ORIGINAL CONTRIBUTION

MMR Vaccination and Febrile Seizures

Evaluation of Susceptible Subgroups and Long-term Prognosis
seiMogens Vestergaard, MD, PhD Anders Hviid, MSci Kreesten Meldgaard Madsen, MD, PhD Jan Wohlfahrt, MSci Poul Thorsen, MD, PhD Diana Schendel, PhD Mads Melbye, MD, DMSci Jrn Olsen, MD, PhD
HE SAFETY OF THE MEASLES, mumps, and rubella (MMR) vaccine is of major public health interest because millions of children are vaccinated every year. Fortunately, the vaccine is generally well-tolerated, rarely associated with serious adverse effects, and may even have nonspecific health benefits.1-5 However, MMR vaccination is followed by a transient increased risk of febrile seizures compared with nonvaccinated children, probably due to vaccine-induced fever.6-10 It may have clinical implications if susceptible children could be identified before the vaccination but no study has been large enough to identify such subgroups. For example, it is unknown whether children with a personal or a family history of seizures are more prone to MMR-induced febrile seizures than children without such history. Febrile

and 4 other national registries. Context The rate of febrile seizures increases following measles, mumps, and rubella (MMR) vaccination but it is unknown whether the rate varies according to personal or family history of seizures, perinatal factors, or socioeconomic status. Furthermore, little is known about the long-term outcome of febrile seizures following vaccination. Objectives To estimate incidence rate ratios (RRs) and risk differences of febrile seizures following MMR vaccination within subgroups of children and to evaluate the clinical outcome of febrile seizures following vaccination. Design, Setting, and Participants A populationbased cohort study of all children born in Denmark between January 1, 1991, and December 31, 1998, who were alive at 3 months; 537171 children were followed up until December 31, 1999, by using data from the Danish Civil Registration System Main Outcome Measures Incidence of first febrile seizure, recurrent febrile seizures, and subsequent epilepsy. Results A total of 439251 children (82%) received MMR vaccination and 17 986 children developed febrile seizures at least once; 973 of these febrile seizures occurred within 2 weeks of MMR vaccination. The RR of febrile seizures increased during the 2 weeks following MMR vaccination (2.75; 95% confidence interval [CI], 2.55-2.97), and thereafter was close to the observed RR for nonvaccinated children. The RR did not vary significantly in the subgroups of children that had been defined by their family history of seizures, perinatal factors, or socioeconomic status. At 15 to 17 months, the risk difference of febrile seizures within 2 weeks following MMR vaccination was 1.56 per 1000 children overall (95% CI, 1.44-1.68), 3.97 per 1000 (95% CI, 2.905.40) for siblings of children with a history of febrile seizures, and 19.47 per 1000 (95% CI, 16.05-23.55) for children with a personal history of febrile seizures. Children with febrile seizures following MMR vaccinations had a slightly increased rate of recurrent febrile seizures (RR, 1.19; 95% CI, 1.01-1.41) but no increased rate of epilepsy (RR, 0.70; 95% CI, 0.33-1.50) compared with children who were nonvaccinated at the time of their first febrile seizure. Conclusions MMR vaccination was associated with a transient increased rate of febrile seizures but the risk difference was small even in high-risk children. The long-term rate of epilepsy was not increased in children who had febrile seizures following vaccination compared with children who had febrile seizures of a different etiology.
JAMA. 2004;292:351-357 www.jama.com (Drs Thorsen and Schendel), Denmark; and National Center on Birth Defects and Developmental Disabilities, Centers for Disease Control and Prevention, Atlanta, Ga (Dr Schendel). Corresponding Author: Mogens Vestergaard, MD, PhD, Department of Epidemiology and Social Medicine, The Danish Epidemiology Science Centre, Aarhus University, Vennelyst Blvd 6, DK-8000 Aarhus C, Denmark (mv@soci.au.dk).

zures are in general associated with Author Affiliations: The Danish Epidemiology Scian ence Centre, Department of Epidemiology and increased risk of epilepsy11-13 but it Social Medicine, Aarhus University, Aarhus (Drs Vestergaard, Madsen, and Olsen), The Danish Epiredemiology Science Centre, Department of Epidemiology Research, Statens Serum Institut, Copenhagen mains unclear if febrile seizures (Dr Melbye and Mr Hviid and Ms Wohlfahrt), and followNorth Atlantic Neuro-Epidemiology Alliances, Departing MMR vaccination carry a particu- ment of Epidemiology and Social Medicine, Aarhus larly high risk. To address these questions, we performed a large population-based cohort study.

