Share this article on:

The Epidemiology of Viral Meningitis Hospitalization in Childhood

Hviid, Anders; Melbye, Mads

doi: 10.1097/EDE.0b013e3181567d31

Background: There have been few long-term population-based studies of viral meningitis, and only a limited number of potential risk factors have been evaluated.

Methods: We estimated the incidence of viral meningitis hospitalization in childhood, and assessed risk factors for this disease through a population-based cohort study comprising all children born in Denmark from 1977 through 2001 (n = 1.5 million). Information on sex, number of children and adults in the household, age of parents at child's birth, degree of urbanization, birth weight, gestational age, and possible complications at birth were linked to the children in the cohort, together with information on hospitalization with viral meningitis. We calculated incidence rates of viral meningitis and estimated rate ratios according to the various risk factors using Poisson regression.

Results: The incidence rate was highest in the first 6 months of life (38.7 per 100,000 person-years), with a peak right after birth (58.7 per 100,000 person-years). A secondary peak was seen among 5-year-old children (15.6 per 100,000 person-years). Overall incidence rates decreased throughout the study period, with outbreaks occurring every 3 to 5 years [decrease in rate ratio per calendar year = 0.95; 95% confidence interval (CI) = 0.94–0.96]. Summer and early fall peaks were present.

We observed independent effects of sex (girls vs. boys: rate ratio = 0.47 [95% CI = 0.43–0.53]), children in the household (eg, living with 3+ younger children vs. living with none: 1.94 [1.22–3.07]), single parenthood (living with 1 parent vs. living with 2 parents: 1.30 [1.12–1.39]), degree of urbanization (children living in the capital vs. children living in small town and rural areas: 1.54 [1.31–1.80]), low birth weight (increase in RR per 500 g reduction in birth weight = 1.05 [1.00–1.09]), prematurity (increase in RR per 1 week reduction in gestational age = 1.03 [1.01–1.04]), and cesarean section (cesarean section vs. vaginal birth: 1.29 [1.12–1.49]).

Conclusions: Incidence of viral meningitis hospitalization is highest immediately after birth with a secondary peak at age 5. Lack of passive maternally acquired antibodies and preferential hospitalization are the likely causes for the peak in infancy. Increased transmission in kindergarten, preschool, and day care could explain the secondary peak. The incidence decreased throughout the 25-year study period perhaps due to improved public hygiene. Among the assessed risk factors, we found the strongest to be male sex, a high number of children in the household, and degree of urbanization.

From the Department of Epidemiology Research, Statens Serum Institut, Copenhagen, Denmark.

Submitted 18 August 2006; accepted 25 July 2007.

Supported by Danish Medical Research Council.

Correspondence: Anders Hviid, Department of Epidemiology Research, Statens Serum Institut, Artillerivej 5, 2300 Copenhagen S, Denmark. E-mail:

Viral meningitis has most commonly been caused by polio virus, nonpolio enteroviruses, lymphocytic choriomeningitis virus, and mumps virus.1 Today, the near-eradication of polio and the wide-scale use of routine mumps vaccination has left the nonpolio enteroviruses as the leading cause of viral meningitis in children, at least in developed countries.2–4

The typical case of childhood viral meningitis is short-lived and benign. However, any meningitis case is the cause of parental and professional alarm and almost all patients with symptomatic viral meningitis are initially hospitalized to ensure that bacterial meningitis (which carries a much greater risk of serious morbidity and mortality) can be excluded.

Previously, only simple epidemiologic characteristics of viral meningitis, (eg, age, calendar period, sex, and season) have been examined in detail. Furthermore, little has been published using population-based data. To determine the incidence of viral meningitis in childhood and to identify and quantify risk factors for hospitalization with this disease, we conducted a cohort study in Denmark. The cohort comprised all children born 1977–2001, with data available from several unique nationwide Danish registries.

Back to Top | Article Outline


Since April 1968, every person living in Denmark has been given a unique identification number in the Danish Civil Registration System.5 Based on this registry we constructed a cohort comprising all children born in Denmark in the period from 1 January 1977, to 31 December 2001. Using the unique personal identification number, we were able to link information on various potential risk factors and information on possible hospitalization with viral meningitis for children in the cohort.

