Skip Navigation

Epidemiology of Primary Varicella and Herpes Zoster Hospitalizations: The Pre-Varicella Vaccine Era

  1. Felix Lin1,a and
  2. James L. Hadler2
  1. 1Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven
  2. 2Infectious Diseases Division, Connecticut Department of Public Health, Hartford, Connecticut
  1. Reprints or correspondence: Dr. James Hadler, Infectious Diseases Division, Connecticut Department of Public Health, 410 Capitol Avenue, MS #11-FDS, P.O. Box 340308, Hartford, CT 06134-0308 (james.hadler{at}po.state.ct.us).

  1. Presented in part: 31st Centers for Disease Control and Prevention National Immunization Conference, Detroit, May 1997 (abstracts 33 and 150).

  • ↵a Present affiliation: University of New Mexico School of Medicine, Albuquerque.

 
Next Section

Abstract

To determine the epidemiology and costs of hospitalization with primary varicella and herpes zoster in the prevaccine era and the usefulness of hospital discharge data to determine the population impact of vaccination on these conditions, statewide hospital discharge data in Connecticut from 1986 to 1995 were analyzed. Annual hospitalizations for herpes zoster were 4-fold higher than for primary varicella (16.1 vs. 4.1/100,000). Overall, 69% and 83%, respectively, had no underlying immunosuppressive conditions. Regarding primary varicella, 53% of patients were aged <15 years, there was a marked winter-spring seasonality, and Hispanics and blacks were 4.1 and 2.6 times more likely than whites to be hospitalized. Regarding herpes zoster, 66.9% of patients were aged >64 years, and there was no seasonality. The mean patient charges in 1995 were $12,819 for primary varicella and $15,583 for herpes zoster. Analysis of population-based hospital discharge data is a feasible means of monitoring the impact of varicella immunization on severe morbidity due to primary varicella and herpes zoster.

In May 1995 and June 1996, respectively, the American Academy of Pediatrics and the Advisory Committee on Immunization Practices from the Centers for Disease Control and Prevention recommended the universal vaccination against varicella for infants, susceptible 11- and 12-year-old children, and other susceptible persons at high risk of exposure or complications [1, 2]. As use of varicella vaccines becomes widespread, the epidemiologic features of primary varicella and herpes zoster are expected to change.

Common perceptions regarding varicella vaccine are as follows: primary varicella is a benign disease for which routine vaccination is of questionable benefit, the duration of immunity conferred by the currently licensed vaccine may be shorter than expected, the vaccine may potentially create a susceptible older population at risk of severe complications from primary varicella, and decreased circulation of varicella could result in an increase in herpes zoster [3–5]. Thus, it is important to effectively describe the epidemiology of complications from primary and latent varicella in the prevaccine and vaccine eras and to monitor the extent to which severe and costly complications continue to occur.

Severe and costly complications from latent infection are of particular interest. Although current data suggest that vaccination of persons with certain immunocompromising conditions with the live attenuated varicella vaccine will prevent more herpes zoster than it will cause [6–9], the long-term impact on the incidence of herpes zoster remains uncertain. On one hand, to the extent that the varicella vaccine fully prevents primary infection, it should also prevent latent infection and its consequences. In addition, given that breakthrough varicella is generally milder than varicella in unvaccinated persons [10], it is conceivable that latent infection will be less consistently established. On the other hand, the vaccine could indirectly increase the incidence of herpes zoster among persons who have a latent varicella infection. It is thought that periodic boosting of cellular immunity by repeated exposure to varicella virus may be a factor in preventing or attenuating herpes zoster [11]. In the future, a decrease of circulating varicella virus from widespread vaccination will reduce the potential for such boosting and could result in more latently infected persons developing zoster. Additionally, the vaccine strain itself can cause herpes zoster [12], although this appears to be uncommon. Because of such concerns, cost-benefit studies of varicella vaccine have not incorporated any impact of vaccination on the incidence and costs of herpes zoster [13–17]. Thus, it is important not only to describe the epidemiology of primary varicella but also to effectively describe the epidemiology and costs of latent varicella infection over time.

In this context, use of comprehensive population-based hospital discharge databases are a potential means of accomplishing such surveillance. Such systems provide a complete record of all hospitalizations and are not subject to underdiagnosis and deficiencies in reporting that limit surveillance systems of outpatient diseases. Additionally, hospitalization databases measure the severe and costly end of the disease spectrum, on which vaccination is likely to have the greatest impact. Although the epidemiology and natural history of herpes zoster in the prevaccine era have been described elsewhere [18–22], there are no published population-based studies making use of hospital discharge data. Additionally, there is scant reference in the medical literature to its cost impact overall or relative to primary varicella [23].

This study was undertaken to obtain population-based estimates of hospitalization rates for primary varicella and herpes zoster in the 10-year period prior to widespread vaccine availability, to determine the epidemiology of hospitalizations for primary varicella and herpes zoster, to obtain measures of trends in the cost of hospitalizations due to primary varicella and herpes zoster, and to determine whether monitoring of population-based hospital discharge data may be an effective way to monitor the impact of immunization on the epidemiology of complications due to primary and latent varicella.