2004 American Medical Association. All rights reserved.

(Reprinted) JAMA, July 21, 2004Vol 292, No. 3 351

Downloaded from jama.ama-assn.org by guest on November 18, 2011

MMR VACCINATION AND RATE OF FEBRILE SEIZURES

METHODS
Study Population

This population-based cohort study was based on a previously described study population2 and includes all children born in Denmark between January 1, 1991, and December 31, 1998, who were alive at the age of 3 months (N = 537171). The cohort was established by means of data from the Danish Civil Registration System and 4 other national registries. All live-born children and new residents in Denmark are assigned a unique personal identification number (civil registry number), which is stored in the Danish Civil Registration System together with information on vital status, emigration, address, and family structure (link to mother and father).14 The registry is updated every week and all changes regarding the status of the above-mentioned variables are required by law. The civil registry number can be used to link individual information in all national registries. We obtained permission from the Danish Data Protection Board before the study was initiated.
MMR Vaccination Status

We determined MMR vaccination status from vaccination data reported to the National Board of Health by general practitioners, who provide MMR vaccinations in Denmark. The general practitioners are reimbursed by the state based on these reports. We retrieved information on vaccinations from January 1, 1991, through

December 31, 1999. The MMR vaccine data are transferred to the National Board was introduced in of Health once a week without specifyDenmark in 1987 and ing the day of vaccination, we had to sea single-antigen lect 1 day as the day of vaccination in measles vaccine has our analyses and we chose Wednesnever been day. Since 1996, vaccination informarecommended. The tion has been recorded with the childs MMR own civil registry number and the invaccine used in formation directly linked with other regDenmark during the istries. Before 1996, vaccination inforstudy period was mation and the age of the child were identical to that used recorded with the civil registry numin the United States and ber of the accompanying adult. We used contained the information from the Danish Civil Regfollowing vaccine istration System to identify the link from strains: Moraten the accompanying adult to the child; (measles), Jeryl Lynn therefore, 98.5% of the children were (mumps), and identified with the use of the childs civil Wistar RA 27/3 (rubella). registry number or the civil registry The national number of the mother or father and vaccination program the age of the child at vaccination. recommended durThe remaining 1.5% of vaccinated chiling the entire study dren were identified based on the civil period that chilregistry number of other relatives and dren should be the childs address at the time of vaccivaccinated twice, at 15 nation. months and at 12 years. Only the first Febrile Seizures and Epilepsy vaccination is relevant Information on febrile seizures and epito the end point lepsy was obtained from the National under study. Because Hospital Register (NHR),15 which conthe vaccination tains information on all patients discharged from Danish hospitals since 1977; outpatients (visits to emergency department and hospital clinics) have been included in the register since 1995. All treatments in Danish hospitals are free of charge for all Danish citizens. Diagnostic information was classified according to the Danish version of the International Classification of Diseases (ICD) as follows: ICD-8 was used from 1977 to 199316 and ICD-10 was used from 1994 to the end of 1999.17 We classified children as having a febrile seizure when they were registered with ICD-8 code 780.21 or ICD-10 code R56.0, were aged between 3 and 60 months at the time of discharge, and had no recorded history of nonfebrile seizures, cerebral palsy, severe head traumas, intracranial tumors, meningitis, or encephalitis. The febrile seizures could not be classified as simple or complex be-

cause the NHR contains no information on number of febrile seizures occurring within the febrile episode, duration of the febrile seizures, and type of febrile seizures (generalized or focal onset). Children were categorized with epilepsy if they had ICD-8 code 345 or ICD-10 code G40.
Potential Effect Modifiers and Confounders

We obtained information on febrile seizures and epilepsy in siblings from the NHR during the period January 1, 1977, to December 31, 1999. Children were labeled with a family history of seizures from the day a sibling was admitted to a Danish hospital or had been in outpatient care with febrile seizures or epilepsy. We obtained information on
352

birth weight and gestational age from the Danish Medical Birth Register18 and the NHR.15 Information on socioeconomic status (as indicated by the employment status of the head of the household) and maternal education was obtained from Statistics Denmark at the time the child was aged 15 months.
Statistical Analysis

To study the association of MMR vaccination with a first episode of febrile seizure, we followed the children from

the age of 3 months until the first diagnosis of febrile seizure registered in the NHR, death, emigration, a diagnosis of epilepsy, cerebral palsy, severe brain injury, brain tumor, meningitis, encephalitis, aged 5 years, or until December 31, 1999, whichever came first. The resulting person-years at risk were aggregated and analyzed using Poisson regression, producing incidence rate ratios (RRs).19 We considered MMR vaccination a time-varying covariate; the children were assigned to the nonvaccinated group until they received the MMR vaccine. From that day, they were included in the vaccinated cohort. We evaluated whether the RR of febrile seizures following MMR vaccination varied between subgroups of children by testing for statistical interaction. All RRs were adjusted for age (3-month categories) and calendar

