Data Reveals Threat of Shingles Epidemic From Vaccine Use;
Spend some time on his interesting website.
Varicella Susceptibility and Incidence of Herpes Zoster among Children and Adolescents
in a Community Under Active Surveillance
Licensure of varicella vaccine by the U.S. Food and Drug Administration in March, 1995 has given rise to concerns that include a potential shift in varicella incidence to susceptible adults and increase in herpes zoster (HZ) incidence. Baseline values prior to widespread vaccination were obtained through distribution of an adolescent survey to all thirteen public middle (7/8th grade) schools in the Antelope Valley, CA health district. Based on 4,216 respondents aged 10 to 14 years, varicella susceptibility is 7.7% (95% C.I., 6.9% to 8.5%) and true cumulative (1987 to 2000) HZ incidence rate is 133 per 100,000 person-years (95% C.I., 95 to 182 per 100,000 person-years).
Since licensure of varicella vaccine by the U.S. Food and Drug Administration in March, 1995, universal vaccination of children, combined with herd protection, has caused dramatic declines in varicella incidence and associated morbidity and mortality of varicella disease [1-3]. This has given rise to unresolved issues that include a potential shift in varicella incidence to susceptible adults and increase in HZ incidence due to a reduction in periodic exogenous exposures to wild-type varicella which previously boosted cell-mediated immunity (CMI) to suppress reactivation [4-6].
One goal of the Adolescent Survey is to assess the number of adolescents still susceptible to varicella in a community with moderate vaccination coverage. Another goal is to investigate the variability of true HZ incidence rates among children and adolescents prior to widespread varicella vaccination, since there are few studies of HZ incidence in the U.S. and only one study that provides true incidence (with pre-varicella person-time removed) among children and adolescents .
In August 2000, all thirteen public middle schools in the Antelope Valley were provided a sufficient quantity of adolescent surveys designed to be included in student enrollment packages and which parents were asked to voluntarily complete and return to the school. After processing enrollments, in September 2000, the schools returned the completed surveys to the Varicella Surveillance Project (VASP). The one-page survey (English on front and Spanish on reverse side) requested the school name, student=s name, date of birth, gender, age at varicella (if applicable), who diagnosed varicella (parent or healthcare provider), age at shingles (if applicable), date and location of varicella vaccination (if applicable or known), duration of residence (in years) in the Antelope Valley, and race/ethnicity. An initial assessment of the responses from each of the schools indicated lack of racial and socio-economic balance, so further effort in April 2001 was made to improve the representation of the sample by encouraging those schools with a poor response to issue the surveys in student classrooms. Again, parents were to complete and return the surveys. Duplicate surveys were eliminated based on exact and probabilistic matching techniques involving the student name and date of birth. Surveys that lacked the student=s date of birth were excluded.
True incidence rates of HZ exclude observation time prior to varicella disease and include observation of person-time from the age of varicella disease until September 2000 (or April 2001 for surveys passed out to classrooms) or age of herpes zoster, whichever event occurs first.
Parents reporting students having a positive history of HZ (shingles) were contacted by telephone if no age (for shingles) was specified to discern if the question regarding shingles was properly understood. Fifty percent of the parents who submitted surveys with an affirmative answer to shingles, including the age that shingles occurred, were contacted to confirm the response and determine if HZ was physician diagnosed.
If we assume middle school students are comprised primarily of adolescents aged 12 and 13 years (in 7th and 8th grade), then the estimated target population is N=10,000 based on an equal proportion of individuals aged 10 to 14 years from the 2000 population census. We also assume the true proportion that is susceptible to varicella is 8% (P=0.08). Suppose we desire a 95% (Z(1Bα) = 1.96) confidence that the estimated proportion differs from the true proportion by no more than 10% of P (ε=0.10). Then the required sample size, n, assuming random selection, must satisfy:
n $ Z(1-α)2 @N @ P @ (1-P)/[(N-1) @ ε2 @ P2 + Z(1-α)2 @ P @ (1-P)]
or n $3,064. We must now adjust for finite population correction (f.p.c.) to reflect the fact that the survey population is finite in size and that sampling is conducted without replacement. The reduced sample size (n=) is given by: n==n/(1+n/N) = 2,345. Future estimates should be updated to reflect the actual proportion of the population still susceptible to VZV.
