional rationale for not delaying childhood vaccinations.

    Acknowledgments

    We acknowledge the contributions of the Marshfield Clinic Research Foundation in Marshfield, Wisconsin, to this project.

    Footnotes

      • Accepted March 17, 2014.
    • Address correspondence to Simon J. Hambidge, MD, PhD, Denver Health Mailcode 1914, 660 Bannock St., Denver, CO 80204; E-mail: simon.hambidge{at}dhha.org
    • Dr Hambidge conceptualized and designed the study, reviewed and interpreted study data, and drafted the initial manuscript; Ms Newcomer contributed to study design, conducted analyses, reviewed and interpreted study data, and critically reviewed the manuscript; Dr Narwaney reviewed and interpreted study data, conducted analyses, and critically reviewed the manuscript; Drs Glanz, Daley, Rowhani-Rahbar, Klein, Lee, Nelson, Lugg, Naleway, Nordin, and DeStefano, and Ms Shoup and Mr Weintraub contributed to study design, reviewed and interpreted study data, and critically reviewed the manuscript; Dr Xu contributed to study design, supervised analyses, and critically reviewed the manuscript; and all authors approved the final manuscript as submitted.

    • Dr Hambidge and Ms Newcomer had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    • The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention or of America’s Health Insurance Plans.

    • Portions of this work were presented during a platform session at the Pediatric Academic Societies annual meeting in Washington, DC, on May 6, 2013.

    • FINANCIAL DISCLOSURE: Dr Hambidge has received royalties, unrelated to this study, from Elsevier for editing a general pediatric textbook; Dr Daley has received an honorarium, unrelated to this study, from McGraw-Hill publishers for writing a textbook chapter on immunizations; Dr Rowhani-Rahbar was a vaccine safety fellow funded by the Centers for Disease Control and Prevention at the time of this study; Dr Klein has received research funding, unrelated to this study, from GlaxoSmithKline, Sanofi-Pasteur, Merck, Novartis, Pfizer, and Protein Science; Dr Naleway has received funding from GlaxoSmithKline; the other authors have indicated they have no financial relationships relevant to this article to disclose.

    • FUNDING: This study was supported through the Vaccine Safety Surveillance and Assessment Projects (contract 200-2002-00732) with American’s Health Insurance Plans, funded by the Centers for Disease Control and Prevention. The Centers for Disease Control and Prevention coauthors (Mr Weintraub and Dr DeStefano) were involved in the design and conduct of the study; analysis and interpretation of the data; and review and approval of the manuscript.

    • POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

    References