JAMA, July 21, 2004Vol 292, No. 3 (Reprinted)

2004 American Medical Association. All rights reserved.

Downloaded from jama.ama-assn.org by guest on November 18, 2011

MMR VACCINATION AND RATE OF FEBRILE SEIZURES

period (1-year categories). In multi- sonal history of febrile seizures, we come of interest (a second episode of variable analyses, we considered constructed a cohort of 10 541 chil- febrile seizures or a first diagnosis of condren who were nonvaccinated at the epilepsy), death, emigration, cerebral founding by sex, number of siblings time of the first febrile seizure. palsy, severe brain injury, brain tumor, with febrile seizures (no siblings, no These meningitis, encephalitis, aged 5 years, or sibchildren were followed up prospec- until December 31, 1999, whichever lings with febrile seizures, 1 sibling tively from the day of the first regis- came first. The RRs were adjusted for age, with tered febrile seizure until the second calendar period, age at first febrile seifebrile seizures, 2 siblings with episode of febrile seizure registered zure, and current vaccination status. febrile in seizures), number of siblings with the NHR, death, emigration, a epidiagnolepsy (no siblings, no siblings with epilepsy, 1 siblings with epilepsy), birth order (1, 2, 3, 4), gestational age in sis of epilepsy, cerebral palsy, severe All analyses were conducted using weeks ( 36, 37-41, 42), birth weight brain injury, brain tumor, meningitis, SAS statistical software version 8.2 (SAS in grams ( 2499, 2500-2999, 3000- encephalitis, aged 5 years, or until Institute Inc, Cary, NC). P .05 was De3499, 3500-3999, 4000), maternal cember 31, 1999, whichever came first. considered statistically significant.

education (postgraduate education, We considered MMR vaccination a RESULTS coltime-varying covariate. The RRs We followed up 537171 children for a lege, vocational training, secondary were school, primary school), and adjusted for age, age at first febrile total of 1.9 million person-years and socioecoseiidentified 17986 children who develnomic status as indicated by the zure, and calendar period. oped febrile seizures at least once; 973 employTo estimate the number of addi- of these febrile seizures occurred within ment status of the head of the tional febrile seizures that occurred 2 weeks of the MMR vaccination. Durhousein ing the study period, 439251 children hold (managers [very high], wage the 2 weeks following MMR vaccina- (82%) received MMR vaccination. earner tion compared with nonvaccinated RRs of Febrile Seizures [high], wage earner [medium], wage chilearner [low], wage earner [minimal], dren (risk difference), we first calcu- After MMR Vaccination Overall, we found that the rate of first unemployed). We had no lated the proportion of vaccinated febrile seizures was 10% higher among information and on birth weight, gestational age at nonvaccinated children that vaccinated children (7445; personbirth, developed socioeconomic status, and maternal febrile seizures at a given age when education for 6.2%, 31.7%, 2.7%, foland lowed up for 2 weeks (cumulative 0.3% of the children, respectively. incidence). The cumulative Data incidences on gestational age at birth was not were calculated by using the availexponenable for children born after tial formula: cumulative incidence=1 December 31, 1996. When evaluating confounding, we used 2 different strategies for the handling of missing values. First, we used the method of single imputation, replacing a missing value with the exp (incidence rate time).21 The for- years at risk, 1151661) compared with

most common value of that variable: 3000 to 3499 g for birth weight, 37 to 41 weeks for gestational

age at birth, wage earner (standard level) for socioeconomic status, and vocational train-

ing for maternal education. Second, we analyzed only those children with complete information on all variables (358702). A priori, we decided to add all variables to the final model that changed the estimate of interest by at least 10% using either strategy.20 None of the variables with missing data qualified. Only age and calendar period were included in the final model. When evaluating effect modification, we analyzed only those children with complete information on the variable of interest. To study the association of MMR vaccination and a second episode of febrile seizures in children with a per-

