Transmission

Varicella is a highly infectious viral disease caused by the varicella zoster virus (VZV) that results from exposure to cases of varicella or herpes zoster. Herpes zoster cases represent a method for regular exposure and reintroduction of VZV into communities that otherwise may not be large enough to sustain endemic transmission of the virus. Herpes zoster is less transmissible than varicella. This may relate both to the limited number of lesions and to modes of transmission; the infection is most commonly localized and is not thought to involve the respiratory tract. Varicella can be transmitted by respiratory droplets, from skin lesions, by direct contact or possibly by aerosolization of virus from skin lesions and also, presumably from lesions in the mouth (enanthem). Studying transmission of varicella has been challenging due to the strong cell association of the virus. VZV has been consistently difficult to culture from the throat but easy to culture from skin lesions. Finally, infection of a pregnant woman in the first 2 trimesters of pregnancy may result in transplacental transmission of VZV resulting in a severe congenital infection in the fetus or newborn known as congenital (fetal) varicella syndrome (Enders et al., 1994).

Varicella transmission has been described from household studies where secondary attack rates among susceptible children following household exposure has ranged from 61% to 100% (Asano et al., 1977; Hope-Simpson, 1952; Ross, 1962). The highest estimates come from small groups of seronegative children involved in postexposure vaccine effectiveness studies in household settings. The lowest estimate is from observations in a general practice in England where children <15 years with a negative disease history were followed after exposure to a primary household case (Hope-Simpson, 1952). Hope-Simpson described transmissibility of varicella to be lower than measles (76%) but higher than mumps (32%). In a large study in the United States of the effect of gamma globulin on modifying varicella infections, Ross described a secondary attack rate of 87% among untreated children aged 6 months to 12 years with a negative disease history (Ross, 1962).

Compared with households, transmission is lower in community settings; the relatively few published estimates available come from varicella outbreak investigations that have included vaccinated as well as unvaccinated children. In these situations, due to continuing exposure, outbreaks in child care centers may ultimately achieve a cumulative attack rate among history and vaccine negative children as high as 80%–88% (Galil et al., 2002b; Izurieta et al., 1997) though, depending on the size of the child care center, this may take weeks or months of transmission. In elementary schools, cumulative attack rates among susceptible children tend to be lower (30%–54%) perhaps due to different mixing patterns resulting in lower risk of exposure (Dworkin et al., 2002). Yorke and London, in analyses of 30 to 35 years of monthly reported measles and varicella cases from New York City and Baltimore, suggested that varicella is only 35%–65% as infectious as measles in the community (Yorke and London, 1973).

Herpes zoster represents a method for reintroduction of VZV into the community in the absence of epidemic varicella. Patients with herpes zoster are less contagious than cases of varicella because their infection is localized and it is not thought to involve the respiratory tract. However, some studies have identified VZV antigen (by PCR) in throat samples of healthy persons with localized herpes zoster. There are few published studies on the transmissibility of herpes zoster probably because of the limited opportunities for exposure of susceptible persons to herpes zoster cases which occur predominantly, among older adults. However, Seiler reported that 11 (15.5%) of 71 susceptible children <15 years of age exposed to herpes zoster in household settings developed varicella (Seiler, 1949). This proportion increased to 17% if infants <1 year of age, who may be protected by maternally acquired antibodies, were excluded from the calculation. Young children with herpes zoster may be more contagious. In a daycare setting, a 3-year-old boy with herpes zoster transmitted in the secondary generation to ∼30% of susceptible children in the day care he attended and to both his susceptible siblings (Reigle and Cooperstock, 1985). The child was reported to “lift his shirt repeatedly to scratch or to show everyone his lesions.” As varicella disease declines in countries implementing universal childhood vaccination programs, transmission from herpes zoster may become more apparent.

Periodicity and seasonality

Worldwide, varicella is an endemic disease that exhibits a marked seasonal pattern in temperate climates and most temperate climates where this has been studied (Bramley and Jones 2000; Degeun et al., 1998; Lee, 1998; Seward et al., 2002; Tobias et al., 1998) (Fig. 40.1). The exception is Singapore where seasonality has not been described from surveillance data (Ooi et al., 1992). In temperate and tropical climates, the peak disease incidence is most commonly reported in the cooler, drier months during winter or spring. Periodicity with interepidemic cycles of 2–5 years is described from many countries (Bramley and Jones, 2000; Degeun et al., 1998; Seward et al., 2000; Tobias et al., 1998) while a time period as long as 15 years between major epidemics has been described from Singapore (Ooi et al., 1992).

Fig. 40.1

Notification of varicella in Scotland, 4-week periods, 1990–1998.

Congenital (fetal) varicella syndrome

This condition was first described in 1947 (Laforet and Lynch, 1947). It is characterized by cicatricial skin lesions and neurologic, eye and skeletal anomalies which may include limb paresis, hypoplasia of upper or lower extremities, chorioretinitis, cataracts and cortical atrophy. Based on a large, multicenter study where 1373 women with varicella and 366 with herpes zoster during pregnancy were prospectively studied, the risk of congenital (fetal) varicella syndrome was found to be approximately 0.4% if varicella infection was acquired in the first 12 weeks of pregnancy and 2.0% from 13–20 weeks (Enders et al., 1994). Maternal zoster was not associated with fetal abnormalities. Based on this study, Enders estimated that the rate of fetal varicella syndrome may be 1.6 per 100 000 births. Enders and Miller used these findings to estimate the number of cases of congenital varicella syndrome that may occur every year by applying a varicella age-specific incidence of 2/1000 population among women of child-bearing age to the number of births per year, taking into account the risk of infection occurring in the first 20 weeks of pregnancy. They calculated that, in the prevaccine era, there may have been 44 cases of congenital varicella syndrome every year in the United States, 8 in England and Wales and 9 in Germany (Enders and Miller, 2000). Incidence is likely to be higher in countries where VZV susceptibility is higher among adults though data are lacking.