In practice, the manner in which the adolescent survey was conducted did not lend itself to strict random sampling techniques. The adolescent survey was included in the enrollment package of students entering middle school; however, additional surveys were subsequently passed out to students in classrooms (clusters) for parents to complete and return. Each completed survey was assigned a sequential number and the survey data were entered into a data base using a computer program designed by project staff.
The 95% confidence intervals reported for varicella susceptibility and herpes zoster incidence are computed based on normal approximation and Poisson distribution, respectively.
Of 11,958 seventh and eighth grade students enrolled in all thirteen public schools in the Antelope Valley, VASP received a total of 4,216 (35%) responses providing parental recall of varicella and HZ histories of students aged 10 to 14 years (mean age 12.9 years). The age distribution consisted of 787 (18.7%) aged 10 or 11 years, 2935 (69.6%) aged 12 or 13 years and 494 (11.7%) were aged 14 years. The response by school widely varied from a low of 12% to a high of 94%. This represents 35% of the public middle (or intermediate) school population and 17% (4,216/24,945) of the 2000 census population given for the 10 to 14 age group. The racial balance is consistent with that of the Antelope Valley population census and the California Basic Education Demographics Survey (CBEDS). (Table 1)
Of the total 4,216 respondents, 3627 (86%) presumably had varicella and 589 (14%) indicated a negative or uncertain history. Two hundred sixty six (6.3%) were presumably vaccinated and 323 (7.7%) were either unvaccinated or uncertain as to vaccination. Thus, the percentage of students susceptible to varicella or uncertain as to susceptibility is 7.7% (95% C.I. 6.9% to 8.5%). Of 220 respondents that did supply a date of vaccination, 45 (20%) gave dates prior to licensure of vaccine (March 1995) and therefore were considered unvaccinated. (Table 2) When stratified by age, 10% of children aged 10 and 11 years were still susceptible, declining to 7% susceptible among those aged 14 years. When stratified by race, African-Americans had a significantly higher percentage (12.8%) of susceptibles (χ2 = 24.5, p< 0.005) and higher percentage (10.8%) vaccinated (χ2 = 21.4, p< 0.005) compared to the other races. Adolescents for which parents indicated both varicella disease and receipt of varicella vaccination were classified under varicella in the 1st row of Table 2.
We tallied the number of parents that indicated their child had both (1) a history of varicella and (2) received varicella vaccination. Since medical professionals would be unlikely to administer varicella vaccine to children with a history of varicella, the parent=s affirmative response to these two questions would indicate (a) a possible breakthrough case of varicella if date of vaccination was specified after the licensure date (March, 1995) and prior to onset of varicella, (b) inability to understand the question regarding receipt of varicella vaccine or (c) possibly incorrect memory recall or confusion of varicella vaccination with other traditional vaccinations. Over one-fourth of the responses by Hispanics (27.7%) and African-Americans (25.2%) incorrectly indicated that students with a history of varicella received varicella vaccination. By contrast, only 3.9% of the responses by Caucasians indicated this inconsistency. Of the 518 respondents indicating both varicella and vaccination, 298 (57.5%) gave no date of vaccination, thus the percentage of breakthrough cases was indeterminable in the adolescent study. Based on active surveillance in 2000, of the 837 verified varicella cases reported of all ages, 141 (16.8%) cases occurred 42 or more days after vaccination and were considered possible breakthrough cases. When stratified by age, 25.4% of breakthrough cases were in children 1 to 4, 18.0% were in children 5 to 9, and 9.5% were in those aged 10 to 14 years.
