Varicella Mortality: Trends before Vaccine Licensure in the United States, 1970–1994
- 1Council of State and Territorial Epidemiologists, Atlanta, Georgia
- 2National Immunization Program, Centers for Disease Control and Prevention, Atlanta, Georgia
- Reprints or correspondence: Dr. Jane F. Seward, National Immunization Program, Centers for Disease Control and Prevention, 1600 Clifton Rd. NE, Mailstop E-61, Atlanta, GA 30333 (jseward{at}cdc.gov).
Abstract
We examined varicella deaths in the United States during the 25 years before vaccine licensure and identified 2262 people who died with varicella as the underlying cause of death. From 1970 to 1994, varicella mortality declined, followed by an increase. Mortality rates were highest among children; however, adult varicella deaths more than doubled in number, proportion, and rate per million population. Despite declining fatality rates, in 1990–1994, adults had a risk 25 times greater and infants had a risk 4 times greater of dying from varicella than did children 1–4 years old, and most people who died of varicella were previously healthy. Varicella deaths are now preventable by vaccine. Investigation and reporting of all varicella deaths in the United States is needed to accurately document deaths due to varicella, to improve prevention efforts, and to evaluate the vaccine's impact on mortality.
Varicella is a highly infectious disease that is preventable by vaccine. Before vaccine licensure in 1995, ∼4 million cases per year resulted in ⩽9300 hospitalizations [1] and 100 deaths each year [2]. Children bore the brunt of the health burden, accounting for 190% of cases, 66% of hospitalizations, and 45% of deaths (Centers for Disease Control and Prevention [CDC], unpublished data); however, the risk of severe complications and death was highest among infants, adults [2], and immunocompromised individuals [3, 4]. Moreover, complications and deaths were described commonly among previously healthy individuals [5–9].
Use of the varicella vaccine, which is recommended for routine use among susceptible people >12 months old [10, 11], is anticipated to alter the epidemiology of varicella by shifting the largest proportion of cases from children to adults. High vaccine coverage in childhood, especially if combined with a catch-up immunization program at adolescence, is expected to lead to a dramatic overall reduction in varicella cases and complications among both children and adults [12].
There has been no long-term, comprehensive analysis of varicella mortality for all age groups in the United States. We, therefore, analyzed national mortality data to characterize varicella deaths among United States residents during 1970–1994, the 25-year period before varicella vaccine licensure. These data will serve as baseline data for an evaluation of the impact of the vaccination program on varicella mortality in the United States.
Methods
Sources of mortality data
We focused on the 25 years before vaccine licensure because there is a 2-year delay in the availability of national mortality public-use data (e.g., 1997 is now available) and because it is too early to evaluate the impact of the varicella vaccination program, which became operational in most state health departments by the end of 1996. We analyzed death records for 1970–1994 in the national Multiple Cause Mortality Database, which is maintained by CDC's National Center for Health Statistics (NCHS). The Multiple Cause Mortality Database includes all deaths in the United States, except for 3 years (1972, 1981, and 1982). In 1972, only 50% of deaths were coded for underlying cause of death, and in 1981 and 1982, only 50% of deaths were coded for multiple conditions in 19 of 50 states. To estimate the actual national deaths, NCHS duplicated records for 1981–1982 in the public-use files, and we duplicated 1972 records for this study.
The cause or causes of death obtained from death records were coded according to the Eighth Revision International Classification of Diseases, Adapted for Use in the United States (ICD-8A) [13] for 1968–1978 and according to the 9th revision of the International Classification of Diseases (ICD-9) [14] for 1979–1994. The underlying cause of death is determined by a computer algorithm that evaluates all the codes for cause of death [15]. The algorithm assigns an underlying cause of death on the basis of conditions and their positions as listed on the death certificates. We analyzed deaths for which varicella (ICD-8A code 0.52 and ICD-9 code 0.52) was determined to be the underlying cause of death for consistency with official death statistics reported from NCHS. The algorithm for assigning varicella as the underlying cause of death and the instructions for using the varicella codes were the same throughout the study period.
