Immunocompetent Children Account for the Majority of Complications in Childhood Herpes Zoster

Herpes zoster (HZ) results from reactivation of latent varicella-zoster virus infection that has been acquired with chickenpox. Although HZ may account for relevant morbidity in adults [1], which is related primarily to postherpetic neuralgia and ophthalmic zoster, HZ in children is believed to be benign unless the child is immunocompromised [2]. We performed a population-based study to assess the incidence of hospitalization due to HZ, to assess the relative proportions of healthy and immunocompromised children who were hospitalized, and to describe the clinical features of associated complications.

Methods and subjects. Active surveillance for HZ in children <17 years of age who were admitted to hospitals for treatment of HZ was conducted during the 2-year period of January 2003 through December 2004; the surveillance included all German pediatric hospitals. Information on each child was ascertained via a detailed questionnaire.

A case of HZ was defined as either hospitalization because of HZ or hospitalization because of any other reason if, ⩽6 days after admission, HZ was identified as the underlying cause. Cases of HZ that erupted while a child was hospitalized for other reasons were excluded. Complications were reported in the following categories: skin infection, ophthalmic zoster, zoster oticus, generalized zoster, and other complications. If meningitis or encephalitis was reported, the diagnosis was summarized as "meningoencephalitis." Children were considered to be immunocompromised if they either received immunosuppressive therapy or had conditions characterized by immunodeficiency.

To assess underreporting, a second source was in place in a defined region—North-Rhine Westphalia (NRW), the most populous federal state of Germany (3,248,227 children ⩽17 of age years during 2003 [http://www.destatis.de]). This second source was based on reports from ∼10,000 general practitioners and pediatricians in primary care. Because these 2 sources can be assumed to be independent and confined to a closed catchment area, capture-recapture incidence estimation was applied.

To compare these estimates with the 10th edition of the International Classification of Diseases (ICD10) inpatient-discharge records of HZ (B01.0–B01.9) in any diagnostic position, all 68 pediatric hospitals in NRW were requested to provide these data for the same 2-year period as the primary study. Reports from the 38 compliant hospitals were used to extrapolate the number of all hospitalizations due to HZ.

Estimations of incidence were based on a population of 14,106,043 children ⩽17 years of age during 2003 in Germany (http://www.destatis.de). Stata 8.2 (Stata) was used for all data management and statistical analyses. Median and interquartile range (IQR) values were used as the average value and measure of variation throughout the analysis. Fisher's exact test was used for statistical comparison of categorical data.

The study was approved by the ethics committee of the medical faculty of Ludwig-Maximilian University (Munich) and by the Bavarian data-protection agency.

Results. During a 2-year period (January 2003 through December 2004), results on 244 children hospitalized because of HZ were reported. Of these children, 144 (59%) had no underlying chronic disease, 78 (32%) were considered to be immunocompromised, 9 (4%) were reported to have atopic dermatitis, and 13 (5%) had other chronic diseases; of the 78 immunocompromised children, 74% had a hemato-oncologic disease, with leukemia being the most frequent diagnosis (n=34).

Of the 244 children studied, 112 (46%) were girls. The median age at admission was 9.0 years (IQR, 4.9–13.1 years), with an equal distribution over all age groups; 10 children were admitted during their first year of life, and 10 children were admitted during their second year of life. The average duration of hospital treatment was 7 days (IQR, 5–10 days); 22 children (9%) were hospitalized for >14 days (table 1).

The age when chickenpox was contracted was reported for 74 children. The overall median time between chickenpox and diagnosis of HZ was 4 years (IQR, 1–7 years); for immunocompetent children it was 5 years, and for immunocompromised children it was 3 years (P=.093). The median age when chickenpox was contracted was 1 year for immunocompetent children and 3 years for immunocompromised children. In 6 of 10 children diagnosed as having HZ during the first year of life, the time of varicella infection was recorded: 5 of these 6 children became infected with varicella during the first 3 months of life; in the remaining case, the mother had contracted varicella during the third trimester. For 1 child, no overt infection with varicella was remembered.

