Five-year cumulative incidence, 1.1% (95% CI, 0.8%-1.4%) for exposed group and 0.5% (95% CI, 0.4%-0.6%) for unexposed group. Ten-year cumulative incidence, 1.7% (95% CI, 1.4%-2.1%) for exposed group and 0.9% (95% CI, 0.8%-1.0%) for unexposed group.
aAdjusted for sex, year of birth, maternal history of epilepsy, malformations, and gestational age.
eAppendix. Hospital Diagnoses (ICD-8 and ICD-10 Codes) for Conditions Examined in This Study
Customize your JAMA Network experience by selecting one or more topics from the list below.
Olsen M, Thygesen SK, Østergaard JR, et al. Hospital-Diagnosed Pertussis Infection in Children and Long-term Risk of Epilepsy. JAMA. 2015;314(17):1844–1849. doi:10.1001/jama.2015.13971
Pertussis is associated with encephalopathy and seizures in infants. However, the risk of childhood epilepsy following pertussis is unknown.
To examine whether pertussis is associated with the long-term risk of epilepsy.
Design, Setting, and Participants
We used individually linked data from population-based medical registries covering all Danish hospitals to identify a cohort of all patients with pertussis born between 1978 and 2011, followed up through 2011. We used the Civil Registration System to identify 10 individuals from the general population for each patient with pertussis, matched on sex and year of birth.
Inpatient or hospital-based outpatient diagnosis of pertussis.
Main Outcomes and Measures
Cumulative incidence and hazard ratio of time to hospital-based epilepsy diagnosis (pertussis cohort vs general population cohort), adjusted for birth year, sex, maternal history of epilepsy, presence of congenital malformations, and gestational age. Unique personal identifiers permitted unambiguous data linkage and complete follow-up for death, emigration, and hospital contacts.
We identified 4700 patients with pertussis (48% male), of whom 90 developed epilepsy during the follow-up. The cumulative incidence of epilepsy at age 10 years was 1.7% (95% CI, 1.4%-2.1%) for patients with pertussis and 0.9% (95% CI, 0.8%-1.0%) for the matched comparison cohort. The corresponding adjusted overall hazard ratio was 1.7 (95% CI, 1.3-2.1).
Conclusions and Relevance
In Denmark, risk of epilepsy was increased in children with hospital-diagnosed pertussis infections compared with the general population; however, the absolute risk was low.
Pertussis, an acute respiratory tract infection caused by the bacterium Bordetella pertussis, is among the most common vaccine-preventable childhood diseases in developed countries. It is characterized by spasms of coughing and a protracted course.1 Although high immunization rates have lowered the number of pertussis cases substantially, widespread outbreaks still occur. Worldwide, an estimated 16 million cases occur each year.2 Almost 50 000 pertussis cases were reported in the United States in 2012,2,3 and an apparent resurgence of pertussis also has been recently reported in other countries.3,4 The cause of this resurgence is unknown but may involve incomplete vaccination rates, use of acellular vaccines, and genetic changes in B pertussis.5 Infants younger than 2 months have the highest rates of pertussis-associated hospitalizations and deaths.6 Although there is no evidence of neurotoxic effects of B pertussis infections, increased intrathoracic and intra-abdominal pressure during coughing can result in central nervous system hemorrhages.1 Together with hypoxemia, this may lead to encephalopathy,7 occurring in 0.3% of infants with pertussis.8
Epilepsy is the most common neurologic childhood disorder,9,10 and its etiology is poorly understood.9,10 In fact, most cases of epilepsy are diagnosed without a known cause.9,11,12
Known triggers of epilepsies include external insults to the central nervous system, including brain injury and central nervous system infections.10 Pertussis, during the acute phase, is associated with seizures8 in infants, but the likelihood of developing epilepsy is unknown. We therefore conducted a population-based cohort study of the association between pertussis and risk of childhood epilepsy.
Our nationwide study was conducted in Denmark (population, 5.6 million). The Danish National Health Service provides tax-supported health care, with free access to hospital-based and primary medical care.