Incidence and seroprevalence

Age and climate

Varicella is a highly contagious disease that occurs worldwide. The epidemiology of varicella varies between temperate and tropical climates. Though incompletely understood, these differences may relate to agent, host, environmental or a combination of these factors (Garnett et al., 1993; Lolekha et al., 2001; Mandal et al., 1998; Seward et al., 2000). Because climatic differences (as distinguished from an “island” effect reflecting reduced risk of exposure) are not observed for other highly contagious diseases such as measles, and VZV is known to be heat labile, a partial explanation may be that heat diminishes the ability of the virus to survive in the environment and thereby decreases transmission. In most temperate climates, >90% of persons are infected by 15 years of age with the highest incidence of disease occurring among children <10 years of age (Seward et al., 2000; Wharton, 1996). In tropical climates, cases are acquired at older ages with a higher proportion of cases and higher susceptibility among adults (Lee, 1998).

Varicella incidence data has been described mainly from developed countries where data are collected from notifiable disease reporting, surveys or studies based on medical record encounters. Prior to the national varicella vaccination program in the United States, total annual varicella incidence measured from national household survey data, averaged over a decade, was 15.0–16.0 cases per 1000 population (Guess et al., 1986; Seward et al., 1998). In other temperate climates, reported incidence using different methods of data collection have been generally lower than US rates varying from <5 to ∼13.0 cases per 1000 total population (Boelle and Hanslik, 2002; Bramley and Jones, 2000; Choo et al., 1995; Fairley and Miller, 1996). Higher rates, especially from a single year of data collection, may reflect an epidemic disease year (Chant et al., 1998) (Table 40.1).

Table 40.1

Varicella age-specific and total incidence, selected studies.

Infectiousness of diseases is reflected in age-specific incidence. As stated by Hope-Simpson following studies of the infectiousness of measles, varicella and mumps in the household “the more infectious the disease, the younger is the age at which an attack is likely to be received” (Hope-Simpson, 1952). Studies from earlier in the twentieth century in the US showed the highest varicella age-specific incidence was among children aged 6–7 years, in the first 2 years of school (Fales, 1928). To understand age-specific disease patterns, single years of age or small age groups should be studied; grouping ages 5–14 years, for example, may mask considerable variations in incidence within this broad age span. Age-specific incidence among children 5–9 years from US national data in the 1970s and 1980s was ∼90/1000 children (Guess et al., 1986; Wharton, 1996). However, by the 1990s, varicella was being acquired at earlier ages. Before introduction of varicella vaccine in the United States in 1995, varicella age-specific incidence had shifted to earlier ages with pre-school aged children (1–4 years) having higher age specific incidence than the 5–9-year-old age group (Seward, 1998) (Table 40.1). These data are consistent with findings in many countries including France, Italy, England and Wales, Scotland and Slovenia where the highest age-specific incidence in the 1990s has been reported among preschool-aged children <5 years (Bramley and Jones, 2000; Degeun et al., 1998; Fornaro et al., 1999; Ross and Fleming, 2000) with peak incidence in some studies as young as 1 or 2 years (Socan et al., 2001, Yawn et al., 1997). These changes in epidemiology are thought to reflect earlier exposure through attendance in child care (Wharton 1996; Yawn et al. 1997; Ross and Fleming, 2000)

Varicella seroprevalence data reflect age-specific disease incidence and in countries where incidence data are not available, VZV age-specific seroprevalence provide an excellent method for understanding the epidemiology of the disease. Comparing results from serosurveys should take into account improvements in sensitivity and specificity of laboratory methods over time. In addition, these studies represent a snapshot of the seroprevalence status in a community; seroprevalence will be higher if samples are obtained after a varicella epidemic. In temperate climates, >90% of adolescents or young adults are VZV seropositive (Fig. 40.2) (Kanra et al., 2002; Kilgore et al., 2003; Seward et al., 2000; Thiry et al., 2002). Slightly lower seroprevalence rates in adolescents (10–15 years) (82%) are reported from Italy whereas a study in South Africa reported a seroprevalence of 81% among adults 20–29 years (Gabutti et al., 2001; Schoub et al., 1985). In tropical climates, serological studies show higher susceptibility among adults reflecting a higher mean age of infection than in temperate climates (Fig. 40.2). In such countries, seroprevalence among adolescents or young adults has varied widely from less than 20% in St. Lucia (an island community in the West Indies) to >90% in tropical areas in Brazil and urban Calcutta, India (Barzaga et al., 1994; Garnett et al., 1993; Lee, 1998; Mandal et al., 1998; Ooi et al., 1992; Reis et al., 2003).

Fig. 40.2

VZV Seroprevalence by age in eight countries.