Preexisting conditions and complications
We defined preexisting conditions as conditions listed on the death record that were likely to have existed before the varicella infection and classified them according to their potential to increase the risk for severe varicella infection. High-risk conditions were defined as those for which the Advisory Committee on Immunization Practices does not recommend vaccine administration because of the risk of severe varicella disease in the vaccine recipient—for example, malignancies and AIDS (comprising human immunodeficiency virus [HIV]/AIDS). We searched death records for additional conditions that may increase the risk for severe varicella infection either due to the underlying condition or to treatment for the condition (e.g., asthma, systemic lupus erythematosus, cystic fibrosis, rheumatoid arthritis and allied conditions, diabetes, and scleroderma). Because of the low proportion of these potentially moderate risk conditions (2.7%) over the 25-year period, we analyzed them with death records of the previously healthy—that is, those that did not list any of the high-risk preexisting conditions (table 1).
Figure 1
Percentage of varicella deaths, by month of death and age group, United States, 1970–1994.
Figure 2
Varicella mortality rates (age-adjusted to year 2000 population), United States, 1970–1994.
We defined varicella-associated complications as conditions listed on the death record that were likely to have occurred as a consequence of varicella infection. We focused on 4 categories: pneumonia, central nervous system (CNS) complications, hemorrhagic conditions, and secondary infections. Pneumonia was defined as either viral or bacterial pneumonia. CNS complications included conditions that affected the CNS as a direct consequence of varicella infection, such as encephalitis, cerebellar ataxia, and Reye's syndrome. Hemorrhagic conditions included purpura and thrombocytopenia. Secondary infections were defined as bacterial infections, including those that affect the CNS, such as brain abscess (table 1).
Data analysis
Analyses were performed with SAS statistical software, version 6.12 (SAS Institute, Cary, NC). We defined children as people ⩽19 years old and adults as people ⩾20 years old. Neonatal deaths included infants <28 days old. We assessed seasonality by examining the number and percentage of varicella deaths, by age group and by month of death. We examined seasonality for all years and for the unduplicated years, to determine if duplicating records distorted findings; however, the results were similar, so we present seasonality for all years. To calculate varicella death rates for the foreign-born population, we used the deaths for 1980–1994 (the only complete 5-year intervals for which foreign-born status was recorded on the death records) and applied the foreign-born population estimates obtained from the Census Bureau's current population reports for 1980 and 1990 [16]. To account for changes in the population age distribution over the 25-year period, we calculated age-adjusted varicella mortality rates by making use of the estimated United States year 2000 population [17]. For calculation of case-fatality rates (CFR), we estimated the number of varicella cases by using the 1970–1994 National Health Interview Surveys [18], which are maintained by NCHS. The National Health Interview Surveys collect information on health-related issues by conducting personal interview surveys annually and uses a nationwide sample of the civilian, noninstitutionalized population in the United States. We calculated age-specific varicella incidence rates for each of the 5-year periods and multiplied those rates by the averaged United States population for the appropriate age group and 5-year period.
Results
From 1970 to 1994, 2262 death records listed varicella as the underlying cause of death, an average of 90 deaths per year; these ranged from 47 deaths in 1986 to 138 in 1973. Table 2 shows the demographic characteristics of decedents whose underlying cause of death was varicella. For the 25-year period, most deaths occurred among children (59.0%) and whites (83.8%). Adults >50 years old accounted for 19.1% of all varicella deaths, and infants <28 days old accounted for 1.2% of all varicella deaths. There was a shift in the age distribution of varicella deaths from children (80.0%) in 1970–1974 to adults (54.3%) in 1990–1994. Whites accounted for most varicella deaths during the study period; however, the proportion of deaths among decedents of races other than white or African American increased from 1.8% to 18.3% during the 25-year period. Foreign-born status was available only for the 5-year intervals from 1980–1984 through 1990–1994. Of the 166 foreign-born decedents in these years, 84.3% were ⩾20 years old.
Table 2
Number and percentage of people who died with varicella as the underlying cause of death, by age group, sex, race, and foreign-born status for all years and for each 5-year interval, United States, 1970–1994.
Seasonality
Varicella deaths showed a strong seasonal pattern consistent with patterns for varicella cases. Overall, 44.2% of deaths occurred between March and May and 8.4% occurred between August and October. Figure 1 shows a seasonal pattern in all age groups, but the pattern was less pronounced among adults ⩾50 years old, compared with adults 20–49 years old and children <20 years old. There was a strong seasonal pattern for deaths regardless of the presence or absence on the death record of a high-risk preexisting medical condition.