Previous SectionNext Section

Methods

Source population

Data from the Connecticut Health Information Management and Exchange (CHIME) Program [24] were obtained for hospitalizations of Connecticut residents with primary varicella or herpes zoster. CHIME is a statewide information service containing specific data on inpatient hospital discharges from all 36 Connecticut general, acute care, and nongovernmental hospitals. The data include inpatient discharge data from Connecticut residents who were treated at hospitals in border states (New York, Massachusetts, and Rhode Island). The input data, submitted by individual hospitals or their data processing services, have been compiled and converted into standard definitions according to established CHIME policies and procedures [24]. Records were requested of all hospitalized Connecticut residents discharged between 1 January 1986 and 31 December 1995 with a principal or secondary diagnosis code of primary varicella or herpes zoster. Primary varicella hospitalization was defined as a hospital discharge with a principal or secondary discharge diagnosis code of varicella (052.0–052.9) according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) [25]. Herpes zoster hospitalization was defined as a hospital discharge with a principal or secondary discharge diagnosis code of zoster (053.00–053.99). Records with diagnostic codes for both primary varicella and herpes zoster were excluded from the analysis, so that the epidemiology and costs of each disease could be examined separately.

Demographic data

Data available through CHIME without individual hospital consent include admission year, calendar quarter of hospitalization, age, race/ethnicity, sex, the principle diagnostic code and up to nine secondary hospital discharge diagnostic codes, length of hospitalization, disposition (death, discharge, transfer), total charge, and payer. A person was considered to have an underlying condition if there was a discharge diagnostic code for a condition known to increase the risk for occurrence or severity of primary varicella or herpes zoster, such as immunosuppression secondary to therapy, malignancy, pregnancy, or other disease (table 1). A hospitalization was considered to have a specific cornplication of primary varicella if there was a varicella-specific discharge diagnostic code that indicated specific organ involvement (052.1, 052.2, 052.7, 052.8) or an ICD-9-CM code indicating a varicella-zoster-associated bacterial superinfection (table 1). Similarly, a hospitalization was considered to have a specific complication of zoster if there was a zoster-specific discharge diagnostic code that indicated specific organ involvement or an ICD-9-CM code indicating a varicella-zoster-associated bacterial superinfection (table 1).

Statistical analysis

Hospitalization rates were calculated with denominator data for Connecticut obtained from the 1990 census [26]. Overall population estimates obtained between the years 1986 and 1995 revealed that the group-specific population of Connecticut was nearly in a steady state [27]. In 1990, the population of Connecticut was 3,287,116 persons, of whom 83.8% were non-Hispanic white, 7.9% were non-Hispanic black, and 6.5% were Hispanic. Percentages of persons with underlying conditions or specific complications were determined by use of the total number of discharges with varicella or herpes zoster as the denominator. A ⩽5% level of probability for a given finding was considered to indicate statistical significance. All statistical analyses were done with Epi Info version 6.04 (Centers for Disease Control and Prevention, Atlanta) and SAS software (SAS Institute, Cary, NC).

Previous SectionNext Section

Results

Over the 10-year study period, there were 1341 identified primary varicella-related hospital discharges, an average of 134 per year (range, 86–178; figure 1). The average annual primary varicella hospitalization rate over the 10-year period was 4.1 hospitalizations per 100,000 population (range, 2.6–5.4). Over the same period, there were 5274 herpes zoster-related hospital discharges, an average of 527 per year (range, 468–596; figure 1). The average annual herpes zoster hospitalization rate over the 10-year period was 16.1/100,000 population (range, 14.5–18.2). The average annual herpes zoster hospitalization rate was 3.9 times higher than the primary varicella hospitalization rate during the same time period. In addition, there were 33 persons with diagnostic codes for both primary varicella and herpes zoster who were excluded from the diseasespecific analyses.

Figure 1
View larger version:
Figure 1

No. of persons hospitalized with primary varicella (PV) and with herpes zoster (HZ), by year, Connecticut, 1986–1995.

Patient characteristics

Tables 2 and 3 show the distribution of primary varicella and of herpes zoster hospitalizations by the available demographic characteristics and by the quarter of year in which they occurred. Fifty-three percent of the primary varicella hospitalizations were of persons <15 years of age, and ∼29% were of persons between 20 and 34 years old. Examination of these 2 age groups by year revealed modest non-statistically significant (P > .05) variation (range, 48%–61% and 19%–36%, respectively).

Figure 2
View larger version:
Figure 2

Average charge (US$) per hospitalization with primary varicella (PV) and with herpes zoster (HZ), by year, Connecticut, 1986–1995.

Table 1
View larger version:
Table 1

International Classification of Diseases, Ninth Revision, Clinical Modification codes used to determine underlying conditions and complications for hospitalizations for primary varicella or herpes zoster (HZ).

Table 2
View larger version:
Table 2

Number of cases and average annual rates of hospital discharges with varicella by selected demographic characteristics and by quarter of year, Connecticut, 1986–1995.

Table 3
View larger version:
Table 3

Number of cases and average annual rates of hospital discharges with herpes zoster by selected demographic characteristics and by quarter of year, Connecticut, 1986–1995.

Marked differences in primary varicella hospitalization rates by race/ethnicity were found (table 2). Overall, Hispanics were 4.1 times and blacks were 2.6 times more likely to be hospitalized than were whites. Higher primary varicella hospitalization rates for Hispanics and for blacks than for whites were found across each age group and each year of the study period.