Hospitalization and deaths

Though often considered a benign childhood disease, varicella may result in serious complications and death. This health burden assumes greater importance as other infectious causes of morbidity and mortality such as polio, measles and H. influenzae are controlled through vaccination (Rawson et al., 2001). Assessing the severe health burden due to varicella in terms of economic and societal costs is important for vaccine policy decision making and to monitor the impact of vaccination programs. Data on population-based varicella mortality, case fatality and hospitalizations are rarely available from developing countries or countries with tropical climates where a higher proportion of cases occur among adults. Additionally, methodological differences in data collection as well as issues such as access to health care should be considered when comparing studies, especially across countries. Finally, calculating case fatality rates and risks of hospitalization for varicella cases is dependent on having accurate incidence data available.

In the United States, in the 5–8 years before licensure of varicella vaccine, varicella resulted in an average of 10 632 hospitalizations and 100 deaths per year (Galil et al., 2002a; Meyer et al., 2000); two-thirds of the hospitalizations and about half the deaths occurred in children. Although varicella is a more severe infection in immunocompromised persons, the majority of severe morbidity and mortality in developed countries occurs among healthy persons. In France (1990–1997), 70% of all varicella deaths and in the United States (1990–1994), 89% of varicella deaths among children and 75% of varicella deaths among adults occurred in persons without underlying high risk medical conditions (including HIV/AIDS, leukemia and other malignancies, other forms of blood dyscrasia and immune deficiencies) (Boelle and Hanslik, 2002; Meyer et al., 2000).

In the US, between 1970 and 1994, before the use of varicella vaccine, crude varicella mortality rates (examining varicella as the underlying cause of death) declined from 0.7 per million population in 1973 to 0.2 in 1986 and then increased to average 0.4 per million population from 1990–1994 (Meyer et al., 2000). Similar crude mortality rates are reported from Australia (0.3/million), France (0.35/million), England and Wales (0.5/million), Scotland (0.5/million) and Singapore (0–0.8/million) (Boelle and Hanslik, 2002; Bramley and Jones, 2000; Chant et al., 1998; Fairley and Miller, 1996; Lam et al., 1993; Rawson et al., 2001). Reflecting the high incidence of varicella in children, varicella mortality rates are highest among children especially those <1 year of age (1.1–3.6 deaths per million population) and lowest among adults (0.2–0.3 deaths per million population).

The risk of dying from varicella is measured by the case fatality rate (CFR). In the US, data from 1970–1994 show higher case fatality rates in infants <1 year and adults ≥20 years of age compared with children 1–9 years (Meyer et al., 2000). Even though CFRs among adults declined substantially from the 1970s to the 1990s, during 1990–1994, adults still had a 27 times higher risk of dying from varicella (CFR 21.3 per 100 000 cases) than children 1–4 years (0.8/100 000 cases). Similar CFRs are reported from France from 1990–97 (CFR 1.0/100 000 for children <15 years and 22.8 for persons ≥15 years) and England and Wales from 1988–1992 for persons <45 years (0.7 for 0–4 years, 1.4 for 5–14 years and 20 for persons 15–44 years of age) (Boelle and Hanslik, 2002; Fairley and Miller, 1996). Though extremely high CFRs (471 and 535) are reported for adults ≥65 years in England and Wales, and France respectively, in this age group, misclassification of varicella with herpes zoster is more likely to occur, as documented in US hospitalization and mortality data (Choo et al., 1995; Galil et al., 2002c).

Population-based mortality data are lacking from developing countries and from countries with tropical climates. Because of the older age of infection and case severity among adults, tropical countries may experience greater morbidity and mortality from varicella and its complications, including congenital varicella syndrome. The eradication of smallpox afforded an opportunity to study varicella in developing countries in the 1970s in more detail than has been possible since. When smallpox was still endemic, varicella was the rash illness most commonly confused with smallpox (Jezek et al., 1978a). For several years after smallpox eradication, heightened surveillance for febrile rash illnesses was conducted. In India, 862 155 varicella cases and 433 varicella deaths were reported from January to December 1976 for a CFR of 5.2 per 10 000 reported cases, 50 times higher than in developed countries (Jezek et al., 1978b). Reasons for the higher case fatality may include incomplete ascertainment of cases (reporting bias for more severe cases) compared with deaths, varicella epidemiology in India, and access to, and quality of, medical care. For 400 deaths where both age and sex were recorded, 80% occurred among adults >15 years and 71% were male, reflecting varicella epidemiology in tropical climates and perhaps a reporting bias and/or medical access differences for adults and for males compared with females (Jezek et al., 1978a).

Varicella hospitalizations represent severe morbidity or health burden due to varicella and its consequences, and the infection control burden due to varicella infections in hospitals. As pointed out by Wharton, “ascertaining the reason for hospitalization from hospital discharge diagnoses, in the absence of additional information from the clinical record, is difficult due to lack of standard procedures for ordering hospital discharge diagnoses” (Wharton, 1996). Some studies have validated hospital discharge codes (Choo et al., 1995). Thus, when comparing studies of hospitalizations, especially across countries, it is important to consider differences in methods that may result in higher or lower estimates including how hospitalizations were identified (from discharge codes or from reviewing medical records), whether the primary discharge or all discharge codes were searched for varicella, the validity of discharge codes and how hospital admission practices, billing and coding may change over time.

Estimates of annual hospitalizations for varicella in the US during the 1970s and 1980s ranged from approximately 4000 to 9000 depending on the dates of the study, the population studied, and study methods (Guess et al., 1986; Wharton, 1996). More recent estimates indicate annual hospitalizations attributable to varicella of 10 632 per year (Galil et al., 2002a) and almost 15 000 per year if all varicella hospitalizations are included (Ratner, 2002). Reflecting the range of these estimates, crude varicella hospitalization rates from the US, France, Australia, Scotland, and England and Wales have varied from approximately 2–6 per 100 000 population with more recent estimates attempting to describe attributable hospitalizations in the US varying from 3.1–4.1 per 100 000 population (Boelle and Hanslik, 2002; Bramley and Jones, 2000; Chant et al., 1998, Fairley and Miller, 1996; Galil et al., 2002a, Ratner, 2002). The majority of varicella hospitalizations occur among children (56%–67%) reflecting the fact that 90% of varicella cases occur among this age group.