Thoracic dermatomes (44%) and trigeminal dermatomes (32%) were the most frequent skin localizations of HZ. In 13 (5%) of the children, 2 distinct dermatome areas were affected. In 8% of the children, a cervical dermatome was affected, and the exanthema was considered to be generalized in 4% of the children. The distribution of affected dermatomes differed significantly between immunocompromised and immunocompetent children (P=.004). Lumbal/sacral dermatomes were seen predominately in immunocompromised children (15/70 [21%]) and less often in immunocompetent children (12/155 [8%]), whereas trigeminal dermatomes were seen predominantly in immunocompetent children (60/155 [39%]) and less often in immunocompromised children (12/70 [17%]).

Rash-related pain was noted in 98 (40%) of the children. The duration of the pain was mostly limited to the duration of HZ rash. In 4 children, the pain persisted after healing of the exanthema, suggesting postherpetic neuralgia; 1 of these 4 children was immunocompetent (pain for ∼14 days), and the other 3 (all with persistent pain for >4 weeks) were immunocompromised.

Complications of HZ were observed in 115 (47%) of the 244 children. Whereas 96 (58%) of 166 immunocompetent children had a specified complication, only 19 (25%) of 66 immunocompromised children did (P<.001). The most frequent complications were the following: superinfection of the skin (in 42 [17%] of the children), ophthalmic zoster (in 29 [12%] of the children, 10 of whom had keratitis), zoster oticus (in 23 [9%] of the children, 11 of whom had facial nerve paralysis [Ramsay Hunt syndrome] and 3 of whom had involvement of the vestibular nerve), and meningoencephalitis (in 22 [9%] of the children) (table 2). Except for generalized HZ, no complications were statistically significantly more common in immunocompromised children. Zoster oticus, ophthalmic zoster, meningoencephalitis, and complications of the upper respiratory tract and of the ear, nose, and throat were reported more often in immunocompetent children (table 2).

Children with generalized HZ (median age, 4.7 years) and with skin infectious complications (median age, 6.5 years) were younger, whereas children with meningoencephalitis (median age, 13.0 years) and zoster oticus (median age, 10.6 years) were older, than the median age of all children (9 years). The difference in age was statistically significant only for meningoencephalitis (P<.001).

At discharge from hospitalization, 7 children had facial-nerve paralysis that subsequently was reported to have persisted for up to 5 months (last available information). Also at discharge from hospitalization, severe scarring was suspected in 6 children, as were hearing loss in 1 child and restricted vision related to keratitis in 1 child.

On the basis of data on the 244 patients in the present study, the incidence of hospitalization in Germany was 0.9/100,000 children/year, up to the age of 17 years. To account for underreporting, the incidence was estimated on the basis of 2 surveillance sources in NRW. A total of 60 individual reports of HZ were identified via hospital surveillance (ESPED) and medical-practice surveillance; hospital surveillance alone identified 46 patients, and the second source identified 16 patients (2 children were identified via both surveillance sources). Thus, the incidence of hospitalization was estimated as 4.1/100,000 children/ year (95% confidence interval [CI], 0.6–7.6/100,000 children/ year), up to the age of 17 years; by comparison, the incidence of hospitalization as extrapolated from 83 hospitalizations for the corresponding ICD10 inpatient-discharge records by 38 (55%) of the 68 pediatric hospitals in NRW was 2.3/100,000 children/ year.

Discussion. To our knowledge, the data reported herein represent the largest population-based series of children hospitalized with HZ and its complications. The primary, and surprising, finding is that immunocompetent children accounted for most hospitalizations and for the majority of complications.

As in previous studies, the occurrence of varicella during the first year of life was confirmed to be a risk factor for HZ during childhood [3, 4]: in the present study, 32% of the patients with HZ had varicella during their first year of life, compared with a German seroepidemiologic study's finding that, in the general population, 9% of children are seropositive at the end of their first year of life [5]. Furthermore, postherpetic neuralgia was observed only rarely (i.e., in only 4 children) in the present study, confirming that it is a rare event during childhood [3, 6]—in contrast with a risk of ∼50% in patients with HZ who are ⩾60 years of age [7]. The 3 children in the present study who had neuralgia for >1 month were immunocompromised.