In contrast to primary varicella, the number and rate of hospitalizations with herpes zoster increased progressively with increasing age. Sixty-seven percent of all herpes zoster-related hospitalizations were of persons aged ⩾65 years, and only 5.2% were of persons <30 years of age. Examination of the percentage of all cases in each of these 2 age groups by year revealed modest and non-statistically significant (P > .05) variation (range, 63%–71% and 4.1%–7.0%, respectively).

Unlike primary varicella hospitalizations, there was a marked difference in the number of cases and crude rates of herpes zoster hospitalization by sex (table 3). Overall, 59.5% of patients were female, and females had 5.1 more cases per 100,000 population than did males (table 3). However, examination of agespecific rates by sex revealed no consistent differences between males and females in any age group. Most of the crude difference appears to be due to the larger female elderly population.

Also in contrast to primary varicella, crude herpes zoster hospitalization rates by race/ethnicity were similar for blacks and whites, but the rate for Hispanics was only one-half that of the other 2 groups (table 3). Examination by age group, however, revealed several notable differences. In all age groups <50 years of age, race/ethnicity-specific rates were higher for blacks and Hispanics than for whites. The relative risk compared with that of whites ranged from 2.6 to 6.2 for blacks and from 1.6 to 3.5 for Hispanics. In contrast, in all age groups ⩾65 years, rates were consistently higher for whites than either for blacks or for Hispanics. For these age groups, the relative risk for blacks and for Hispanics compared with that for whites ranged from 0.53 to 0.98 and from 0.01 to 1.11, respectively.

Temporal characteristics

Month of admission was grouped into four admission quarters: January–March, April–June, July–September, and October–December. Overall, 73.2% of varicella hospitalizations were in the 6-month period from January through June, the winter and spring quarters. This seasonality was evident in each year of the study period and all age groups. In contrast, seasonality was not observed in herpes zoster hospitalizations (table 3). This lack of seasonality persisted across all years of the study period and across all age groups, including persons <30 years old, among whom 51.5% of cases occurred in the first half of the year.

The number of primary varicella hospitalizations increased substantially and, with the exception of 2 years, progressively between 1986 and 1995 (figure 1). Overall, there was a 31% increase in the number of primary varicella hospitalizations from the first 5-year interval, 1986–1990, to the second, 1991–1995. Similarly, the number of herpes zoster hospitalizations increased annually between 1986 and 1991 but then leveled off thereafter. Overall, there was only an 8.8% increase in the number of herpes zoster hospitalizations from the first 5-year interval to the second.

Underlying conditions

Of the 1341 primary varicella hospital discharges, a total of 230 (17.2%) had an underlying condition recorded (table 4). The percentage of primary varicella hospitalizations with at least one underlying condition increased over time, from 12% (n = 70) of all primary varicella hospitalizations between 1986 and 1990 to 21% (n = 162) between 1991 and 1995. Most of this increase was due to human immunodeficiency virus (HIV), which increased from 2.9% (n = 17) to 8.4% (n = 64) during the two 5-year time periods, and pregnancy, which increased from 1.5% (n = 8) to 3.6% (n = 28). These two conditions accounted for 71% (65/91) of the increase in the absolute number of persons with an underlying condition. However, only one-half of the increase in primary varicella hospitalizations from the first 5 years to the second can be readily explained by increases in persons with underlying conditions.

Table 4
View larger version:
Table 4

Underlying conditions and outcome among persons hospitalized with primary varicella and with herpes zoster, Connecticut, 1986–1995.

Of the 5274 hospital discharges with herpes zoster, 31.4% were found to include at least one underlying condition (table 4). Among the 2067 underlying conditions for these 1657 persons, malignancies, including leukemia, accounted for 65.7% of the underlying conditions, and HIV infection accounted for 20.5%. The percentage of herpes zoster hospitalizations with at least one underlying condition increased over the study period, from 29.1% between 1986 and 1990 to 33.6% between 1991 and 1995. This increase can be entirely explained by increasing HIV prevalence, which occurred in 4.6% of all cases in the first 5 years and 11.1% of all cases in the second. We also observed an inverse association of underlying conditions with age. Among persons aged <50 years, 60.5% had an underlying condition, whereas among those aged 50–64 years, 35.9% had an underlying condition. The percentage further decreased to 26.4% among those aged 65–79 years and to 16.1% among those aged ⩾80 years.

Complications

A total of 546 (40.7%) primary varicella hospitalizations had at least one coded varicella-specific complication, and an additional 80 (6.0%) had at least one possible varicella-related bacterial infection. Overall, there were 733 directly coded varicella-related complications among the 626 primary varicella hospitalizations.

Varicella hemorrhagic pneumonitis was the most commonly coded specific complication, occurring in 16.6% (n = 223) of all primary varicella hospitalizations. Hemorrhagic pneumonitis was particularly a complication in adults aged ⩾20 years (32.7% vs. 4.4%; P < .001). A total of 185 bacterial skin-related complications, including impetigo, cellulitis, and/or septicemia, occurred in 12.7% (n = 170) of all primary varicella hospitalizations. Examination of each complication by year revealed no difference in the annual frequency, except hemorrhagic pneumonitis. Over the 10-year period, the frequency of varicella hemorrhagic pneumonitis increased from 9.2% (n = 8) of all cases in 1986 to peaks of 20.9% (n = 37) in 1993 and 18.5% (n = 27) in 1995.