Varicella mortality rates
Varicella mortality rates have been consistently higher among children <15 years old than among adults; the highest rates occurred among infants <12 months old. The overall varicella mortality rate showed a pattern of decline followed by an increase (figure 2). The pattern of change in mortality rates differed by age group (table 3). Mortality rates among children <10 years old declined through the 1970s and early 1980s, with the greatest decline among children 5–9 years old, but they then increased in the late 1980s and early 1990s.
Table 3
Five-year average varicella mortality rates, by age group, sex, and race, United States, 1970–1994.
In contrast to the overall decline in mortality rates among children from 1970–1974 to 1990–1994, mortality rates have increased among adults, neonates, and foreign-born adult residents. Adult mortality rates increased 83% from 0.17 per million population during 1970–1974 to 0.31 per million population during 1990–1994. Among neonates, the varicella mortality rate increased from 0.59 per million live births in 1970–1974 to 2.47 in 1990–1994. There also has been an increase in age-specific varicella mortality rates among foreign-born residents. During 1980–1984, when information on place of birth became available, mortality rates were higher among foreign-born than among United States-born residents who were ⩾45 years old (3.48 vs. 1.06) and among those 20–44 years old (2.36 vs. 0.74), but not among those <20 years old (2.36 vs. 2.38). Mortality rates increased among both foreign-born and United States-born residents during 1990–1994 and were consistently higher for foreign-born residents who were ⩾45 years old (5.11 vs. 1.16), 20–44 years old (2.58 vs. 1.56), and <20 years old (4.16 vs. 3.31).
Mortality rates have been similar for males and females throughout the study period. Mortality rates among male patients and female patients declined through the 1970s and early 1980s, then increased for males in the late 1980s and for females in the 1990s. Mortality rates were higher among whites than among African Americans, and mortality rates declined in the early 1970s and 1980s; however, rates began to increase among African Americans in 1990–1994. Mortality rates for people in all other racial categories combined have been higher than rates for either whites or African Americans and have been steadily increasing.
CFR
Overall, the CFR were highest among adults, followed by infants, with the lowest rates among children 1–4 years old, followed closely by children 5–9 years old. With the exception of children 15–19 years old, CFR declined between 51.3%–72.2% in all age groups, with the greatest decline occurring among adults. The fluctuation in CFR among the 15–19-year-old children reflects the small numbers of deaths in this age group. In 1990–1994, adults had a risk 25 times greater and infants had a risk 4 times greater of dying from varicella than did children 1–4 years old (table 4).
Table 4
Varicella case-fatality rate, by age group and year of death (5-year interval) and 95% confidence intervals (CIs), United States, 1970–1994.
Preexisting conditions
Overall, 622 (27.5%) varicella death records listed a preexisting high-risk condition; malignancies accounted for 88.4% of all high-risk conditions; however, the proportion of preexisting conditions varied during the study period. Malignancies were listed on 22.5% of varicella death records in 1970–1974, 35.9% in 1975–1979, and 11.4% in 1990–1994. HIV/AIDS codes first appeared on varicella death records during 1985–1989. By 1990–1994, HIV/AIDS was listed on 2.1% of varicella death records for children ⩽19 years old, 13.5% for those 20–49 years old, and 2.8% for people >50 years old. Most varicella deaths occurred in previously healthy individuals, ranging from 66.5% of child varicella deaths and 56.8% of adult varicella deaths in 1980–1984 to 88.8% and 74.7% of child and adult varicella deaths, respectively, in 1990–1994.
Complications
Overall, the most common complication among people who died from varicella was pneumonia (27.6%), followed by CNS complications (21.1%), secondary infections (8.6%), and hemorrhagic conditions (4.8%). Fatal pneumonia complications affected all age groups (range, 21.3% for 10–19-year-olds to 32.5% for people ⩾20 years old). Over the 25-year period, pneumonia complications declined among people who died from varicella, especially among decedents with preexisting high-risk conditions, but also among decedents who were previously healthy adults (60.4%). Yet among previously healthy children who died of varicella, the proportion with pneumonia complications was relatively stable at ∼20% (table 5).
Table 5
Varicella complications among varicella decedents, stratified by age group, risk status, and 5-year interval, United States, 1970–1994.