Parents indicated that the students had resided in the Antelope Valley a mean of 8.8 years. African-Americans had the lowest mean of 6.5 years and those Caucasian/Non-Hispanic had the highest mean of 10 years. A majority, 52%, indicated having resided in the Antelope Valley for 10 or more years, 25% resided 5 to 9 years and 24% resided less than 5 years.
Of those respondents indicating they experienced varicella, approximately half were physician diagnosed and half were diagnosed by the parent.
The adolescent survey indicated that 51% of varicella cases occurred before the age of 5 years; 46% occurred between the ages of 5 to 9 years; and 3% occurred on or after age 10 years.
A total of 39 (0.9%) cases of HZ occurred in respondents during the 14-year period from 1987 to 2000. Crude incidence is 72/100,000 person-years (39 cases/54,222 person-years) among the cohort aged <15 years and the true incidence is 1.8 times higher, or 133/100,000 person-years (95% C.I., 95 to 182 per 100,000 person-years) (39 cases/29,249 person-years) among those with a previous history of varicella. (Table 3) Of 20 (51%) parents contacted regarding their affirmative answer to HZ, 19 (95%) indicated HZ was physician diagnosed; one (5%) parent indicated familiarity with HZ symptoms and therefore did not seek physician confirmation. Restricting the data to the prelicensure era, the cumulative (1987 to 1995) true HZ incidence rate is 145 per 100,000 person-years (95% C.I., 86 to 228 per 100,000) (18 cases/12,457 person-years) among children <10.
The 4,216 respondents in the Adolescent Study exceeded the goal sample size of 2,335. The cumulative varicella susceptibility of 7.7% (95% CI, 6.9% to 8.5%) among individuals aged <15 years in the Antelope Valley, CA health district is consistent with the 7% VZV susceptibility among 12- to- 19-year-olds measured by serological testing (National Health and Nutrition Examination Survey, NHANES III, 1988-94) and similar to the cumulative 8.8% theoretically susceptible reported by Finger et al in the 1991 to 1992 Kentucky Behavioral Risk Factor Surveillance System (BRFSS)-based study . Of the 4,216 respondents aged 10 to 14 years in the Antelope Valley, CA Adolescent Survey, if we exclude individuals that cannot be stratified by age of immunity due to unspecified (unknown) age of varicella onset or date of varicella vaccination, 114 (2.7%) and 84 (1.9%), respectively, the cumulative age-specific percentages of varicella susceptibility can be determined among the remaining cohort of 4,018 (95.3%). Since only the cohort aged 10 to 14 years was available in the Antelope Valley study, determination of the age-stratified percentage susceptible relies on parental recall of varicella experience beginning in 1986 to 1990 for those aged <1 year (age 0). These percentages are compared to those derived in the BRFSS-based study  in Figure 1. The similar results may indicate that the limitations of these two studies, including use of parental recall, may have had no significant effect on the percentage of varicella susceptibility among children and adolescents.
The percentage susceptible that we calculated in this study is biased high since those uncertain as to varicella vaccination were all assumed to be unvaccinated. If high varicella vaccine coverage levels can be achieved and maintained among school entry children, only a relatively small percentage of children and adolescents may enter adulthood still susceptible to varicella with the potential of increased morbidity and mortality of varicella disease.
The high vaccination rate among African-Americans may indicate some selective sampling bias or a true difference relative to other ethnic groups.
Interestingly, active surveillance of varicella in the Antelope Valley population indicates the mean age of first cases of varicella (i.e., excluding both breakthrough cases of varicella reported among those vaccinated and second varicella cases occurring among those individuals reporting a previous history of varicella) has increased from 6.9 years in 1995 to 8.1 years in 2000. However, since decline in incidence was observed across all age groups, the burden of varicella disease (in 2000) was lower than in the prelicensure era.