For all ages combined, overall rates of hospital admission per 1000 varicella cases have ranged from 2.2 to 4.7 in national studies in the US and France (Boelle and Hanslik, 2002; Galil et al., 2002a; Ratner, 2002). The highest rate from France included principal and associated varicella hospitalizations. In the 1990s, among children <13 with varicella, a hospitalization rate of 5.5 per 1000 cases was reported from Minnesota (Yawn et al., 1997). The risk of hospitalization varies by age. The pattern of age-specific risks for hospitalizations is similar to that of age-specific case fatality rates with infants and adults having higher risks of hospitalization than young children. Except in France where the risk for hospitalization is higher in infants than in adults, studies consistently report the highest risk of hospitalization in adults, an increased risk in infants and the lowest risk in children 1–4 years or 5–9 years of age (Boelle and Hanslik, 2002; Fairley and Miller, 1996; Galil et al., 2002a; Guess et al., 1986; Wharton, 1996).

Studies of varicella hospitalizations in developing countries are sparse and population-based hospitalization rates are commonly not available for comparison with data from developed countries. However, similar to reported varicella mortality described above (Jezek et al., 1978a), studies of varicella hospitalizations in tropical climates also describe a high proportion of hospitalized cases among adults and also males, which may, in part, reflect hospital admission practices (Seward et al., 2000).

The HIV epidemic may also be expected to influence the epidemiology of varicella. In countries with high HIV prevalence, varicella may cause more severe morbidity and mortality however there are few population-based data examining these issues. A retrospective review of all children admitted to the only isolation facility in Durban, South Africa from 1986–1996, showed a decline in all disease mortality of 86% over the study period mainly attributed to a decline in measles deaths. However, between 1994 and 1996, 15% of varicella admissions and 75% of varicella deaths occurred in HIV co-infected children (Jeena et al., 1998).

Methodological issues

Studies of herpes zoster (HZ) incidence, morbidity and mortality have used various methodologies; the most reliable data come from prospective cohort studies. Herpes zoster incidence has been described from surveillance data where there is likely to be significant under-reporting of cases, retrospective cohort studies that use medical records thus missing cases that do not seek medical attention, and surveys that may encounter non responders. Thus, differences in the method of data collection must be taken into account when comparing HZ incidence data. In addition, because HZ incidence increases dramatically with age, comparing HZ incidence rates across studies needs to take this into account by standardizing rates to a reference population. Other methodological issues arise, especially when assessing disease burden due to HZ. Herpes zoster is more common among elderly adults, an age group that is likely to have chronic medical conditions that may cause lengthy hospitalizations or death; therefore, it is important to describe the attributable health burden or mortality from HZ. Otherwise, coincidental HZ infections that occur during hospitalization and that may be listed in hospital discharge codes and on death certificates may inflate health burden and mortality due to HZ. MacIntyre et al. (2003) reported that only 40% of 4718 hospitalizations that listed HZ in discharge codes had HZ as the primary diagnosis, and that these cases had significantly shorter hospitalization stay and were less likely to die (1%) than those with a secondary HZ diagnosis (6%).

Seasonality and clustering

Because HZ results from reactivation of latent VZV, the disease does not occur in epidemics and periodicity is not described. Most studies report no seasonal variations in the incidence of HZ (Brisson et al., 2001; Hope-Simpson, 1965; McGregor, 1957; Ragozzino et al., 1982); however several studies report seasonality with peak incidence in the summer and some authors speculate that this may be associated with ultraviolet radiation that peaks in the summer months (Glynn et al., 1990; Wilson, 1986; Zak-Prelich et al., 2000).

Although small clusters of HZ cases have been reported (Hope Simpson, 1965; Palmer et al., 1985; Schimpff et al., 1972), most authors report no clustering, or if clustering occurred they were considered to represent coincidental episodes of disease (Brisson et al., 2001; Donahue et al., 1995; Glynn et al., 1990; Hope-Simpson, 1965; Wilson, 1986). A few authors suggest that clusters may represent a clinical manifestation of exogenous reinfection, or stimulation of endogenous VZV reactivation, (Palmer et al., 1985; Schimpff et al., 1972).

Secular trends

Several studies have suggested that HZ rates are increasing. In a study in the United States spanning 15 years from 1945–1959, Ragozzino et al. (1982) reported an increase in the annual age standardized (1970 US white population) incidence from 1.1 to 1.5 per 1000 person years. In his study in the 1990s, Donahue et al. (1995) reported that once the HZ incidence rates he observed were standardized to the 1970 US white population, they increased to 2.9 per 1000 person–years, more than double that reported by Ragazzino. The increase was not associated with aging of the population or with an increase in immunosuppressed individuals. In addition, from 1979 to 1997, increases in HZ incidence of 35% and 21% have been described in Canada and England, respectively (Brisson et al., 2001).