In 75% of the immunocompromised children, there was no specific complication, suggesting that the immunocompromised children had been hospitalized predominantly for preemptive virostatic therapy; in contrast, immunocompetent children accounted for 84% of the reported complications. There were a remarkable number of patients with ophthalmic zoster, zoster oticus, and meningoencephalitis, constituting 64% of the complications in immunocompetent children. Meningoencephalitis is generally believed to be a rare complication of HZ, with a benign course in immunocompetent patients and a less favorable prognosis in immunocompromised patients [7, 8]. All of the patients in the present study recovered completely.

Ophthalmic zoster, irrespective of the child's age or the severity of the rash, is known to be associated with complications in ∼50% of patients who do not receive virostatic treatment [9]. Treatment reduces the complication rate to ∼20% [10]. In the present study, all 29 patients with ophthalmic zoster, including the 10 who also had keratitis, had been treated with acyclovir. However, 1 immunocompetent child with a suspected vision impairment was discharged from hospitalization.

Zoster oticus is rarely reported in the pediatric literature [11]. Yet, in the children in the present study, zoster oticus with facial-nerve paralysis constituted a considerable number of the complications; 7 children had persistent facial-nerve paralysis after they were discharged from hospitalization. It has been reported that facial paralysis due to zoster oticus has a less favorable prognosis than does Bell palsy [12].

The incidence of hospitalization because of HZ must be seen in the context of the incidence of HZ in children, which is ∼1/ 1000 children/year [4]. If we consider only hospital reports of HZ—for which the incidence is 0.9/100,000/year—then ∼1/100 children with HZ are hospitalized in Germany. Because, in the present study, 39% of the hospitalizations were related to complications in immunocompetent children, it must be assumed that ⩾1/250 immunocompetent children with HZ will be hospitalized because of complications, an incidence that is consistent with the results of 2 population-based studies: Guess et al. [3] reported 2 hospitalizations in 168 immunocompetent children with HZ, and Petursson et al. [13] reported 0 in 86 immunocompetent children.

Although our estimate of the incidence of hospitalization (0.9/100,000 children/year), which is based on cases reported by hospitals, is within the range reported by others [3, 14], it must be assumed that this figure reflects the minimum incidence. The capture-recapture estimate suggests that the reported number of hospitalizations is less than a quarter of all true cases. The 95% CI for the capture-recapture estimate, however, was wide, because there were only 2 linkable cases.

We validated the 2-source estimate of incidence by using ICD10 inpatient-discharge records—diagnosis data and extrapolated, to all hospitals, the number of patients with HZ who were reported by the 38 compliant hospitals; because the assignment of a specific ICD10 discharge-from-hospitalization diagnosis depends more on the administrative facilities of the hospitals than on the number of patients with HZ, this extrapolation seems to be justified. Therefore, the true incidence of hospitalizations because of HZ is likely to be ⩾2.3/100,000 children/ year, which is 2.5 times higher than the estimate based only on active hospital surveillance.

General varicella vaccination has been recommended in Germany only since September 2004. Data reported by Hardy et al. [15] suggest that the risk of HZ may be lower in vaccinated children. Although more data with longer follow-up are needed on immunocompetent children, it is possible that the varicella-vaccination program will also reduce HZ in vaccine recipients.

In conclusion, the present study has shown that • ⩾1/100 children with herpes zoster are hospitalized in Germany; • Immunocompetent children account for most of these hospitalizations; • Although preemptive therapy is a common reason for hospitalization of immunocompromised children, complications are the main reason for hospital admission of immunocompetent children. • ⩾ 1/250 healthy children with HZ will acquire a complication requiring hospitalization, with bacterial superinfection, meningoencephalitis, herpes ophthalmicus, and zoster oticus—with half of these complications being accompanied by Ramsay Hunt syndrome—accounting for the majority of these hospitalized patients.

• Received March 22, 2007.
• Accepted June 1, 2007.

• © 2007 by the Infectious Diseases Society of America