CNS complications occurred in 44.1% of varicella deaths in 1970–1974, but only in 5.7% in 1990–1994. CNS complications were more common among children who died of varicella who were <15 years of age and among decedents of all ages who were previously healthy. Nonetheless, fatal varicella complications varied by age and risk status over time. Among children who died of varicella who were previously healthy, the proportion with CNS complications declined from 64.0% in 1970–1974 to 8.9% in 1990–1994. Although fatal CNS complications were less common among previously healthy adult decedents than among child decedents, the proportions also declined, from 11.9% in 1970–1974 to 1.4% in 1990–1994. Reye's syndrome, which is included in our definition of CNS complications, occurred almost exclusively among children <15 years old who died of varicella and declined sharply in the early 1980s. During 1970–1979, there were 170 (16.6%) varicella deaths records that listed possible Reye's syndrome, compared with 28 (2.3%) during 1980–1994.
In contrast to the decline in CNS complications and pneumonia reported in varicella deaths over the 25 years under study, hemorrhagic conditions and secondary infections increased. Hemorrhagic conditions reported in varicella death records increased from 2.5% in 1970–1974 to 8.0% in 1990–1994, whereas secondary infections increased from 3.1% to 13.1%, most prominently among previously healthy children.
Discussion
This is the first long-term analysis of deaths due to varicella in the United States to describe mortality rates, CFR, complications, and preexisting conditions among both child and adult decedents. This study demonstrates important changes in varicella mortality in the United States before the availability of varicella vaccine. Most striking was the shift in the preponderance of varicella deaths from children in the 1970s to adults in the 1990s. During the 25-year time period under study, as deaths declined among children, who made up the majority of cases and deaths in the 1970s, the number of adult deaths increased 3-fold, and, consequently, the relative proportion of adult deaths increased. The increase in adult deaths is consistent with data from the United Kingdom that indicate increases in both the adult varicella mortality rate and the proportion of varicella cases among adults [19, 20].
Increased mortality among adults may partly reflect increased migration to the United States of people from tropical countries, such as Mexico, the Philippines, China, Vietnam, and India, where adults are more likely to be susceptible to varicella [21]. Varicella death rates were higher among adults born in other countries, compared with adults born in the United States. During the time when place-of-birth information was available on the death records (1985–1994), ∼20% of all adult varicella deaths occurred among foreign-born adults.
The increase in neonatal deaths, although relatively few in number, contrasts with a decline in neonatal mortality in the United Kingdom over a similar time period (1967–1990) [22]. The increase in neonatal deaths parallels the increase in adult deaths, which occurred while adult CFR declined, which suggests that adult varicella cases have increased in the United States over the 25-year period. This may have resulted in more mothers acquiring infection during pregnancy or around the time of delivery. The increase in neonatal deaths in the United States may be due to varicella-zoster immune globulin not being offered appropriately to neonates whose mothers are infected with varicella around the time of delivery. Alternatively, it could reflect an increase in deaths in the neonatal period from congenital varicella syndrome (not identifiable on death records with a specific ICD code) secondary to maternal infection during the first 2 trimesters of pregnancy.
The decline in mortality rates among children in the 1970s and 1980s occurred concurrently with improved medical treatment and survival of people with malignancies [23], effective antiviral therapies for varicella zoster virus infections, and improved treatment of life-threatening varicella complications. In addition, reported cases of Reye's syndrome declined dramatically after publication of the association between aspirin and Reye's syndrome and common viral illnesses, often influenza or varicella [24, 25].
The dramatic decline in varicella CNS complications among child decedents paralleled decreases in reports to the CDC of cases of postinfectious encephalitis during the late 1970s and early 1980s [26] and of varicella-associated Reye's syndrome cases [27] during the 1980s. Studies conducted in the United States in the 1970s [28] reported higher rates of encephalitis as a complication of varicella than those conducted in the early 1990s [29]. Similarly, a study in the United Kingdom reported a marked decline in deaths due to encephalitis among children between 1967 and 1985 [19]. It is plausible that Reye's syndrome was misdiagnosed as encephalitis. A postmortem study of 32 children who died of varicella during 1952–1977 found that 12 decedents had acute encephalopathy compatible with Reye's syndrome, but only 2 had definite encephalitis, which suggests that true encephalitis is a rare event in fatal varicella [30]. After CNS complications declined, pneumonia became the most common lethal varicella complication among healthy children and adults, followed by secondary infections. The increase in secondary infections among child decedents may reflect a true increase in severe secondary infections, especially invasive group A streptococcal infections, which have been associated with varicella [31], or it may reflect improved laboratory diagnosis.