Dermatomal distribution

Most studies report no laterality to the HZ rash with both the right and left sides affected equally. The rash affects mainly one dermatome with the thoracic, cervical, lumbar, and cranial dermatomes accounting for over 90% of the dermatomes affected (di Luzio Paparatti et al., 1999; Guess et al., 1985; Hope-Simpson, 1965; McGregor, 1957; Petursson et al., 1998; Ragozzino et al., 1982). Hope-Simpson noted that the HZ rash distribution correlates with the distribution of the varicella rash that follows a centripetal distribution. The rash affects mostly the trunk (thoracic 50%), and face (cranial 13%) with similar percentages for the cervical (14%) and lumbar (13%) regions, and spares the limbs. Similar findings were reported for HZ rashes in children (Guess et al., 1985; Takayama et al., 2000).

Incidence

Age

It is estimated that about 10%–30% of people will get HZ during their lifetime, resulting in about 300 000–900 000 cases of HZ in the US, and 260 000 cases in the UK each year (Brisson et al., 2001; Schmader, 2001). The incidence of HZ varies markedly in published studies. The total population incidence of HZ, unadjusted for age differences between populations, ranges from 1.3 to 4.8 per 1000 person–years (Table 40.3) (Chidiac et al., 2001; McGregor, 1957; Ragozzino et al., 1982). The differences in the rates observed across studies may be due to differences in the methods used for ascertainment of cases, case definitions, determination of population at risk, health systems capturing cases, and age distribution of population studied (that may be skewed or not reflective of the overall population) rather than to true population differences. In a prospective cohort study of all patients in his general practice in Cirencester, UK from 1947–1962, Hope-Simpson reported an annual incidence rate of 3.4 per 1000 person–years (Hope-Simpson, 1965). Much lower incidence rates were found in a community-based study in the US in which cases were identified retrospectively from medical records in a central diagnosis system that included all providers in the community (Ragozzino et al., 1982). The denominator included the population of Rochester, Minnesota, obtained from census data. Other studies used well defined populations for denominators such as enrollees in a health plan in Massachusetts (Donahue et al., 1995), attendants at general practices in Scotland (McGregor, 1957), England, (Hope-Simpson, 1965), and Italy (di Luzio Paparatti et al., 1999). The study in France (Chidiac et al., 2001) was a prospective study that did not include cases from pediatricians or from residents in institutions, and used the population from the census for the denominator. Therefore, it is important to adjust rates of studies that report age-specific incidence to a standard population for comparison purposes. The differences in incidence rates for two studies (Donahue et al., 1995; Hope-Simpson, 1965) narrowed from 2.1 and 3.4 per 1000 person–years respectively to 3.2 and 3.3 per 1000 person–years as a result of adjusting the crude rated to the 2000 US population, highlighting the need for such adjustments before comparing rates between studies.

Table 40.3

Herpes zoster age-specific and total incidence, selected studies.

The majority of studies report that a small proportion of persons infected with HZ may experience a second or a third episode. Hope-Simpson (1965) reported that the rate of second or more infections in persons previously infected is similar to that of first infection in the general population with 8 (4.1%) of 192 cases reporting a second episode, and one person a third episode, resulting in a rate of 3.1 per 1000 person–years. These findings were supported by other studies with recurrence ranging from 1.7%–5.2% (Donahue et al., 1995; Ragozzino et al., 1982; Wilson, 1982). A much higher recurrence rate of 45% was reported in one English study in persons with a first infection occurring over the age of 45 years (Edmunds et al., 2001).

All published studies report that the incidence of HZ increases with increasing age (Brisson et al., 2001; Chidiac et al., 2001; Donahue et al., 1995; McGregor, 1957; Hope-Simpson, 1965; Ragozzino et al., 1982; Opstelten et al., 2002) (Table 40.3). Most studies report an approximately 10 fold increase in risk with increasing age; Hope-Simpson reported incidence increasing from 0.74 among children less than 10 years of age to 10.1 per 1000 person–years among individuals aged 80–89 years. This marked increase in incidence with age results in up to 50% of people who live to 85 years acquiring HZ (Brisson et al., 2001; Schmader, 2001).

Because children may not have acquired varicella and therefore may not yet be at risk for reactivation of VZV, the incidence of HZ also depends on varicella epidemiology. Therefore, comparison of HZ incidence rates in children need to take into account both the age distribution of children in the studies and the age-specific incidence of varicella which may vary across populations. Hope-Simpson, 40 years ago, speculated that the increase in HZ incidence in persons 0–20 years reflected the increase in the number of children infected with VZV, placing them at risk for HZ. He suggested that the incidence observed in the third decade, by which time almost everyone in temperate climates is latently infected with VZV, reflects the incidence in a maximally infected population. Studies of HZ in children report annual incidence rates ranging from 0.25 to 1.15 per 1000 population for children <10 years and from 0.43 to 1.60 for those <20 years (Chidiac et al., 2001; Guess et al., 1985; Hope-Simpson, 1965; Petursson et al., 1998) (Table 40.3). Other studies used different age groups such as <5, and 5–14 years, or <14 years making comparison between studies difficult (Brisson et al., 2001; Donahue et al., 1995; McGregor, 1957; Ragozzino et al., 1982). Guess et al. (1985) in Rochester, Minnesota, during the period 1960–1981, reported that the incidence of HZ increased with age from 0.20 cases per 1000 person–years in children less than 5 years of age to 0.63 cases per 1000 person–years in those 15 to 19 years of age. In contrast, Petursson et al. (1998) in Iceland, during 1990–1996, reported rates that ranged from 0.80 per 1000 person-years in children less than 5 years to 1.80 in those 15–19 years of age (Table 40.3). These four fold differences in rates of HZ in children in narrow (5 year) age groupings are more likely to signify differences in study methods including completeness of ascertainment than differences between populations (Guess et al., 1985; Petursson et al., 1998). Further studies to understand such differences are needed. Maternal varicella during pregnancy and varicella or variable exposure during the first few months of life are associated with HZ in childhood.