Back to Top | Article Outline

Viral Meningitis

Information on hospitalization with viral meningitis was obtained from the Danish National Hospital Register.6 From 1977 to 1993, ICD-8 (International Classification of Diseases, 8th Revision) has been used to code diagnosis, and since 1994, ICD-10 (International Classification of Diseases, 10th Revision). We used the codes 045.xx, and A87.x to identify cases of viral meningitis. Note that these codes do not include mumps meningitis, which would have been a major contributor to viral meningitis hospitalizations before the introduction of general measles-mumps-rubella vaccination in Denmark in 1987. Consequently, the nonpolio enteroviruses are the likely cause of the majority of cases identified by these diagnostic codes in a developed country such as Denmark.

Back to Top | Article Outline

Demographic Risk Factors

We included the following demographic risk factors: sex; number of younger children, older children, and adults in the household; degree of urbanization of the community of the household; and age of father and mother at birth of child. This information was obtained from the Danish Civil Registration System. Sex of the child is encoded in the unique personal identification number. The Danish Civil Registration System contains links to mother and father (if known), enabling calculation of mother's and father's age at birth (<20, 20–24, 25–29, 30–34, 35+ years). The Registration System contains information on place of residence, both current and past, for all people living in Denmark. We identified all places and dates of residence for children in the cohort. For each residence we identified the degree of urbanization of the community of the household, the number of other household members, and their ages. If the number of household members changed, or if at least 1 household member changed age category (younger or older than cohort child, and adult) during the time at a given residence, this time interval was further divided to accommodate these changes. Thus, for each child in the cohort at any time in the follow-up period, we obtained the following information: number of younger children in the household (0, 1, 2, 3+); the number of older children in the household (0, 1, 2, 3+); number of adults in the household (1, 2, 3+); and degree of urbanization of the community of the household [small towns or rural areas (population less than 10,000), medium-sized towns (population 10,000–99,999), large cities (population 100,000+), capital suburbs (Copenhagen suburbs), and capital (Copenhagen)].

Back to Top | Article Outline

Birth-Related Risk Factors

Information on birth-related risk factors was obtained from the Danish Medical Birth Registry.7 These risk factors included child's birth weight (in 500 g categories plus <1000 g and 5000+ g), child's gestational age (<31, 31–33, 34–36, 37–39, 40–41, ≥42 weeks), 5-minute APGAR score (<6, 6–7, 8–10), use of forceps (yes or no) or vacuum pump at delivery (yes, no), and cesarean section at delivery (yes, no). Data on birth weight were missing for 6.1% of children; on gestational age for 13.1%; on APGAR score 10.6%; on use of forceps 10.1%; on use of vacuum pump 10.1%; and on cesarean section, 9.7%.

Back to Top | Article Outline

Statistical Analysis

Children in our cohort contributed person-time to follow-up from birth or 1 January 1977, whichever occurred last, until first viral meningitis hospitalization, death, disappearance or emigration, attaining 15 years of age, or 31 December 2001, whichever occurred first. The resulting incidence rates (IRs) according to age, calendar period, season, demographic factors, and birth characteristics were further analyzed with Poisson regression (log-linear regression on the viral meningitis counts using the logarithm to the follow-up time as offset), producing estimates of rate ratios (RR) and their associated 95% confidence intervals (CIs).8 We evaluated the possible confounding of these RRs by comparing the results from the univariate models with results from multivariate models containing the risk factor and the possible confounder. If the risk factor RRs changed by more than 10%, we included the confounding variable in the final model. When adjusting for the potential confounding effect of variables with missing values, we used the method of single imputation, replacing a missing value with the most common value.

Back to Top | Article Outline


A total of 1,540,832 live-born children were included in the cohort. The follow-up of 35,242 children (2%) was prematurely terminated because of death (n = 13,975), emigration (n = 20,837), or disappearance from the Registration System (n = 430). During 15,299,937 person-years of follow-up we identified 1642 viral meningitis hospitalizations (374 cases during 1,493,124 person-years among infants up to the first birthday and 1268 cases during 13,699,472 person-years among 1–14 year olds). The mean age ± SD at hospitalization with viral meningitis was 4.9 ± 3.8 years. Few of the viral meningitis cases had supplementary diagnostic information. Only 365 cases had at least 1 supplementary diagnosis, and among these no etiologic diagnoses were prevalent.