In contrast to this decline, we are unable to explain the increase in mortality that occurred among adults and children from the mid-1980s. Although the number of death records that mentioned HIV/AIDS increased during this time, most deaths occurred among otherwise healthy children and adults. This suggests that among healthy people, there may be less awareness of the potential severity of varicella, which could result in delays in seeking health care, in complications not being recognized, or in aggressive therapy for complications not being offered as early to this group. It is also possible that the decrease in deaths among high-risk people reflects health care providers' vigilance in administering antiviral treatment to people with malignancies and other preexisting high-risk conditions. Furthermore, high-risk conditions may have existed but were not reported. For example, information collected for death records is limited to medical conditions, with no information provided on immune suppressive medical therapies, such as systemic steroid use. Therefore, this study may underestimate the number of decedents with preexisting high-risk conditions.
There are several limitations associated with using death records. These include coding errors, antemortem diagnostic errors, inadvertent omissions, underreporting of preexisting conditions, such as HIV infection [32], unavailability of medical records to the certifying physician, death record completion before medical workup, difficulty in determining the underlying cause of death when several disease processes are involved, and misunderstanding of the certification process [33].
Three other limitations should be considered when interpreting the data we report. First, misclassification may have occurred in assigning the underlying cause of death. Misclassification could result in underestimating varicella deaths if complications of varicella, such as pneumonia or group A streptococcal infections, were assigned as the underlying cause of death and varicella was not listed on the death records or was listed as a contributing cause of death. On the other hand, misclassification could overestimate varicella deaths if disseminated herpes zoster, especially among people who were older [28, 29] and/or immunocompromised was diagnosed and recorded as varicella. We found some evidence of misclassification in the less-pronounced seasonality among deaths in older adults, which suggests that varicella diagnosis codes in this age group may have a lower specificity [29]. Furthermore, it can be difficult to clinically distinguish varicella from disseminated herpes zoster. Although there are no published studies on the validity of the varicella code as the underlying cause of death on death certificates, quality assurance of death certificate data is maintained by trained nosologists who code conditions at the state level and, in turn, by nosologists at NCHS who periodically review data from a sample of the submitted death certificates. Furthermore, we compared decedents with varicella listed as the underlying cause of death with those with varicella listed as a contributing cause of death and found similar trends in mortality over time, which suggests that the changes observed are real and not attributable to a change in coding. Currently, the CDC is collaborating with one large state to assess the accuracy of the varicella code on death certificates by reviewing medical records.
Second, several ICD codes for medical conditions of interest in this study lack specificity. For example, Reye's syndrome is coded under the ICD-8A in a broad category, “other disease of the brain,” and there is no specific code for varicella encephalitis. This limited our ability to monitor varicella deaths with these 2 CNS complications. Third, information collected for death records is limited to medical conditions, with no information on immunosuppressive medical therapies, such as systemic steroid use. Therefore, this study may underestimate the number of decedents with preexisting high-risk conditions.
It is too early to evaluate the impact of the varicella vaccine on mortality. Nevertheless, with the availability of a safe and effective vaccine, preventing varicella deaths through vaccination should be a public health priority. Physicians play a key role by ensuring that their patients are properly vaccinated. A high proportion of vaccinated people in the population is needed to achieve herd immunity for the protection of people who can not be vaccinated and who remain at high risk of serious disease or death, such as infants and immunocompromised people. Coverage levels must increase substantially above the 1998 national vaccine coverage of 43% among children 19–35 months old, to reduce disease burden, including mortality, among all age groups [34]. Physicians should be aware that most varicella deaths occur among healthy people, that adults as well as children die from this disease, and that varicella deaths continue to occur, despite the availability of an effective vaccine [5, 6]. Physicians are encouraged to report all varicella-related deaths to state health departments [35]. Investigating and reporting all varicella-related deaths will provide more accurate and complete data on people who die from varicella, such as immunocompromising conditions, therapies, vaccination status, and childhood residence, which can be used to improve prevention efforts.
Acknowledgments
We thank Ken Kochanek, of the National Center for Health Statistics, Division of Vital Statistics, for technical assistance in interpreting International Classification of Diseases codes and mortality data, and Barry Sirotkin, of the National Immunization Program, Division of Data Management, for compiling the 25-year varicella mortality file.
- Received February 14, 2000.
- Revision received April 21, 2000.
- © 2000 by the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 6.288
5-Yr impact factor: 6.103
For The Media
For Authors
- Services for authors
- Self-archiving policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