Complications

Herpes zoster may lead to complications such as persistent pain (postherpetic neuralgia), bacterial infection of the lesions, pneumonia, encephalitis, and hemorrhagic complications. A number of risk factors have been found to be associated with complications, the most important of which are older age and immunosuppression. Persons >64 years have about eight times the risk of complications compared to those <25 years (Galil et al., 1997). Other factors include trigeminal distribution of the HZ rash, involvement of more than one dermatome (di Luzio Paparatti et al., 1999; Galil et al., 1997), severe pain, rash or prodromal symptoms at HZ presentation (Choo et al., 1997; Dworkin et al., 2001; Nagasako et al., 2002; Whitley et al., 1998), persisting abnormal sensations in the affected dermatome (Decroix et al., 2000) and scarring, presumed to be a consequence of rash severity (Battcock et al., 1990; Bowsher, 1999). Patients who presented with severe or incapacitating pain and a large number of lesions were 18 times less likely to achieve resolution of both acute neuritis and HZ-associated pain (Whitley et al., 1999).

Postherpetic neuralgia (PHN), a chronic pain syndrome, is the most common complication of HZ (Johnson, 2002; Lojeski and Stevens, 2000; Ragozzino et al., 1982; Stankus et al., 2000). The variability in the intensity and duration of the pain has made it complicated for researchers to agree on a standard definition making it difficult to compare results across studies. Studies have variously defined onset of PHN at rash onset or rash resolution. Study end points vary from 1, 3, 6 and 12 months after rash onset or resolution. At one month past rash onset, reported proportions of persons with PHN range from 6.5% to 45% (Haanpaa et al., 2000; Opstelten et al., 2002), at 3 months, the range is from 7.2% to 25% (Haanpaa et al., 2000; Helgason et al., 2000), and at 12 months, 3.4% to 10% (Bowsher, 1999; Helgason et al., 2000). The varying proportions in PHN at the different time periods may be affected by several factors including age and percent of immunosuppressed in the population under study.

The risk of PHN increases with increasing age although the magnitude of the effect varies across studies. PHN is more common among adults older than 50 years (Bowsher, 1999; Decroix et al., 2000; di Luzio Paparatti et al., 1999; Dworkin et al., 2001). It is estimated that about 27–68% of HZ cases over 60 years of age experience PHN (Dworkin et al., 2001; Kurokawa et al., 2002). In an Iceland study, older age was a significant and independent predictor of PHN; persons 55–74 years were 4.2 times more likely to develop PHN one month after rash onset compared to those younger than 55 years; while those over 75 years of age were 10.7 more likely to develop PHN (Opstelten et al., 2002). An even stronger association with older age was reported in a U.S. study where age ≥50 years compared to age <50 years was associated with a 14.7-fold higher prevalence of PHN at 1 month and a 27.4-fold higher prevalence at 2 months after developing HZ (Choo et al., 1997). Finally, in Singapore, Goh and Khoo (1997) reported that 20% of patients older than 50 years suffered from PHN compared to 7% of persons less than 30 years of age. Some of the difference in magnitude of the effect of older age on PHN may be due to the fact that baseline comparison groups used in the above studies differed. Furthermore, some studies controlled for other risk factors (Choo et al., 1997; Opstelten et al., 2002) while others did not (Goh and Khoo, 1997).

Herpes zoster hospitalizations and complications in healthy children are rare. Guess et al. (1985) found low morbidity among children compared to adults. Among 173 HZ cases in person <20 years of age, there were no occurrences of post-herpetic neuralgia or other late complications and only 2 (1%) were hospitalized. Furthermore, in Iceland, none of the 112 (118 episodes) cases <20 years of age developed moderate or severe pain during the acute illness or postherpetic neuralgia (95% CI, 0 to 0.03) (Petursson et al., 1998).

Hospitalizations

Fewer population-based data are available on HZ hospitalizations and no studies report hospitalization rates that are adjusted for age and high risk conditions. As with HZ incidence, the hospitalization rates increase with increasing age and high rates are seen in persons with suppressed immune systems. In the US, statewide hospital discharge data from Connecticut from 1986–1995 showed an annual crude HZ hospitalization rate of 16.1 per 100 000 person years (Lin and Hadler, 2000). A much lower crude incidence rate was observed in Northern California (2.1 per 100 000 health maintenance organization members) (Coplan et al., 2001), and an intermediate rate of 4.4 per 100 000 was reported in England for 1995 to 1996 (Brisson and Edmunds, 2003). However, because the Connecticut study included all HZ diagnoses rather than HZ as the primary discharge diagnosis, hospitalization rates may have been over-estimated by including individuals with coincidental HZ (Lin and Hadler, 2000). In this study, 31.4% of all hospitalized HZ cases and the majority of HZ hospitalizations among person <50 years of age (61%) had at least one underlying condition that increased the risk or severity of HZ; the majority of these were immunocompromising conditions (82% malignancies, and 8% HIV infection). In contrast, only 8% of the HZ hospitalized cases in England had at least one underlying condition; malignancies accounted for 87% and HIV infection for 6% of the conditions listed (Brisson and Edmunds, 2003). Finally, 67% of the hospitalized cases in the Connecticut study were 64 years or older, while only 55% of those in the California study were 60 years or older emphasizing the importance of age adjustment when comparing studies.