In Figure 1 we present incidence rates of viral meningitis according to age. The IR is greatest in the first 6 months of life (38.7 per 100,000 person-years), with a peak right after birth (58.7 per 100,000 person-years). A secondary peak is noteworthy among 5-year-old children (15.6 per 100,000 person-years).



Figure 2 presents rate ratios according to calendar period among infants and children. Overall there was a relative decrease in incidence rates [decrease in rate ratio per calendar year = 0.95 (95% CI = 0.94–0.96)]. Throughout the period there is evidence of an epidemic pattern with outbreaks every 3–5 years.



Figure 3 shows rate ratios according to season among infants and children. For both age groups the rate ratios are lowest in winter-spring and greatest in summer-autumn.



In Table 1 we present rate ratios according to demographic risk factors. We observed a marked protective effect of female sex [girls vs. boys: RR = 0.47 (95% CI = 0.43–0.53)]. The number of younger children in the household was a risk factor for 1 to 14 year olds [eg, living with 3 or more vs. none: RR = 1.94 (1.22–3.07)]. Conversely, the number of older children in the household was a risk factor for infants [living with 1 vs. none: RR = 2.28 (1.80–2.90)]. Living with only 1 adult in the household (single parents) was a risk factor [living with 1 vs. 2: RR = 1.30 (1.12–1.49)]. Increasing degree of urbanization was associated with increasing risk, which was most marked among infants [eg, capital area vs. small towns or rural areas: RR = 2.00 (1.47–2.74)].



Table 2 presents rate ratios according to birth-related risk factors. Low birth weight was a risk factor among infants [increase in RR per 500 g reduction in birth weight = 1.12 (1.04–1.21)] and low gestational age was a risk factor throughout childhood [increase in RR per 1 week reduction in gestational age = 1.03 (1.01–1.04)]. Comparing APGAR scores less than 8 with scores of 8 to 10 resulted in increased risk of viral meningitis for scores 6 to 7 but not less than 6. Finally, cesarean section was a risk factor [children delivered by cesarean section vs. children vaginally delivered: RR = 1.29 (1.12–1.49)]. Among children born from 1994 through 2001 we had further information on whether cesarean section was before or during labor. We found that cesarean section before labor was a risk factor [ cesarean section before labor vs. vaginal: RR = 1.73 (1.12–2.67)] whereas cesarean section during labor was not [RR = 1.00 (0.59–2.67)].



We estimated the birth weight trends excluding the outlier groupings of less than 1000 g and 5000 g or more. This yielded a stronger trend in the increased RR per 500 g reduction in birth weight: among infants, 1.15 (1.05–1.25) and 1 to 14 year olds, 1.04 (0.99–1.10). We further estimated the trends among children born at 37 to 41 weeks of gestation. In this group of children there was little or no effect of birth weight overall [increase in RR per 500 g reduction in birth weight = 1.01 (0.95–1.07)] or among infants [RR = 1.05 (0.95–1.18)].

Thus, we identified the following risk factors for viral meningitis hospitalization: age, calendar period, season, sex, number of children in the household, number of adults in the household, urbanization, gestational age, APGAR score, and cesarean section.

Back to Top | Article Outline


We evaluated the epidemiologic pattern of viral meningitis hospitalization in childhood in Denmark according to age, calendar period, season, demographic factors, and birth characteristics. Denmark is a highly industrialized Nordic country with a high standard of living and free healthcare for all. In a setting such as Denmark and during the 1977–2001 study period, mumps virus and nonpolio enterovirus would have been responsible for the majority of viral meningitis cases. Considering the diagnostic codes (which excludes mumps meningitis) used here, and assuming no diagnostic code misclassification, the majority of cases are likely caused by nonpolio enterovirus.