mula is based on the assumption that nonvaccinated children (10 541; perthe son-years at risk, 793568) during the incidence rate is constant during the study period (RR, 1.10; 95% CI, pe1.05-1.15), after adjusting for age and riod of interest. The age-specific calendar period. However, the rate of cumufebrile seizures increased during the first lative incidence was calculated (RR, 2.46; 95% CI, 2.22-2.73) and secsepaond week (RR, 3.17; 95% CI, 2.89rately for vaccinated and 3.49) following vaccination only nonvaccinated (FIGURE 1); thereafter the rate was close children within each subgroup and to that for nonvaccinated children. the Overall, the RR of febrile seizures risk difference calculated as the within 2 weeks of MMR vaccination was differ2.75 (95% CI, 2.55-2.97) compared ence between the cumulative inci- with nonvaccinated children. We found dences. Confidence intervals (CIs) no statistically significant difference in for the RR of febrile seizures in the 2 weeks the risk difference were calculated following vaccination between subusgroups of children characterized by ing the Delta method.22 family history of seizures, sex, birth orTo evaluate the long-term der, gestational age at birth, birth prognosis weight, or socioeconomic factors, comof febrile seizures following MMR pared with nonvaccinated children vacwithin the subgroup under study cination compared with febrile (FIGURE 2). The highest RR was found seizures of a different etiology, we categorized children with febrile seizures into 3 groups according to the vaccination status at the time of the first febrile seizure: 10 541 children were nonvaccinated, 973 children had been vaccinated within the previous 2 weeks, and 6472 children were vaccinated more than 2 weeks ago. These children were followed up from the day of the first registered febrile seizure until the out(Reprinted) JAMA, July 21, 2004Vol 292, No. 3 353

2004 American Medical Association. All rights reserved.

Downloaded from jama.ama-assn.org by guest on November 18, 2011

MMR VACCINATION AND RATE OF FEBRILE SEIZURES

Figure 1. Adjusted Rate Ratios of Febrile Seizures According to Time Since MMR Vaccination vs Nonvaccinated Children in a Cohort of Children Born in Denmark, 1991-1998
4.0

3.0

2.0

1.0

0.5

9-26

27-52

53-104 105-156 157-260

rate was close to that observed for nonvaccinated children. This finding is consistent with previous reports6-10 and is expected since MMR vaccination often induces fever,1 a necessary cause of febrile seizures. Farrington et al7 found an increased rate of febrile seizures up to 35 days after vaccination with the Urabe mumps strain but the rate was increased no longer than 2 weeks for the Jeryl Lynn vaccine, which was used in our study. Family history of seizures, preterm birth, low birth weight, and male sex are risk factors for febrile seizures23 but the RR of febrile seizures following MMR vaccination did not vary significantly according to these factors in this study. The highest RR was found among siblings of children with epilepsy; a

4-fold increased rate of febrile seizures following MMR vaccination was MMR indicates measles, mumps, and rubella. Rate ratios are adjusted for age and calendar period. Point esobserved compared with nonvaccitimates are given with error bars representing 95% confidence intervals. nated siblings of children with epilepsy. However, our statistical power in among siblings of children with a to 20 months, and 0.64 per 1000 (95% seizures hisin the 2 weeks following CI, 0.22-1.40) for children vaccinated tory of epilepsy who had a 4-fold inat 21 to 23 months. MMR vaccicreased rate of febrile seizures in The highest risk difference was found nation compared with the 2 among children with a personal hisnonvaccinated weeks following vaccination comtory of febrile seizures (19.47 per 1000; children was 1.56 per pared with a 2.7-fold increased rate 95% CI, 16.05-23.55) and for chil1000 (95% CI, of 1.44-1.68) for children dren with a family history of febrile seifebrile seizures following zures (3.97 per 1000; 95% CI, 2.90vaccinated at 15 vaccination in to 17 months, 1.46 per 5.40; TABLE 1). siblings of children with no history of 1000 (95% CI, epilepsy (P value for 1.10-1.91) for children Long-term Prognosis of Febrile Seizures interaction=.09). vaccinated at 18 Following MMR Vaccination Among the 10 541 children with a We found that children who experipersonal history of febrile seizures, 175 enced febrile seizures within 2 weeks children had a recurrent febrile seiof MMR vaccination had a 19% inzure within 2 weeks of the MMR vaccreased rate of recurrent febrile seicination. The RR of febrile seizures zures (RR, 1.19; 95% CI, 1.01-1.41) but in no increased rate of epilepsy (RR, 0.70; the 2 weeks following vaccination 95% CI, 0.33-1.50) during up to 105 was months of follow-up. The reference 2.75 (95% CI, 2.32-3.26) after group consisted of children who had adjustnot been vaccinated when having their ing for age, age at the first febrile first febrile seizure (TABLE 2). seiCOMMENT zure, and calendar period, The incidence rate of febrile seizures compared with nonvaccinated children with a was increased in the 2 weeks followpering MMR vaccination and thereafter the sonal history of febrile seizures.
Time Since Vaccination, wk