The rates of HZ hospitalizations increase sharply among the elderly (Chant et al., 1998; Coplan et al., 2001; Lin and Hadler, 2000). In one US study, HZ hospitalization rates increased from 21.3 per 100 000 populations among individuals <30 years of age to 1604.5 per 100 000 among those 85 years of age or older. The steepest increase in rates occurred among those over 64 years of age (Lin and Hadler, 2000). Similar patterns are seen in other studies and other countries with hospitalization rates increasing after age 50 or 60 years though rates vary across studies (Brisson et al., 2001; Brisson and Edmunds, 2003; MacIntyre et al., 2003). In Australia, HZ hospitalization remained stable at about 25 per 100 000 populations among persons <60 years of age, then increased from about 50 per 100 000 for the 60–64 age groups to over 300 for the persons 85 years and older (MacIntyre et al., 2003). In England and Canada, HZ hospitalization rates increased from 2 and 1 per 100 000 population in children aged <5 years to 148 and 86 in adults older than 64 years of age, respectively (Brisson et al., 2001).

Some studies of HZ hospitalization did not report rates. In Singapore, between 1993 and 1994, HZ accounted for 3% of total hospitalizations with a mean age of 50 years (range 23 months to 88 years); 58% of HZ hospitalizations were older than 50 years of age (Oh et al. 1997). In contrast, HZ hospitalization in Australia accounted for 0.08% of all hospitalizations between 1998 and 1999 with a mean age of 69 years and 53% were older than 50 years (MacIntyre et al., 2003). Chant et al. (1998) reported that more women in Australia than men were hospitalized for HZ (ratio of women to men was 1.4:–1.6:1), while Oh et al. (1997) reported no sex differences in hospitalization in Singapore.

Deaths

Studies on HZ mortality are few and study methods may not be comparable. As with incidence, hospitalizations, and complications, HZ mortality is more common among the elderly and those with suppressed immune systems. Schmader (2000) surmized based on clinical experience and the absence of HZ-related deaths in cohort studies, that HZ mortality appears to be an infrequent event at least among healthy persons. In Australia during 1971–1993, Chant et al. (1998) reported that 92% of HZ deaths per year occurred among persons older than 65 years. The average crude death rate for the study period was 0.068 per 100 000 population and was 10 times higher for the elderly >65 years (mortality rate 0.60). Females had more than double the rate of death compared to males however these crude mortality rates are not age adjusted (Chant et al., 1998). In Lin’s study in the US, 5.3% of all HZ-related hospitalizations resulted in death and about 52% of those who died had at least one underlying high risk condition, including malignancies; leukemia, and HIV. The risk of death was higher for those with underlying high risk conditions (8.7%) than for those without any high risk conditions (3.7%). In a more recent study in Australia, 4% of 4718 hospitalized for HZ died; however, death was much less common in the group with primary HZ hospitalization (1%) compared to the group with HZ as the secondary diagnosis for hospitalization (6%) (MacIntyre et al., 2003). Finally, in a study in England and Wales from 1993 to 2000, the overall HZ mortality rate was 0.094 per 100 000 person–years however, the number of HZ deaths decreased from 64 in 1993–1994 to 40 in 1999–2000 (Brisson and Edmunds, 2003). The risk of death was <0.014 per 100 000 person–years in persons aged <64 years then increased sharply to 0.566 for persons older than 64 years. Finally, similar results were observed for case fatality ranging from <2 per 100 000 HZ cases in persons <65 years of age to 61 per 100 000 HZ cases in persons older than 64 years of age (Brisson and Edmunds, 2003).

Impact of vaccination

The impact of child and adult vaccination against VZV on the incidence of HZ and PHN remains to be determined. In the US, where there are 9 years of experience with a universal varicella vaccination program, varicella incidence, hospitalizations, and mortality have declined dramatically (Seward et al., 2002; Nguyen et al., 2005; Zhou, 2005). In a study using data from a large Health Maintenance Organization, there was no change in the overall or age-specific incidence of HZ between 1992 and 2002 in an area with about 70% varicella vaccination coverage and a decline in varicella disease that started in 1999 (Jumaan et al., 2005). Survey data from 2000 onwards from Massachusetts show an increase in herpes zoster (Yih et al., 2005).

In 1965, Hope-Simpson hypothesized that the long interval observed between the infection with varicella disease, resulting in the establishment of the latent virus, and the reactivation to cause HZ may be due to internal and external boosting. The external boosting hypothesis, suggesting the need for frequent exposures to VZV to maintain immunity against reactivation of the latent virus has attracted attention, especially with the licensure of the varicella vaccine. Some researchers have speculated that a universal varicella vaccination program, with its associated decline in varicella disease, may have the unintended effect of increasing the incidence of HZ. They note that the decline in varicella disease due to the vaccination program leads to fewer opportunities for persons with a latent virus to boost their immunity through exposure to children with wild-type VZV infection.