The cumulative (1986 to 2000) true HZ incidence rate of 133/100,000 person-years (95% C.I. 95 to 182 per 100,000 person-years) among children age <15 years in the Antelope Valley population, based on 39 cases occurring during an observation period of 29,249 person-years, compares with the cumulative (1990 to 1992) rate of 133/100,000 person-years (95% C.I. 98 to 176 per 100,000 person-years) in the population-based Harvard Community Health Plan Study  conducted among children <14, based on 49 cases occurring during an observation period of 36,842 person-years. These rates are similar to the cumulative (1947 to 1962) rate of 138/100,000 person-years among 455 individuals aged 10 to 19 years reported in the 16-year study conducted in a medical practice in Cirencester, England, based on 10 cases occurring during an observation period of 7,280 person-years .
The cumulative HZ incidence obtained in the Antelope Valley study represents the mean of the incidence rate among children <10 in the prelicensure era (1987 to 1995) and the incidence rate among children 10 to 14 in the post-licensure period (1996 to 2000). The fact that HZ incidence in the Antelope Valley, CA study compares favorably with the only other historical study reporting true incidence , suggests that individuals aged 10 to 14 years in the post-licensure period have not as yet been influenced by increasing varicella vaccination levels. While CMI to varicella-zoster virus (VZV) in this cohort has persisted sufficiently to suppress reactivation, significant decline in exogenous exposures has occurred only during the past two years (since 1999). Decline in CMI resulting in reactivation may occur sooner than suggested by laboratory measurements of CMI to VZV in individuals. Such measurements conducted in the prelicensure era were confounded and enhanced due to boosting that individuals received from periodic exogenous exposures to natural varicella in the community.
The similar true HZ incidence rates reported in the Antelope Valley, CA and Boston, MA communities , despite differences in study methodology, suggest that the limitations inherent to these studies may have had no significant effect on true HZ rates in children and adolescents. Additionally, considering the HZ incidence reported by Hope-Simpson , studies of HZ incidence rates based on low observations times (while they exhibit wide confidence intervals) may accurately characterize the population due to the mixing of age groups and societal behaviors within the study sample that are representative of the population.
In the near future, it will be interesting to investigate post-licensure incidence rates of HZ among children <10 with a previous history of varicella. Introducing varicella vaccine may reduce the exogenous boosts this cohort receives in elementary school and result in an earlier decline in CMI.
Limitations of this study include the fact that only a cross-sectional sample of the adolescent population was surveyed, although both racial and socio-economic balance was similar to that of the population. Due to the retrospective nature of this study, parents would likely recall the age of children with recent episodes of varicella disease more accurately than early episodes of disease. None of the cases were laboratory confirmed. Since HZ in children and adolescents is relatively mild compared to older adults, many cases may not come to the attention of parents, school nurses, or healthcare providers. Ascertainment of only more serious cases of HZ among children and adolescents by healthcare providers and other age-related biases could further confound observations of incidence rate and advancing age.
Parents appeared to have excellent recall of age of onset and dermatome affected by herpes zoster in their children. Of those parents interviewed by telephone concerning an affirmative response to herpes zoster, all but one parent indicated that prior to seeking healthcare, symptoms of herpes zoster (shingles) were largely unknown. Parents= learning experience at the physician’s office, combined with the fact that shingles was considered to be a rare event in children, likely contributed to enhanced memory of the event.
Since shifts in varicella incidence to older age groups or increases as low as 20% in HZ disease among adults with a prior history of wild-type varicella could potentially counterbalance the medical cost-effectiveness of varicella vaccination [6,10,11], careful post-licensing active surveillance of both varicella and herpes zoster diseases will assist public health authorities in evaluating vaccine use strategies.
We wish to thank all thirteen public schools, including their principals and health clerks, in the Antelope Valley for their participation in this project, including: Almondale Middle School, Challenger Middle School, Del Sur Elementary School, Gifford C. Cole Middle School, High Desert School, Hillview Middle School, Joe Walker Middle School, Juniper Intermediate School, Mesa Intermediate School, New Vista Middle School, Parkview Intermediate School, Piute Middle School, and Shadow Hills Intermediate School.