Risk Difference of Febrile Seizures Among Subgroups of Vaccinated Children

The risk difference of febrile

this subgroup was limited and further studies are needed to determine whether the siblings of children with epilepsy are more likely to experience a febrile seizure after MMR vaccination than other children, or the finding is merely due to chance. The RR of febrile seizure was not modified by a family history of febrile seizures. Overall, our data suggest that MMR vaccination and the other indicators for febrile seizures follow a multiplicative
354

model; the rate of febrile seizures in all subgroups of children is approximately 2.75 times higher within 2 weeks of MMR vaccination than it would have been had the children not been vaccinated. The absolute increase in incidence of febrile seizures following vaccination depends therefore on the

underlying risk of febrile seizures in each subgroup. In Denmark, most children are vaccinated against MMR at age 15 to 17 months when the incidence rate of febrile seizures is peaking.24 At this age, the number of children experiencing febrile seizures within 2 weeks was 1.56 more per 1000 vaccinated children compared with nonvaccinated children. No previous studies have cal-

JAMA, July 21, 2004Vol 292, No. 3 (Reprinted)

2004 American Medical Association. All rights reserved.

Rate Ratio

Downloaded from jama.ama-assn.org by guest on November 18, 2011

MMR VACCINATION AND RATE OF FEBRILE SEIZURES

culated the risk difference according brile seizures was collected prospec- The information on vaccination was to tively and independent of parental reported to the National Board of Health age at vaccination, but 2 studies recall. on a weekly basis but without inforfound We expect the data quality of the mation on the exact day of vaccinathat approximately 0.33 febrile sei- MMR tion. We chose Wednesday as the day zures were attributable to 1000 vaccination to be high because the of vaccination. Because children in Dendoses genmark are vaccinated Monday thru Friof MMR vaccine overall.6,7 eral practitioners are reimbursed only As expected, we found the highest afrisk ter reporting immunization data to difference in children with a the personal National Board of Health. history of febrile seizures. The underlying risk of febrile seizures in these children is high; approximately one third will have at least 1 episode of recurrent

Figure 2. Adjusted Rate Ratios of Febrile Seizures Within 2 Weeks Following MMR Vaccination for Children With Specific Characteristics vs Nonvaccinated Children With the Same Characteristics

febrile seizures before they reach 5 Strata years of age.25 In this very high-risk group, All Children Siblings With Febrile Seizures we found 19 additional febrile seizures within 2 weeks of the vaccination per 1000 children compared with nonvaccinated children aged 15 to 17 months. The Advisory Committee on Immunization Practices has suggested that the benefits of administering MMR vaccine to children with a personal or family history of convulsions substantially outweigh the risks, and these children should be vaccinated following the recommendations for children who have no contraindications.26,27 We found no increased rate of epilepsy among children who had febrile seizures after MMR vaccination compared with children who had febrile seizures of a different etiology. The rate of recurrent febrile seizures was slightly increased, possibly because the MMR

Rate Ratio P Value for (95% Confidence Interval)Interaction 2.75 (2.55-2.97)

vaccination introduced an extra febrile episode during the window of highest susceptibility and the total number of febrile episodes is a well known risk factor for recurrence.28 We know of only 1 study6 evaluating the clinical outcome of children with febrile seizures following MMR vaccination. No increased rate of subsequent seizures was found in 41 children with febrile seizures following MMR vaccination compared with 521 children who had

No Siblings Pensioner, Student No Siblings With Febrile Mothers Educational Level Seizures Primary Education 1 Sibling With Febrile Secondary Education Seizures Vocational Training 2 Siblings With Febrile College Seizures Postgraduate Education Siblings With Epilepsy No Siblings No Siblings With Epilepsy 1 Siblings With Epilepsy Sex Boy Girl Birth Order 1 2 3 4 Gestational Age, wk <37 37-41 42 Birth Weight, g <2500 2500-2999 3000-3499 3500-3999 4000 Socioeconomic Status Manager (Very High) Wage Earner (High Level) Wage Earner (Medium Level) Wage Earner (Standard Level) Wage Earner (Other) Unemployed,

2.68 (2.41-3.00) 2.78 (2.51-3.07) 2.84 (2.23-3.60) 3.08 (1.49-6.34)

2.68 (2.40-2.99) 2.73 (2.48-3.01) 3.95 (2.63-5.94)