There were a total of 25 acute hospital-based deaths related to primary varicella, averaging 2.5 deaths per year, and 1.9 deaths for every 100 hospitalizations. Overall, 68.0% (n = 17) of those who died were male and 40% (n = 10) were <20 years of age. Persons who died were more likely than all hospitalized persons to have an underlying condition (48% vs. 17%), a varicella-specific complication (48% vs. 17%), or a bacterial skinrelated infection (24% vs. 13%). Seventeen deaths were in patients with underlying illnesses, including HIV (n = 3), leukemia (n = 5), other malignancy (n = 6), other blood dyscrasia (n = 1), and congenital immunodeficiency (n = 2).

A total of 1985 herpes zoster hospitalizations (37.6%) had 2079 coded herpes zoster-specific complications. Nervous system complications and ophthalmic complications accounted for 57% and 21% of these complications, respectively. Two hundred seventy-seven persons (5.3%) died during herpes zoster-related hospitalization, resulting in an average rate of 27.7 deaths per year. At least one underlying condition was present in 52% of deaths. Underlying conditions included leukemia (n = 61), other malignancies (n = 135), HIV (n = 24), other blood dyscrasias (n = 10), and immunodeficiencies of childhood (n = 13). The risk of death during hospitalization was higher in persons with underlying conditions than in those without (8.7% vs. 3.7%;P < .05).

Hospital charge

The average unadjusted annual charge per primary varicella hospitalization nearly quadrupled over the 10-year period, from $3240 to $12,819 (figure 2). Higher charges were associated with older age, underlying conditions, and varicella-specific complications. The total charges for 1995, the year before vaccination became widely available, were slightly >$ 1.9 million.

The mean unadjusted charge per herpes zoster hospitalization for the study period was $12,834, which is $5037 more than the mean unadjusted charge for primary varicella hospitalizations. Furthermore, the mean unadjusted charge per herpes zoster hospitalization increased 2.2-fold between 1986 and 1995, from $7159 to $15,583 (figure 2). This increase in charge occurred despite a decrease in the mean length of stay from 11.6 days to 9.3 days over the study period. Mean unadjusted charges per herpes zoster hospitalization were higher in persons with underlying conditions than those without ($16,587 vs. $11,115) but did not vary in any consistent pattern by age. In 1995, the total unadjusted charge for all herpes zoster hospitalizations was $8,182,235, 4.3 times higher than the $1,922,913 in total unadjusted charges for hospitalization with primary varicella during the same year.

Previous SectionNext Section

Discussion

This study documents that there were marked differences in the epidemiology of hospitalizations with primary varicella and with herpes zoster in the prevaccine era. Of particular note, the number of hospitalizations with herpes zoster was nearly 4 times that with primary varicella and there were >10-fold the number of deaths in any given year. In any future evaluations of the potential cost-effectiveness of universal varicella vaccination, it will be critical to consider the additional impact of vaccination on medical costs due to herpes zoster as well as to varicella. In addition, the study showed that analysis of hospital discharge data may be an efficient, feasible, and potentially invaluable means to monitor the impact of varicella vaccination on the occurrence of severe disease due to both primary varicella and herpes zoster.

The increase in primary varicella hospitalizations before vaccine licensing has not been previously described. Although much of the increase in hospitalizations can be attributed to an increase in prevalence of persons with underlying conditions that may increase the severity of varicella or its consequences, in particular HIV infection, this accounts for no more than 50% of the increase in hospitalizations. Part of the increase could also result from the availability of intravenous acyclovir and recommendations for its use to treat persons at high risk of severe consequences of varicella [28–31]. Of interest, a similar increase beyond that expected from an increase in HIV was not seen in hospitalizations with herpes zoster, suggesting that the increase in varicella hospitalizations was not artifactual and that some other factor(s) might have caused it. Despite the increase over time, the average annual rate of primary varicella hospitalizations was consistent with rates based on national hospital discharge samples [32, 33].

Blacks and Hispanics had higher primary varicella hospitalization rates compared with whites, independent of age or year of hospitalization. In Connecticut, most blacks and Hispanics live in urban areas and in relatively crowded living conditions. Crowding may lead to exposure to varicella at younger ages and higher rates of primary varicella among younger children but would not be expected to result in higher rates of hospitalization in all age groups. Furthermore, HIV and other underlying conditions explain only a minority of cases in each race/ethnic group.

Whether there was differential access to primary outpatient health care between these generally poorer socioeconomic groups and the referent white group and a resulting effect on subsequent hospitalization rates could not be directly examined in this data set. The only other published population-based studies to report rates of varicella hospitalization by race/ethnicity were ones conducted of military recruits [34, 35]. In these studies, both blacks and persons of Puerto Rican origin also had higher rates of hospitalization than whites, but it was hypothesized that they had higher rates of susceptibility to varicella before entry into service. The military recruit studies were unable to examine race and ethnicity as factors for more than one age group. Regardless of the reasons for higher rates of hospitalization, these data suggest that it will be particularly important to achieve high levels of vaccine-induced immunity in these populations, especially in Connecticut.

Nearly 30% of the cases were in adults 20–34 years of age, the peak childbearing ages; ∼63% in this group were in women. Not surprisingly, these cases also exhibited the same seasonality seen in children. This is consistent with the observation that many adults are exposed to varicella by their children [36], in particular those caring for an ill child. Effective vaccination efforts among children should greatly diminish the risk of exposure among susceptible adults.