Modeling the impact of universal varicella vaccination predicts an initial increase in the incidence of HZ that will occur within 50 years until HZ declines as vaccinated cohorts replace those with a history of varicella disease (Brisson et al., 2002; Garnett and Ferguson, 1996). Although the incidence of HZ is much lower than that of varicella, some studies suggest that the health burden due to HZ is greater than that due to varicella because of the higher rates of complications, hospitalizations and deaths (Lin and Hadler, 2000; Brisson and Edmunds, 2003; Chant et al., 1998; MacIntyre et al., 2003). This has been raised as a serious potential concern in other developed countries considering a varicella vaccination program. However, the health burden on HZ may be overestimated. In a study in England, varicella and HZ hospitalization rates were similar at 4.5 and 4.4 per 100 000 population, respectively (Brisson and Edmunds, 2003). Furthermore, in another study in England and Canada, the reported age specific proportion of cases hospitalized and in-patient days hospitalized were only slightly higher for HZ compared to varicella (Brisson et al., 2001). However, because HZ occurs mainly in the elderly compared to varicella, the overall burden on in-patient hospitalization for HZ is reported to be considerably higher (Brisson et al., 2001; Brisson and Edmunds, 2003; MacIntyre, 2003). In a US study of HZ hospitalizations, the authors did not differentiate between primary vs coincidental cause of HZ hospitalization or death (Lin and Hadler, 2000). Coincidental cases of HZ may be quite common especially among elderly hospitalized adults with long hospital stays, some of whom will die from other causes (MacIntyre et al., 2003). This may contribute to overestimation of the health burden due to HZ.

Several studies reported lower HZ rates among persons exposed to varicella cases (Solomon et al., 1998; Terada et al., 1995; Thomas et al., 2002), compared to those without any exposure. Thomas showed that contacts with ≥5 varicella cases were associated with a strong protective effect against HZ after controlling for occupational and social (OR = 0.29, 95% CI 0.10–0.8) compared to those with no contacts. The two other smaller studies with methodological limitations reported that pediatricians had lower HZ incidence than the general population (Terada et al., 1995), and dermatologists or psychiatrists (Solomon et al., 1998). Terada based his findings on small numbers of cases; while Solomon’s findings were based on low response rates (31% paediatricians) and found no differences between dermatologists and psychiatrists, groups considered to have different rates of exposure to varicella cases. Finally, in all three studies, protection was observed for groups with more exposure to varicella than is generally experienced by the general population.

Other studies have used exposure to children as a proxy for exposure to varicella. Two observational studies in England reported that household or occupational exposure to children was associated with a lower HZ incidence (Brisson et al., 2002; Thomas and Hall, 2004). Brisson reported that people living with children had a significantly lower rate of developing HZ (RR, 0.75 95% CI 0.63–0.89). In a subsequent analysis of the same data, Thomas reported that individuals who reported working with young children (primary school teachers, nursery nurses, playgroup leaders and other child care providers) were significantly less likely to develop HZ after adjusting for age, sex, ethnicity and a child living in the household (RR, 0.70, 95% CI 0.58–0.85). However, it is not clear how other factors that could have contributed to such findings were adjusted for in the analysis including the fact that people with other medical health problems, who are at a higher risk for HZ, are less likely to live with children. Furthermore, Thomas and Hall (2004) in their review of risk factors for HZ report that women in general have higher incidence rates of HZ compared to males; they comment that these findings conflict with the suggested reduced HZ risk associated with exposure to varicella cases or children.

The protection against HZ due to varicella exposure is plausible and is supported by immunologic studies. One reported that 71% of adults who had household exposure to varicella experienced a boost in cellular immune responses (Arvin et al., 1983), and the other found that vaccinated leukemic children with household exposure to varicella were less likely to develop HZ than vaccinated leukemic children without such an exposure (Gershon et al., 1996; Hardy et al., 1991). Yet, other issues still remain unknown, including factors that contribute to immune boosting, the duration of protection from such exposures, and other factors that influence VZV reactivation. Herpes zoster affects mainly the elderly, and if there is a protection from exposure to varicella, the duration of this protection is not known. However, modeling, based on the assumption that the boosting effect lasts for 20 years, suggests that an increase in HZ incidence will occur as early as five (Garnett and Ferguson, 1996) to seven (Brisson et al., 2002) years following the implementation of a mass vaccination program.

Vaccination for prevention of herpes zoster

As populations age and the survival of people with chronic and immunocompromised conditions improves, the incidence of HZ and PHN is expected to increase. Several studies have suggested that declines in VZV-specific CMI increase the risk and severity of HZ (Arvin 1996; Miller, 1980; Oxman, 1995). Early studies on administration of a higher titer varicella vaccine to older adults 55+ have shown that vaccinated persons experience an increased VZV-specific CMI responses to levels typical of those observed in younger persons, in whom the incidence and severity of HZ are lower (Levin et al., 1998; Levin, 2001). A recent study reported on the boosting in VZV-specific cell-mediated immunity from a booster dose of VZV vaccine administered ≥5 years after the first dose (Levin et al., 2003). In 2005, the results of a large placebo-controlled clinical trial to test whether a shingles vaccine in persons 60 years and older would prevent or reduce the risk or severity of HZ and its complications became available. The vaccine was 51.3% efficacious in reducing herpes zoster incidence, 66.6% efficacious in reducing the incidence of postherpetix neuralgia (largely through preventing herpes zoster) and 61.1% efficacious in reducing the burden of illness due to herpes zoster (Oxman et al., 2005). This vaccine, now licensed in the USA with expected licensure in other countries, has the potential to reduce the incidence of HZ or reduce/attenuate the severity of PHN, and complications of HZ. (Johnson et al., 2002; Schmader, 2001; Gilden, 2005). In the U.S. Zoster vaccine was recommended for use in all adults ≥60 years without a contraindication in October 2006 (CDC, 2006).