2.93 (2.21-3.89) 2.66 (2.20-3.21) 2.88 (2.56-3.25) 2.73 (2.40-3.10) 2.82 (2.35-3.38)

.68 .90

.09 2.84 (2.26-3.57) 3.02 (2.60-3.51) 2.64 (2.23-3.13) 2.90 (2.54-3.30) 2.35 (1.93-2.86) 2.61 (2.20-3.10)

2.74 (2.48-3.02) 2.78 (2.50-3.09)

.82

.34

2.69 (2.42-2.98) 2.65 (2.36-2.99) 3.21 (2.67-3.85) 2.47 (1.75-3.48)

3.11 (2.33-4.16) 2.93 (2.67-3.22) 2.97 (2.26-3.90)

2.75 (2.41-3.14) 2.69 (2.17-3.34) 2.73 (2.45-3.06) 2.93 (2.46-3.48) 2.59 (1.93-3.48)

.54 .95

.84

febrile seizures in the absence of vaccination.6 However, the statistical power of this study was limited, in particular

3 4 56

Rate Ratio (95% Confidence Interval)

when evaluating the rate of subse- MMR indicates measles, mumps, and rubella. Vertical dashed line represents the overall rate ratio (RR) for febrile seizures within the 2 weeks following MMR vaccination compared with nonvaccinated children. Point quent epilepsy. estimates are given with error bars representing 95% confidence intervals. The strengths of our study include *The RRs are adjusted for age and calendar period. The analyses including siblings were additionally adjusted for the total number of siblings. Children with missing values were excluded when the effect of the variable its concerned was evaluated. size and population-based nature. Test for interaction was performed by a test for trend. When evaluating the possible effect modification by siblings with febrile seizures or by siblings with epilepsy, children without siblings were not included in the test The for interaction. follow-up was virtually complete, which eliminates bias due to nonresponse. Information on MMR vaccinations and fe2004 American Medical Association. All rights reserved.
(Reprinted) JAMA, July 21, 2004Vol 292, No. 3 355

Downloaded from jama.ama-assn.org by guest on November 18, 2011

MMR VACCINATION AND RATE OF FEBRILE SEIZURES

Table 1. Cumulative Incidence and Risk Difference of Febrile Seizures Within 14 Days for Vaccinated and Nonvaccinated Children at 15 to 17 Months No. of Febrile Seizures Within 14 Days per 1000 Children Cumulative Incidence Characteristic All children Children with personal history of febrile seizures Siblings of children with history of febrile seizures Siblings of children with history of epilepsy Nonvaccinated 0.90 11.50 2.62 1.59 Vaccinated 2.46 30.97 6.60 4.97 Risk Difference (95% CI) 1.56 (1.44-1.68) 19.47 (16.05-23.55) 3.97 (2.90-5.40) 3.38 (1.90-5.92)

Children with birth weight

2500 g 37 weeks

1.24 1.36

2.93 3.19

1.69 (1.18-2.41) 1.83 (1.27-2.62)

Children with gestational age at birth

Abbreviation: CI, confidence interval.

Table 2. Adjusted Rate Ratios of Recurrent Febrile Seizures and Subsequent Epilepsy* Recurrent Febrile Seizures MMR Vaccination Status at Time of First Febrile Seizure None Within 14 days 14 days prior No. of Children 10 541 973 6472 No. of Recurrent Febrile Seizures 2753 236 918 Person-Years at Risk 23 560 2212 12 675 Rate Ratio (95% Confidence Interval) 1.00 1.19 (1.01-1.41) 1.10 (0.96-1.26) No. of Epilepsy 251 9 95 Person-Years at Risk 41 310 3825 21 938 Rate Ratio (95% Confidence Interval) 1.00 0.70 (0.33-1.50) 0.92 (0.59-1.43) Subsequent Epilepsy

Abbreviation: MMR, measles, mumps, and rubella. *Among 17 986 Danish children who had febrile seizures between 1991 and 1999. Children were categorized according to vaccination status at time of first febrile seizure. Rate ratios are adjusted for age, calendar period, age at first febrile seizure, and current vaccination status.

day, we have misclassified some vaccinations by up to 2 days. Previous studies have shown that the attenuated viruses in the MMR vaccine cause fever in approximately 10% of nonimmune vaccinees between 5 and 12 days after immunization.1,29 Thus, the rate of febrile seizures is probably not elevated during the first 4 days following vaccination. We have previously validated the quality of febrile seizure registration in the NHR in a cohort of 6624 children born between 1991 and 1992 and followed up until age 10 years.30 We collected information about febrile seizures in the cohort using a parental questionnaire. All potential febrile seizures were confirmed by diagnostic

telephone interview or review of medical records. We found that 323 children (4.9%) in the cohort fulfilled the criteria for febrile seizures and 231 of those were registered in the NHR (completeness, 71.5%; 95% CI, 66.3%-76.4%). Among the 249 children registered with febrile seizures in the NHR, we confirmed the diagnosis in 231 children