A total of 82 hospitalizations (6%) with primary varicella occurred among middle-aged and elderly adults (⩾50 years). Primary varicella in this age group has rarely been described. It is possible that these cases could represent misclassified cases of herpes zoster, considering that we found that hospitalizations with herpes zoster are 52 times more common than primary varicella in this age group. However, whereas herpes zoster shows no seasonality and its highest age-specific hospitalization rates occur among whites, the varicella cases in older adults that we detected showed a seasonal distribution, with 60% occurring in the first and second quarters of the year, and higher age-specific rates among race/ethnic minorities than whites. Thus, it appears that a substantial percentage may be actual cases of primary varicella acquired from contact with other primary cases of varicella, rather than misclassified cases of herpes zoster or cases of primary varicella acquired from exposure to other adults with herpes zoster. Correspondingly, many of these cases might be preventable with effective vaccination programs directed at children.

Over the 10-year study period, 82.8% of all hospitalized persons did not have an underlying condition. This finding highlights the need to vaccinate all infants and susceptible children if the cost of severe complications with varicella is to be prevented.

Few studies have examined trends in charges for hospitalizations related to varicella. In this study, average unadjusted charges increased at a rate that was much greater than the rate of inflation (296% increase in charges vs. 37.7% increase from inflation from 1986 to 1995) at a time when the rate of varicellarelated hospitalizations was also increasing. This suggests that cost savings from preventing cases of varicella may now be greater than in 1990, when a widely cited cost-benefit study of varicella vaccination in the United States was done [13]. The reasons for the dramatic increase in charges are not entirely clear. Although charges increased 4-fold, the average length of stay in days increased by only 17%, from 5.3 days in 1986 to 6.18 days in 1995. It is possible that the increase in charges may reflect in large part the actual rate at which health care costs were outstripping inflation.

Analysis of comprehensive population-based hospital discharge data appears to be a feasible, simple, and potentially sensitive way to monitor the impact of vaccination on the occurrence of the costly complications of primary varicella and their epidemiology. The epidemiology of varicella hospitalizations is distinct from that of herpes zoster, even in older age groups in which misclassification would be expected to be most likely. Additionally, the age, sex, and race/ethnic distributions of persons hospitalized with varicella appear to be fairly consistent over time. Furthermore, the brief declines in incidence of hospitalizations in 1986, 1991, and 1995 are consistent with periodic drops observed from studies of varicella incidence [32].

However, there are several major limitations to use of such data. Factors other than simple occurrence of complicated primary varicella may influence rates of hospitalization. In our study, 50% of the progressive increase could not be readily explained. In addition, since 1995, Connecticut has made a full transition to managed care. The impact of managed care on rates of hospitalization with varicella is unknown, although continued analysis of trends in hospitalization with herpes zoster, the incidence of which may not change much in the first vaccine-era years, may shed light on this. It will be important to better understand these dynamics to completely understand changes in the number of hospitalizations over time.

This study provides one of the first opportunities to compare the epidemiology, complications, and costs of herpes zoster with its parent disease, primary varicella. In the 10 years before the availability of varicella vaccine, there were nearly 4 times the number of hospitalizations with zoster compared with varicella and >4 times the total hospital costs. Only one other published study has presented data comparing overall impact of herpes zoster and primary varicella hospitalizations. In a recent review of 6 years of hospital discharge data in two states in Australia, zoster accounted for 6-fold more hospital bed-days per year than did varicella (7300 vs. 1200 days) [23].

Unlike with primary varicella, hospitalization rates for herpes zoster in the prevaccine era were stable except for the dynamic of the HIV epidemic. This is consistent with information obtained from CHIME—that there was little change in the way hospitalization data were obtained during the study period. It also suggests that the changes in hospitalizations with primary varicella over time were not artifactual.

This study confirms previous findings of a very strong association of incidence of herpes zoster with increasing age [19, 21] and provides, for the first time, population-based age-, sex-, and race/ethnic group-specific hospitalization rates with herpes zoster. Of interest for future surveillance is that there were a substantial number and rate of hospitalizations with herpes zoster in persons <30 years of age. This may be a target group in which to initially monitor the impact of decreased circulation of varicella virus on the occurrence of zoster.

Although many more women than men were hospitalized with zoster, the finding that there were no differences in agespecific hospitalization rates by sex is consistent with those of population-based studies of outpatient zoster, which have also shown no differences in age-specific rates [19, 21].

The finding that blacks and Hispanics <50 years of age are more likely to be hospitalized with zoster than are whites has not previously been noted. The reasons for this are not clear and are similar to our findings that blacks and Hispanics are more likely to be hospitalized with varicella than are whites. Higher hospitalization rates in these groups persisted even when persons with underlying conditions were eliminated from the analysis. Of further interest, unlike our findings for hospitalizations with primary varicella, the differences in race/ethnicityspecific zoster hospitalization rates were reversed in persons >65 years, an age group in which whites were more likely to be hospitalized than blacks or Hispanics. This difference persisted when persons with underlying conditions were eliminated from the analysis. The reasons for this shift in race/ethnicity-specific hospitalization rates with age are unknown and merit further study.

The lack of seasonality with zoster hospitalizations was remarkable and was observed across all years and all age groups. This was in contrast to documented seasonality of hospitalizations with primary varicella and with the finding of a study of zoster based in clinical practice [37], but it was in agreement with other studies of zoster [38–40].