We observed a biphasic age pattern, with a primary peak in infancy and a secondary peak at preschool age. The primary peak is likely due to infants' lack of previous exposure and immunity; they are depending solely on their mother's antibody repertoire through passive maternally acquired antibodies. The secondary peak could be explained by increasing opportunity for exposure in day care, preschool, and kindergarten. Both types of peaks have previously been reported independently and together.9–11 Furthermore, hospitalization propensity may have exaggerated the infant peak somewhat relative to other age groups, among whom similar symptoms would be less likely to warrant hospitalization.

Vestergaard et al12 reported on enterovirus infections in Denmark in 1997–2000 based on nationwide laboratory data. In 1997–1999, roughly 80% of about 100 yearly enterovirus-positive patients were less than 15 years of age. This age distribution for enterovirus infections is similar to what has been observed elsewhere.13 The year 2000 was an outbreak year, with 306 patients and more adult cases than in previous years.

We observed a decreasing secular trend, with outbreaks every 3 to 5 years, and a seasonal distribution, with a summer and early fall peak, similar to previous reports.9,14 The secular pattern reflects a number of factors. One possible set of causes is the endemic circulation of some serotypes, the epidemic cycling of other serotypes, and the possible emergence of “new” serotypes.12,15 Second, improvements in public hygiene during the 25-year study period may have contributed. Enterovirus is primarily transmitted through the fecal-hand-oral route, and increased public hygiene is a likely explanation for the overall decrease in viral meningitis throughout the study period. The seasonal distribution is what has typically been observed in temperate climates, with the cause not well understood.

The incidence rates of viral meningitis in our study are somewhat lower than rates reported elsewhere. Khetsuriani et al14 reported US average annual hospitalization rates of 213 per 100,000 infants, 14 per 100,000 1 to 4 year olds, and 14 per 100,000 5 to 19 year olds. Except in infancy, these numbers are not dissimilar to our overall incidence rate of 10.7 per 100,000 person-years.

Almost twice as many boys as girls were hospitalized with viral meningitis. This is consistent with earlier reports9; a male predominance is also seen for other infectious diseases.16,17 Because Denmark is a highly industrialized society, behavioral and sociocultural factors or preferential hospitalization of boys are unlikely to explain the difference, and a biologic explanation is more plausible. However, little research in sex-related difference in immune function in children has been conducted.

We found that younger children in the household (younger siblings most likely) increased the risk for 1–14 year olds and that older children (older siblings, most likely) increased the risk for infants (likely a direct result of increased probability of contact with an infectious case).18 Because older siblings of infants are likely to be themselves rather young children,19 our results are consistent with the underlying epidemiology of viral infection represented by the observed age pattern in our study.

Children living with only 1 adult (a single parent, most likely) were at an increased risk of viral meningitis compared with children living with 2 adults (parents, most likely). Given the egalitarian nature of Danish society, this is unlikely to represent an association between a possible socioeconomic disadvantage and infection. A more likely explanation is that children of single parents more often have to use day-care facilities, resulting in increased probability of contact with an infectious case.18

Our finding of an increasing risk of viral meningitis with increasing population density is a likely result of urban crowding that increases the opportunity for contact with an infectious case.

We observed an increased risk of viral meningitis associated with poor birth outcomes represented by low birth weight, prematurity, some APGAR scores, and cesarean section. The effects of birth weight (which was not present in term children) and gestational age is likely a result of preterm infants having lower levels of passively acquired maternal antibodies.20 The inconsistencies of the APGAR score results renders them uninterpretable. The association between cesarean section and viral meningitis is intriguing. In a subpopulation we found the cesarean section effect to be restricted to cesarean section before onset of labor, with no association if cesarean section was conducted during labor. If the same pattern exists in the whole population, this suggests that the effect may not be directly related to mode of delivery (cesarean section vs. vaginal) but rather to certain circumstances surrounding cesarean section before labor. Cesarean sections before labor onset are often administered in problematic pregnancies. However, the effect was not confounded by birth weight, gestational age, or 5-minute APGAR score—factors often associated with problematic pregnancies. Hospitalization propensity should also be considered as an explanation for the association between poor birth outcomes and viral meningitis.