(predictive value of a positive registration, 92.8%; 95% CI, 88.8%-95.7%). We believe it is unlikely that MMR vaccination status influences the threshold for hospitalization or the coding of febrile seizures. Any misclassification of febrile seizures is likely to be nondifferential and will therefore bias the RR toward 1.0.21 In fact, we found that the RR of febrile seizures following MMR vaccination was virtually the same during the period 1991 to 1994 (2.68; 95% CI, 2.38-3.02) compared with the period 1995 to 1998 (2.79; 95% CI, 2.55-3.05), although outpatients were included in the latter period only. The Danish national vaccination program recommends that children be vaccinated with MMR at age 15 months and provides vaccinations free of charge. Overall, vaccination coverage was found to be 82%, which increased during the study period. The effect of vaccination may be confounded by variables related both to avoidance of vaccination and to the outcome of interest. We adjusted our results for several potential confounders but found little

change in the estimate of interest. However, the strongest argument against serious confounding is that the risk of febrile seizures was almost the same for nonvaccinated and vaccinated children outside the time frame of 2 weeks following vaccination. MMR vaccination is an effective health intervention. The 3 diseases and
356

their neurological sequelae are at high risk of febrile seizures, and the rarely long-term rate of epilepsy was not inobserved today in countries with creased in children who had febrile seihigh zures following MMR vaccination comvaccination coverage.31,32 Our study pared with children who had febrile showed that the transient increased seizures of a different etiology. rate of febrile seizures was restricted to 2 Author Contributions: Dr Melbye had access to all of weeks following vaccination, the risk the data in the study and takes full responsibility for the integrity of the data and the accuracy of the data difference was small even in analysis. children Study concept and design: Vestergaard, Hviid, Madsen,
Wohlfahrt, Melbye, Olsen.

JAMA, July 21, 2004Vol 292, No. 3 (Reprinted)