Since 1968, a unique 10-digit civil personal registration number has been assigned to all residents of Denmark by the Central Office of Civil Registration. The numbers are used in all Danish registries, permitting unambiguous individual-level linkage of data from all sources used in this study. The Civil Registration System also made it possible to identify the mothers of children in the pertussis and comparison cohorts.13
We used the Danish National Patient Registry (DNPR) to identify all individuals with inpatient stays or hospital outpatient or emergency clinic visits who were born and received a diagnosis of pertussis from January 1, 1978, until December 31, 2011. The DNPR has tracked hospitalizations since 1977 and outpatient and emergency department visits at all hospitals in Denmark since 1995. Data recorded in this registry include CPR numbers, dates of admission and discharge, and up to 20 diagnoses, classified according to International Classification of Diseases, Eighth Revision (ICD-8) codes until 1993 and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes thereafter. Routine pertussis vaccination was introduced in 1961 as a diphtheria-tetanus-pertussis whole-cell vaccine administered at ages 5, 6, 7, and 15 months. In August 1969, the whole-cell vaccine began to be administered as a separate injection in reduced dosages at ages 5 and 9 weeks and 10 months. In January 1997, a pertussis toxoid vaccine replaced the whole-cell vaccine in a diphtheria-tetanus-pertussis toxoid-inactivated polio combination administered at ages 3, 5, and 12 months.14
We used the Civil Registration System to identify 10 individuals from the general population for each patient in the pertussis cohort. Members of the comparison cohort were matched to patients in the pertussis cohort on sex and year of birth and had to be alive at the date of the pertussis diagnosis (referred to as the index date for comparison cohort members).
The study outcome was defined as a first-time diagnosis of epilepsy recorded in the DNPR during the follow-up period. Most children with epilepsy in Denmark are expected to have an inpatient or outpatient hospital contact.15 We did not use prescription records to define childhood epilepsy, as physicians refrain from or postpone antiepileptic drug treatment in about 15% of new-onset cases.16
For members of the pertussis and comparison cohorts, we obtained the history of hospital contacts for meningitis or brain tumors, as well as maternal history of epilepsy. The Danish Medical Birth Registry provided information on birth order, self-reported maternal smoking during pregnancy (available starting in 1991), cesarean delivery, gestational age, birth weight, and 5-minute Apgar score. Because congenital malformations are associated with increased risk of epilepsy and also may be associated with respiratory tract infections, we ascertained from the DNPR all diagnoses of congenital malformations detected during the first year of life.
The diagnostic codes used to identify conditions in this study are listed in the eAppendix in the Supplement.
We followed up each child for a maximum of 15 years from his or her pertussis diagnosis date/index date until the date of the first epilepsy diagnosis, emigration, death, or December 31, 2011, whichever came first. We computed the 10-year cumulative incidence of epilepsy and constructed cumulative incidence curves and used Cox proportional hazards regression to estimate relative risks for epilepsy among patients in the pertussis cohort compared with members of the general population cohort. The analyses were adjusted for sex, calendar year of birth, maternal history of epilepsy (yes/no), gestational age at birth (≤36, >36-≤38, >38-≤41, and >41 weeks), and congenital malformations (yes/no). We repeated the overall analyses in subgroups according to age at pertussis diagnosis (0 to ≤1, >1 to ≤5, >5 to ≤35, and >35 months), short vs long admissions for pertussis, gestational age categories, and birth before vs after the introduction of the pertussis toxoid vaccine in 1997. We used length of pertussis hospital admissions as an indicator of pertussis severity and defined long hospital admissions for pertussis as length of stay longer than the 75th percentile separately for the ICD-8 and ICD-10 periods. This allowed us to address changes in admission length over time as well as changes in registration practices in the 2 periods. The pertussis toxoid vaccine, which became available in Denmark in 1997, is thought to induce better immunity than the previously used whole-cell vaccine,17 potentially ameliorating the clinical course of pertussis.
Variation of estimates across subgroups was evaluated using tests of heterogeneity.18 The statistical significance threshold of the 2-sided testing was set at P < .05. In a sensitivity analysis restricted to participants born after 1990, we included maternal smoking during pregnancy (yes/no). Time from pertussis diagnosis was the underlying time scale in all analyses, and the assumption of proportional hazards was graphically verified.