This is the first epidemiologic study of viral meningitis of this size and with such a wide range of potential risk factors. The cohort design has several strengths. It was based on information from nationwide Danish registries that contain mandatory reportable information on all Danish residents and are considered very complete. Information on covariates was collected independently of the outcome; thus, selection bias and differential recall are unlikely to have influenced the results. Furthermore, the personal identification number that linked all registries and the daily updated information regarding people who moved or died makes follow-up very complete. Taken together with the large size of the cohort we consider the results to be very robust.

In conclusion we found the incidence of hospitalization with viral meningitis to peak in the first 6 months of life with a secondary peak during preschool age. The incidence decreased throughout the study period. Among the risk factors we assessed, sex of the child, number of children in the household, and degree of urbanization were the strongest.

Back to Top | Article Outline


1. Meyer Hm Jr, Johnson RT, Crawford IP, et al. Central nervous system syndromes of “vital” etiology. A study of 713 cases. Am J Med. 1960;29:334–347.
2. Connolly KJ, Hammer SM. The acute aseptic meningitis syndrome. Infect Dis Clin North Am. 1990;4:599–622.
3. Sawyer MH. Enterovirus infections: diagnosis and treatment. Semin Pediatr Infect Dis. 2002;13:40–47.
4. Rotbart HA. Viral meningitis. Semin Neurol. 2000;20:277–292.
5. Malig C. The civil registration system in Denmark, IIVRS Technical Paper, No. 66. International Institute for Vital Registration and Statistics, Bethesda, Maryland, 1996.
6. Andersen TF, Madsen M, Jorgensen J, et al. The Danish national hospital register. A valuable source of data for modern health sciences. Dan Med Bull. 1999;46:263–268.
7. Knudsen LB, Olsen J. The Danish medical birth registry. Dan Med Bull. 1998;45:320–323.
8. Clayton D. Hills M. Poisson and logistic regression. In: Statistical Models in Epidemiology. 1st ed. Oxford, United Kingdom: Oxford University Press; 1993:227–236.
9. Lee KY, Burgner D, Lee HS, et al. The changing epidemiology of pediatric aseptic meningitis in Daejeon, Korea from 1987 to 2003. BMC Infect Dis. 2005;5:97.
10. Kirschke DL, Jones TF, Buckingham SC, et al. Outbreak of aseptic meningitis associated with echovirus 13. Pediatr Infect Dis J. 2002;21:1034–1038.
11. Bottner A, Daneschnejad S, Handrick W, et al. A season of aseptic meningitis in Germany: epidemiologic, clinical and diagnostic aspects. Pediatr Infect Dis J. 2002;21:1126–1132.
12. Vestergaard HT, Johnsen CK, Bottiger B. An unusual enterovirus outbreak in Denmark: clinical characteristics and molecular epidemiology. Scand J Infect Dis. 2004;36:840–847.
13. Nairn C, Clements GB. A study of enterovirus isolations in Glasgow from 1977 to 1997. J Med Virol. 1999;58:304–312.
14. Khetsuriani N, Quiroz ES, Holman RC, et al. Viral meningitis-associated hospitalizations in the United States, 1988–1999. Neuroepidemiology. 2003;22:345–352.
15. Diedrich S, Schreier E. Aseptic meningitis in Germany associated with echovirus type 13. BMC Infect Dis. 2001;1:14.
16. Jensen-Fangel S, Mohey R, Johnsen SP, et al. Gender differences in hospitalization rates for respiratory tract infections in Danish youth. Scand J Infect Dis. 2004;36:31–36.
17. Green MS. The male predominance in the incidence of infectious diseases in children: a postulated explanation for disparities in the literature. Int J Epidemiol. 1992;21:381–386.
18. Reintjes R, Pohle M, Vieth U, et al. Community-wide outbreak of enteroviral illness caused by echovirus 30: a cross-sectional survey and a case-control study. Pediatr Infect Dis J. 1999;18:104–108.
19. Hviid A, Stellfeld M, Wohlfahrt J, et al. The impact of pre-school booster vaccination of 4–6-year-old children on pertussis in 0–1-year-old children. Vaccine. 2006;24:1401–1407.
20. Malek A, Sager R, Kuhn P, et al. Evolution of maternofetal transport of immunoglobulins during human pregnancy. Am J Reprod Immunol. 1996;36:248–255.
© 2007 Lippincott Williams & Wilkins, Inc.