2004 American Medical Association. All rights reserved.

Downloaded from jama.ama-assn.org by guest on November 18, 2011

MMR VACCINATION AND RATE OF FEBRILE SEIZURES

Acquisition of data: Vestergaard, Madsen, Thorsen, 5. Aaby P, Samb B, Simondon F, et al. Non-specific Melbye. beneficial effect of measles immunisation: analysis of Analysis and interpretation of data: Vestergaard, Hviid, mortality studies from developing countries. BMJ. Madsen, Wohlfahrt, Schendel, Melbye, Olsen. 1995;311:481-485. Drafting of the manuscript: Vestergaard. 6. Barlow WE, Davis RL, Glasser JW, et al. The risk Critical revision of the manuscript for important inof seizures after receipt of whole-cell pertussis or tellectual content: Vestergaard, Hviid, Madsen, measles, mumps, and rubella vaccine. N Engl J Med. Wohlfahrt, Thorsen, Schendel, Melbye, Olsen. 2001;345:656-661. Statistical Analysis: Hviid, Wohlfahrt. 7. Farrington P, Pugh S, Colville A, et al. A new method Obtained funding: Vestergaard, Madsen, Thorsen, for active surveillance of adverse events from diphSchendel, Melbye. theria/tetanus/pertussis and measles/mumps/ Administrative, technical, or material support: rubella vaccines. Lancet. 1995;345:567-569. Vestergaard, Thorsen, Schendel. 8. Griffin MR, Ray WA, Mortimer EA, Fenichel GM, Supervision: Vestergaard, Melbye, Olsen. Schaffner W. Risk of seizures after measles-mumpsFunding/Support: This study was supported by grant rubella immunization. Pediatrics. 1991;88:881-885. 22-02-0207 from the Danish Medical Research Coun- 9. Miller C, Miller E, Rowe K, et al. Surveillance of cil. The Danish National Research Foundation funds symptoms following MMR vaccine in children. Practhe activities of the Danish Epidemiology Science titioner. 1989;233:69-73. Centre. 10. Hirtz DG, Nelson KB, Ellenberg JH. Seizures folRole of the Sponsors: The Danish Medical Research lowing childhood immunizations. J Pediatr. 1983;102: Council and the Danish National Research Founda14-18. tion did not participate in the design and conduct of 11. Annegers JF, Hauser WA, Shirts SB, Kurland LT. the study, in the collection, analysis, and interpretaFactors prognostic of unprovoked seizures after tion of the data, or in the preparation, review, or apfebrile convulsions. N Engl J Med. 1987;316:493proval of the manuscript. 498. Acknowledgment: We thank Lars Pedersen, MSci, 12. Nelson KB, Ellenberg JH. Predictors of epilepsy in and children who have experienced febrile seizures. N Engl Anders Riis, MSci, Department of Clinical EpidemiolJ Med. 1976;295:1029-1033. ogy, Aarhus University Hospital, Denmark, for help13. Ellenberg JH, Nelson KB. Sample selection and the ing us to create Figure 2. natural history of disease: studies of febrile seizures. JAMA. 1980;243:1337-1340. 14. Malig C. The civil registration system in DenREFERENCES mark. Technical Papers IIVRS. 1996;66:1-6. 1. Peltola H, Heinonen OP. Frequency of true ad15. Andersen TF, Madsen M, Jorgensen J, et al. The verse reactions to measles-mumps-rubella vaccine: a Danish National Hospital Register: a valuable source double-blind placebo-controlled trial in twins. Lanof data for modern health sciences. Dan Med Bull. 1999;46:263-268. cet. 1986;1:939-942. 2. Madsen KM, Hviid A, Vestergaard M, et al. A popu- 16. World Health Organization. Manual of the International Statistical Classification of Diseases, Inlation-based study of measles, mumps, and rubella juries, and Causes of Death. Geneva, Switzerland: vacWorld Health Organization; 1967. cination and autism. N Engl J Med. 2002;347:14771482. 17. World Health Organization. The International Statistical Classification of Diseases and Related Health 3. Patja A, Davidkin I, Kurki T, et al. Serious adverse events after measles-mumps-rubella vaccination dur- Problems, 10th Revision. Geneva, Switzerland: World ing a fourteen-year prospective follow-up. Pediatr In- Health Organization; 1993. fect Dis J. 2000;19:1127-1134. 18. Knudsen LB, Olsen J. The Danish Medical Birth Registry. Dan Med Bull. 1998;45:320-323. 4. Makela A, Nuorti JP, Peltola H. Neurologic disorders after measles-mumps-rubella vaccination. Pediatrics. 2002;110:957-963. 19. Clayton D, Hills M. Statistical Models in Epidemiology. New York, NY: Oxford University Press; 1993. 20. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health. 1989;79: 340-349. 21. Rothman KJ, Greenland S. Modern Epidemiology. 2nd ed. Baltimore, Md: Williams & Wilkins; 1998. 22. Agresti A. Categorical Data Analysis. New York, NY: John Wiley & Sons; 1990. 23. Vestergaard M, Basso O, Henriksen TB, et al. Risk factors for febrile convulsions. Epidemiology. 2002; 13:282-287. 24. Stafstrom CE. The incidence and prevalence of febrile seizures. In: Baram TZ, Shinnar S, eds. Febrile Seizures. London, England: Academic Press; 2002. 25. Offringa M, Bossuyt PM, Lubsen J, et al. Risk factors for seizure recurrence in children with febrile seizures: a pooled analysis of individual patient data from five studies. J Pediatr. 1994;124:574-584. 26. Leads from the MMWR. Pertussis immunization; family history of convulsions and use of antipyretics: supplementary ACIP statement. JAMA. 1987; 257:2894. 27. Watson JC, Hadler SC, Dykewicz CA, et al. Measles, mumps, and rubellavaccine use and strategies for elimination of measles, rubella, and congenital rubella syndrome and control of mumps: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep. 1998;47(RR-8):1-57. 28. Knudsen FU. Frequent febrile episodes and recurrent febrile convulsions. Acta Neurol Scand. 1988; 78:414-417. 29. Update: vaccine side effects, adverse reaction, contraindications, and precautions: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep. 1996;45(RR-12):135. 30. Vestergaard M. Genetic and Prenatal Epidemiology of Febrile Convulsions [PhD thesis]. Aarhus, Denmark: University of Aarhus; May 31, 2002. 31. Peltola H, Davidkin I, Paunio M, et al. Mumps and rubella eliminated from Finland. JAMA. 2000;284: 2643-2647. 32. de Quadros CA, Olive JM, Hersh BS, et al. Measles elimination in the Americas: evolving strategies. JAMA. 1996;275:224-229.

2004 American Medical Association. All rights reserved.

(Reprinted) JAMA, July 21, 2004Vol 292, No. 3 357

Downloaded from jama.ama-assn.org by guest on November 18, 2011