Persons with underlying conditions such as HIV, leukemia, and other malignancies are at a greater risk of latent varicella virus reactivation and more serious complications. A higher percentage of hospitalized persons with herpes zoster than with varicella had underlying conditions (31.4% vs. 17%), although in both cases, most hospitalizations were in persons with no identified underlying conditions. Of interest, the percentage of hospitalizations in persons with underlying conditions steadily decreased with age, falling to 16% in the ⩾80-year age group. This is consistent with the observations of others that most of the severe morbidity from zoster results from age-related waning immunity rather than the increase in underlying conditions, such as malignancy, that occurs with age [21]. Therefore, if most severe morbidity from herpes zoster is to be prevented, either vaccination of children will have to be effective in minimizing the extent of initial latent infection or it will be necessary to boost waning immunity in older age. The planned clinical trial to test the efficacy of varicella vaccine in boosting immunity in older persons in the Veterans Administration system will attempt to determine whether the latter strategy works [41].

The financial cost of hospitalization with herpes zoster has not been previously described. Of interest, the mean charge and mean length of stay of herpes zoster are substantially higher than the mean charge and mean length of stay of hospitalization with primary varicella. In addition, the percentage of hospitalizations that ended in death was >2 times higher with zoster than with varicella (5.3% vs. 1.9%). Thus, even among persons needing hospitalization, zoster in the prevaccine era appears to have more morbidity than varicella.

Given that widespread vaccination of children has begun and that a clinical trial of varicella vaccine in adults to prevent herpes zoster will soon begin, it is likely that the epidemiology of herpes zoster will begin to change. With the current severity and costs of hospitalizations with herpes zoster and the uncertainty of the impact of widespread childhood vaccination on zoster incidence, it becomes important to establish surveillance and monitor the population-based impact of vaccination. In this context, analysis of comprehensive population-based hospital discharge data appears to be a feasible, simple, and potentially sensitive way to monitor the impact of vaccination on the occurrence of the costly complications of herpes zoster and their epidemiology. The epidemiology of hospitalization with herpes zoster is distinct from that of varicella. Additionally, the age, sex, and race/ethnic distribution of persons hospitalized with zoster appears to be fairly consistent over time. Furthermore, the incidence of hospitalizations in the past 10 years has been constant, modified only by the dynamics of the HIV epidemic.

This analysis of hospital discharge data has several limitations. Most importantly, raw data from CHIME were analyzed without any attempt to validate the discharge diagnosis codes, determine whether primary varicella or herpes zoster was the main reason for initial hospitalization, or determine the percentage of hospital costs that might be attributed to primary varicella or herpes zoster. Although this was the most practical approach to data analysis and future potential use of such data for surveillance, it could result in several types of misclassification. We are not aware of any studies that attempt to examine the sensitivity and specificity of a hospital discharge diagnosis of herpes zoster. However, a recent health maintenance organization-based outpatient study of herpes zoster in which case finding was based on diagnostic codes, including ICD-9 codes, found an 89%–96% positive predictive value of a diagnostic code of herpes zoster [22]. In addition, a recent validation study of varicella-related hospital discharge data for persons aged ⩾13 years in New York found that the positive predictive value of varicella listed as a discharge diagnosis was 87% as a measurement of varicella-related hospitalization (B. Wallace, New York State Department of Health, personal communication). Thus, false-positive misclassification appears unlikely to have a significant impact on our results. We have no measures of the extent to which primary varicella or herpes zoster hospitalizations may have been falsely miscoded as other events and, thus, the extent to which our data underestimate the magnitude of varicella-related hospitalizations. Studies are needed to provide a full context for interpreting the magnitude of the problem described by analysis of hospital discharge data.

Because validation studies were not done, it is possible that primary varicella or herpes zoster could have been coincidental during hospitalization for an unrelated admission diagnosis. Thus, our study may have overestimated the overall burden of disease, especially among persons with underlying conditions who have longer hospitalizations. Nonetheless, there would have been a substantial cost in case management and in hospital epidemiologic considerations for any inpatient case of incidental primary varicella or herpes zoster, especially in a person who was immunosuppressed [32, 42]. In the absence of doing chart reviews, similar limitations apply to the data on underlying conditions and complications.

Changes in methods used for data collection by CHIME could also result in artifactual trends in hospitalizations. Fortunately, there was little change in methods used by CHIME during this time, minimizing the potential for methodologic changes to create surveillance artifact [24] (CHIME, personal communication).

These data were collected from Connecticut hospitals, and, without similar studies in other states, it is difficult to know whether our results are generalizable to other states or the nation as a whole. The fact that Connecticut is fairly heterogeneous in race/ethnic and socioeconomic makeup and has both densely and less-densely populated urban and suburban areas should enable these data to be generalizable to much of the country. In addition, the overall hospitalization rates for primary varicella and herpes zoster are fairly consistent with those obtained from the national hospital discharge database [15, 32]. However, the findings that Hispanic and black race/ethnic minorities and childbearing women have higher rates of hospitalization and that cases of primary varicella appear to be occurring among the middle-aged and elderly need confirmation. Similar studies are needed in other states to confirm our findings if this type of surveillance is to be used to monitor the impact of varicella vaccination on the epidemiology of primary varicella and herpes zoster in the nation as a whole.