We used SAS version 9.2 (SAS Institute Inc) for all analyses. The study was approved by the Danish Data Protection Agency, which protects the privacy of individuals whose data are recorded in Danish registries. No informed consent was required for this study.
We identified 4700 patients with pertussis (48% male), of whom 53% were diagnosed before age 6 months. Numbers of patients varied according to calendar period of birth, with most patients (n = 1059 [23%]) born during the years 1993-1997 and fewest during 2003-2007 (n = 430 [9%]) and 2008-2011 (n = 203 [4%]). Prevalence of preterm birth was higher in the pertussis cohort (11%) than in the general population cohort (5%). Apgar scores of 7 or below, congenital malformations, history of meningitis, history of brain tumor, maternal history of epilepsy, and maternal smoking during pregnancy were distributed equally between the pertussis and comparison cohorts (Table 1). The median and 75th percentile lengths of hospitalizations for pertussis were 6 and 12 days during 1978-1993 and 3 and 7 days during 1994-2011.
In the pertussis cohort, 90 children were diagnosed with epilepsy (incidence rate, 1.56 [95% CI, 1.55-1.57] per 1000 person-years), compared with 511 children in the comparison cohort (incidence rate, 0.88 [95% CI, 0.88-0.88] per 1000 person-years). The cumulative incidence of epilepsy at age 10 years was 1.7% (95% CI, 1.4%-2.1%) for patients in the pertussis cohort and 0.9% (95% CI, 0.8%-1.0%) for members of the comparison cohort (Figure). The corresponding crude hazard ratio (HR) was 1.8 (95% CI, 1.4-2.2), and the adjusted HR was 1.6 (95% CI, 1.3-2.1) (Table 2). The HR did not vary during the follow-up period, and epilepsy cases accumulated evenly over time. Hazard ratios varied from 1.7 to 2.3 in the predefined “age at diagnosis” categories younger than 3 years. Patients older than 3 years when diagnosed with pertussis were not at increased risk of epilepsy compared with the general population cohort (HR, 1.0 [95% CI, 0.5-1.8]). The HR was 1.3 (95% CI, 1.0-1.8) for short pertussis admissions and 2.1 (95% CI, 1.5-2.9) for long admissions. Epilepsy risk was increased for patients with pertussis born preterm (weeks 33-36: HR, 1.6 [95% CI, 0.8-2.8]), although not statistically significantly, as well as at term (weeks 39-41: HR, 2.0 [95% CI, 1.5-2.6]) compared with members of the general population cohort of identical gestational age. The HR for postterm patients (>41 weeks) was 0.6 (95% CI, 0.1-1.3). The HR for children born after 1997 was 1.1 (95% CI, 0.5-2.0). According to tests of heterogeneity, there was no statistically significant variation of HRs in the subgroup analyses (all P values >.05). Adjustment for maternal smoking in the subcohort born after 1991 did not affect the HR (Table 2).
Although the absolute risk of epilepsy was low, we found an increased risk of childhood-onset epilepsy associated with a hospital contact for pertussis, compared with children from the general population. To our knowledge, this is the first study to examine risk of epilepsy following pertussis. According to these data, epilepsy risk did not vary according to the specified categories of age at pertussis diagnosis. It has been reported that fewer preterm children born between 1990 and 2004 were vaccinated for B pertussis compared with term children born during this period.19 In line with these findings, the proportion of children born preterm was approximately twice as high in the pertussis cohort as in the comparison cohort. However, our data did not suggest a stronger association between pertussis and epilepsy in children born preterm compared with those born at term. The decreased relative risk for epilepsy observed among the postterm children should be interpreted with caution owing to low precision of the estimate and a relatively high baseline risk of epilepsy in this subgroup.