Previous SectionNext Section

Acknowledgments

We thank Jane Seward, Karen Galil, and Paul Kilgore of the Centers for Disease Control and Prevention for their helpful review and comments.

Previous SectionNext Section

Footnotes

  • Grant support: Centers for Disease Control and Prevention (Childhood Immunization Program: H23/CCH104484-08).

  • Received February 8, 1999.
  • Revision received January 24, 2000.
Previous Section
 

References

  1. 1.↵
    Committee on Infectious Diseases, American Academy of Pediatrics. Recommendations for the use of live attenuated varicella vaccine. Pediatrics 1995;95:791-6.
    Abstract/FREE Full Text
  2. 2.↵
    Centers for Disease Control and Prevention. Prevention of varicella: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Morb Mortal Wkly Rep 1996;45(RR-11):1-36.
    Medline
  3. 3.↵
    1. Plotkin SA
    . Varicella vaccine [commentary]. Pediatrics 1996;97:251-3.
    Abstract/FREE Full Text
  4. 4.
    1. Newman RD,
    2. Taylor JA
    . Reactions of pediatricians to the recommendation for universal varicella vaccination. Arch Pediatr Adolesc Med 1998;152:792-6.
    Abstract/FREE Full Text
  5. 5.↵
    1. MacFarlane LL,
    2. Sanders ML,
    3. Carek PJ
    . Concerns regarding universal varicella immunization: time will tell. Arch Fam Med 1997;6:537-41.
    Abstract/FREE Full Text
  6. 6.↵
    1. Hardy IB,
    2. Gershon A,
    3. Steinberg S,
    4. et al
    . The incidence of zoster after immunization with live attenuated varicella vaccine: a study in children with leukemia. N Engl J Med 1991;325:1545-50.
    MedlineWeb of Science
  7. 7.
    1. LaRussa P,
    2. Steinberg S,
    3. Gershon AA
    . Varicella vaccine for immunocompromised children: results of collaborative studies in the United States and Canada. J Infect Dis 1996;174 Suppl 3:S320-3.
    Medline
  8. 8.
    1. Gershon A,
    2. LaRussa P,
    3. Steinberg S,
    4. Mervish N,
    5. Lo SH,
    6. Meier P
    . The protective effect of immunologic boosting against zoster: an analysis in leukemic children who were vaccinated against chickenpox. J Infect Dis 1996;173:450-3.
    Abstract/FREE Full Text
  9. 9.↵
    1. Boyer M,
    2. Tete MJ,
    3. Guest G,
    4. Gagnadoux MF,
    5. Rouzioux C
    . Varicella and zoster in children after kidney transplantation: long-term results of vaccination. Pediatrics 1997;99:35-9.
    Abstract/FREE Full Text
  10. 10.↵
    1. Watson BM,
    2. Piercy SA,
    3. Plotkin SA,
    4. Starr SE
    . Modified chickenpox in children immunized with the Oka/Merck varicella vaccine. Pediatrics 1993;91:17-22.
    Abstract/FREE Full Text
  11. 11.↵
    1. Levin MJ
    . Can herpes zoster be prevented? Eur J Clin Microbiol Infect Dis 1996;15:1-3.
    CrossRefMedlineWeb of Science
  12. 12.↵
    1. Lawrence R,
    2. Gershon AA,
    3. Holzman R,
    4. Steinberg SP
    . The risk of zoster after varicella vaccination in children with leukemia. N Engl J Med 1988;318:543-8.
    MedlineWeb of Science
  13. 13.↵
    1. Lieu TA,
    2. Cochi SL,
    3. Black SB,
    4. et al
    . Cost effectiveness of a routine varicella vaccination for US children. JAMA 1994;271:375-81.
    Abstract/FREE Full Text
  14. 14.
    1. Huse DM,
    2. Meissner Lacey MJ,
    3. Oster G
    . Childhood vaccination against chickenpox: an analysis of benefits and costs. J Pediatr 1994;124:869-74.
    CrossRefMedlineWeb of Science
  15. 15.↵
    1. Beutels P,
    2. Clara R,
    3. Tarmani G,
    4. Van Doorslaer E,
    5. Van Damme P
    . Costs and benefits of routine varicella vaccination in German children. J Infect Dis 1996;174 Suppl 3:S335-41.
    Medline
  16. 16.
    1. Scuffham PA,
    2. Lowin AV,
    3. Burgess MA
    . The cost-effectiveness of varicella vaccine programs for Australia. Vaccine 1999;18:407-15.
    CrossRefMedlineWeb of Science
  17. 17.↵
    1. Scuffham P,
    2. Devlin N,
    3. Eberhart-Phillips J,
    4. Wilson-Salt R
    . The cost-effectiveness of introducing a varicella vaccine to the New Zealand immunisation schedule. Soc Sci Med 1999;49:763-79.
    CrossRefMedlineWeb of Science
  18. 18.↵
    1. Burgoon CF,
    2. Burgoon JS,
    3. Baldridge GD
    . The natural history of herpes zoster. JAMA 1957;164:265-9.
    Abstract/FREE Full Text
  19. 19.↵
    1. Ragozzino MW,
    2. Melton LJ,
    3. Kurland LT,
    4. Chu CP,
    5. Perry HO
    . Population based study of herpes zoster and its sequelae. Medicine 1982;61:310-6.
    Medline
  20. 20.
    1. Guess HA,
    2. Broughton DD,
    3. Melton LJ,
    4. Kurland LT
    . Epidemiology of herpes zoster in children and adolescents: a population study. Pediatrics 1985;76:512-7.
    Abstract/FREE Full Text
  21. 21.↵
    1. Donahue JG,
    2. Choo PW,
    3. Manson JE,
    4. Platt R
    . The incidence of herpes zoster. Arch Intern Med 1995;155:1605-9.
    Abstract/FREE Full Text
  22. 22.↵
    1. Miller E,
    2. Marshall R,
    3. Verdien J
    . Epidemiology, outcome and control of varicella-zoster infection. Rev Med Microbiol 1993;4:222-30.
  23. 23.↵
    1. Chant KG,
    2. Sullivan EA,
    3. Burgess MA,
    4. et al
    . Varicella-zoster virus infection in Australia. Aust N Z J Public Health 1998;22:413-8.
    CrossRefMedlineWeb of Science
  24. 24.↵
    1. Morris M
    . CHIME data usage: CHIME program services and benefits for its members. Wallingford, CT: Connecticut Hospital Research and Education Foundation; 1990.
  25. 25.↵
    Public Health Service and Health Care Financing Administration. International classification of diseases, 9th revision, clinical modification. Salt Lake City: Med Index Publications; 1992.
  26. 26.↵
    Bureau of the Census, United States Department of Commerce. 1990 census of the population: general population characteristics, Connecticut Washington, DC: US Government Printing Office; 1992. document CP-1-8.
  27. 27.↵
    Bureau of the Census. Intercensual estimates of the population of states by age, sex and race: 1970–96. Washington, DC: Bureau of the Census; 1997.
  28. 28.↵
    1. Balfour HH Jr.
    . Intravenous acyclovir therapy for varicella in immunosuppressed children. J Pediatr 1984;104:134-6.
    MedlineWeb of Science
  29. 29.
    1. Nyerges G,
    2. Meszner Z,
    3. Gyarmati E,
    4. Kerpel-Fronius S
    . Acylcovir prevents dissemination of varicella in immunocompromised children. J Infect Dis 1988;157:309-13.
    Abstract/FREE Full Text
  30. 30.
    1. Balfour HH Jr.,
    2. Rotbart HA,
    3. Feldman S,
    4. et al
    . Acyclovir treatment of varicella in otherwise healthy adolescents. J Pediatr 1992;120:627-33.
    CrossRefMedlineWeb of Science
  31. 31.↵
    1. Wallace MR,
    2. Bowler WA,
    3. Murray NB,
    4. Brodine SK,
    5. Oldfield EC III.
    . Treatment of adult varicella with oral acyclovir. Ann Intern Med 1992;117:358-63.
    Abstract/FREE Full Text
  32. 32.↵
    1. Wharton M
    . The epidemiology of varicella-zoster virus infections. Infect Dis Clin North Am 1996;10:571-81.
    CrossRefMedlineWeb of Science
  33. 33.↵
    1. Guess HA,
    2. Broughton DD,
    3. Melton LJ,
    4. Kurland LT
    . Population based studies of varicella complications. Pediatrics 1986;78(Suppl):723-7.
    Abstract/FREE Full Text
  34. 34.↵
    1. Herrin VE,
    2. Gray GC
    . Decreasing rate of hospitalization for varicella among young adults. J Infect Dis 1996;174:835-8.
    Abstract/FREE Full Text
  35. 35.↵
    1. Gray GC,
    2. Palinkas LA,
    3. Kelley PW
    . Increasing incidence of varicella hospitalizations in the United States Army and Navy personnel: are today's teenagers becoming more susceptible? Should recruits be vaccinated? Pediatrics 1990;86:867-73.
    Abstract/FREE Full Text
  36. 36.↵
    Centers for Disease Control and Prevention. Varicella-related deaths among adults—United States, 1993. MMWR Morb Mortal Wkly Rep 1997;46:409-12.
    Medline
  37. 37.↵
    1. Hellgren L,
    2. Hersle K
    . A statistical and clinical study of herpes zoster. Gerontol Clin (Basel) 1966;8:70-6.
    Medline
  38. 38.↵
    1. McGregor RM
    . Herpes zoster, chickenpox, and cancer in general practice. Br Med J 1957;1:84-7.
    FREE Full Text
  39. 39.
    1. Kurland LT
    . Descriptive epidemiology of selected neurological and myopathic disorders with particular reference to a survey in Rochester, Minnesota. J Chronic Dis 1958;8:378-418.
    CrossRefMedline
  40. 40.↵
    1. Hope-Simpson RE
    . The nature of herpes zoster: a long-term study and a new hypothesis. Proc R Soc Lond 1965;58:9-20.
  41. 41.↵
    1. Brunell PA
    . Possible role of varicella vaccine in preventing herpes zoster. Pediatr Infect Dis J 1999;18:842-3.
    CrossRefMedlineWeb of Science
  42. 42.↵
    1. Gray AM,
    2. Fenn P,
    3. Weinberg J,
    4. Miller E,
    5. McGuire A
    . An economic analysis of varicella vaccination for health care workers. Epidemiol Infect 1997;119:209-20.
    CrossRefMedline
| Table of Contents

This Article

  1. J Infect Dis. (2000) 181 (6): 1897-1905. doi: 10.1086/315492
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 205 (5)
  1. Journal of Infectious Diseases
  1. Alert me to new issues

Most