The excess epilepsy risk among patients in the pertussis cohort relative to the comparison cohort appeared to be attenuated after 1997, but there were relatively few cases during this period, and the risk was not statistically significantly different than during the earlier birth period. However, the apparent attenuation also was associated with the introduction of the pertussis toxoid vaccine in 1997. Similar vaccination coverage (more than 95% for at least 1 dose of pertussis vaccine) has been reported for the period before (1994-1996) and after (1997-2001) introduction of the pertussis toxoid vaccine. It is important to note that the new vaccination schedule introduced in 1997 left the youngest infants unvaccinated for a longer period compared with the previous schedule.14 However, data indicate better immunity from the toxoid vaccine compared with the whole-cell vaccine schedule it was replacing17,20 and thus potentially a milder clinical course of the B pertussis infection. Thus, a recent (2015) US study indicates lower relative infant mortality following pertussis in infants vaccinated with the pertussis toxoid than in unvaccinated individuals (1997-2008; adjusted odds ratio, 0.17 [95% CI, 0.04-0.73]) compared with the corresponding relative mortality in infants vaccinated with the whole-cell vaccine (1991-1996; adjusted odds ratio, 0.67 [95% CI, 0.11-2.82]).21 On the other hand, improvements in hospital care during the study period may have decreased the complication rate.
The lengths of hospital stay for pertussis declined between the 1978-1993 period and the 1994-2011 period. This may indicate decreased pertussis severity, as well as secular trends in treatment practices and a general decrease in length of hospitalization over time. However, there was no association between length of stay as a proxy for severity and epilepsy risk.
The following potential limitations are important for interpreting the results of this observational follow-up study. Our study was based on nationwide population-based registries, taking advantage of the unique opportunity for data linkage provided by the Danish Civil Registration System. Thus, follow-up for death, emigration, and hospital contacts was virtually complete. The Danish health care system, which provides care for pertussis and epilepsy, was publicly funded with equal access for all Danish citizens throughout the study period. This minimized selection biases, potentially affecting probabilities of receiving pertussis or epilepsy diagnoses. The positive predictive value of epilepsy diagnoses recorded in the DNPR is 81%, based on medical record review and strict diagnostic criteria developed by the International League Against Epilepsy as reference.15 Any misclassification is unlikely to be associated with an earlier infection with pertussis, and thus any bias resulting from this nondifferential outcome misclassification would lead to underestimation of the association between pertussis and epilepsy.22 An Australian study found a high positive predictive value (97%) of ICD-10 pertussis discharge diagnoses in a 2003 study of children undergoing diagnostic tests for pertussis.23 The DNPR captures data only for children who were hospitalized or treated in hospital-based outpatient departments. Thus, the study results may not necessarily generalize to children with mild pertussis who are not seen in hospital settings.
Potential pathophysiological mechanisms underlying the observed association include hypoxic brain damage from coughing, perhaps via increased intrathoracic and intra-abdominal pressure and central nervous system hemorrhages.1 Also, the role of immunity and inflammatory processes in epilepsy is increasingly recognized, following the identification of proinflammatory markers and autoantibodies in several epileptic disorders of unknown etiology.24 However, the basis for these processes has not been established. Too few of the patients with pertussis were diagnosed with complications such as postpertussis seizures to examine the effects of complications on risk of epilepsy in this study.
In Denmark, risk of epilepsy was increased among children with hospital-based diagnoses of pertussis infections compared with the general population; however, the absolute risk was low.
Corresponding Author: Morten Olsen, MD, PhD, Department of Clinical Epidemiology, Aarhus University Hospital, Olof Palmes Allé 43-45, 8200 Aarhus N, Denmark (firstname.lastname@example.org).
Author Contributions: Dr Olsen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Olsen.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Nielsen.
Obtained funding: Sørensen.
Administrative, technical, or material support: Sørensen.
Study supervision: Sørensen.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. The Department of Clinical Epidemiology, Aarhus University Hospital, receives funding for other studies from companies in the form of research grants to (and administered by) Aarhus University.
Funding/Support: The study was supported by grants from the Program for Clinical Research Infrastructure (PROCRIN) established by the Lundbeck Foundation and the Novo Nordisk Foundation.
Role of the Funders/Sponsors: The study funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.