The Varicella Active Surveillance Project under the Los Angeles County Department of Health Services (LACDHS), Acute Communicable Disease Control unit, is supported by a grant provided by the Centers for Disease Control and Prevention (CDC, Atlanta Georgia).
 Seward JF, Watson BM, Peterson CL et al. Varicella Disease After Introduction of Varicella Vaccine in the United States, 1995-2000. JAMA 2002; 287:606-611.
 Wise RP, Salive ME, Braun MM et al. Postlicensure Safety Surveillance for Varicella Vaccine. JAMA 2000 Setp.; 284(10):1271-1279.
 Vazquez M, LaRussa PS, Gershon AA, Steinberg SP, Freudigman K, Shapiro ED. The effectiveness of the varicella vaccine in clinical practice. N Engl J Med 2001; 344:955-960
 Spingarn RW, Benjamin JA. Universal vaccination against varicella. NEJM 1998 March (Letter); 338(10):683.
 Wack RP. More on varicella immunization. N Engl J Med 1998 June (Letter): 338(26):1927.
 Edmunds WJ, Brisson M, Rose JD. The epidemiology of herpes zoster and potential cost-effectiveness of vaccination in England and Wales. Vaccine 2001, April 30; 19(23-24):3076-3090.
 Donahue JG, Choo PW, Manson JE, Platt R. The incidence of Herpes Zoster. Arch Intern Med 1995 August; 155:1608.
 Finger R, Hughes JP, Meade BJ, Pelletier AR, Palmer CT. Age-specific incidence of chickenpox. Public Health Reports 1994 Nov.-Dec.; 109(6):750-755.
 Hope-Simpson RE. The nature of herpes zoster: A long-term study and a new hypothesis. Proc R. Soc. Med. 1965. 58:9-20.
 Brisson M, Gay NJ, Edmunds WJ, Andrews NJ. Exposure to varicella boosts immunity to herpes-zoster: implications for mass vaccination against chickenpox. Vaccine 2002; 20(19-20):2500-2507
 Thomas SL, Wheeler JG, Hall AJ. Contacts with varicella or with children and protection against herpes zoster in adults: a case control study. The Lancet, July 2, 2002. (Accessed January 15, 2003 at http://image.thelancet.com/extras/01art6088web.pdf.)
Table 1. Adolescent Survey, CBEDS, and Population Census Stratified by Race
1Of 4,216 responses, 197 indicated race as AOther@ or did not answer.
22000-2001 California Basic Educational Demographics Survey for 13 Public Schools with 7th and 8th Grades.
32000 Population Census for individuals aged 10 to 14 years.
Table 2. Adolescent Survey Responses regarding Varicella Disease, Receipt of Vaccination, and Percentage Susceptible stratified by Race and Combined Races
1 Of the 589 responses, 92 indicated AUnknown.@
2 Of the 323 responses, 123 indicated AUnknown.@
Table 3. Adolescent Survey Responses regarding Shingles (Herpes Zoster) Disease and Cumulative (1987 to 2000) True and Crude Incidence Rates of HZ among Children <15
Figure 1. Comparison of Age-Specific Percentages of Varicella Susceptibility
Incidence of Herpes Zoster among Children and Adolescents
in a Community with Moderate Varicella Vaccination Coverage
Active surveillance for herpes zoster (HZ) was conducted for two years (2000-2001) in the Antelope Valley community of 320,000 residents among 290 public and private schools, daycares, and healthcare providers. The true ascertainment-adjusted HZ incidence rate is 307 per 100,000 person-years and 138 per 100,000 person-years among children <10 and individuals aged 10 to 19, respectively. The unadjusted rate among vaccinated children (unadjusted) is 9.5 per 100,000 person-years and an estimated 22 per 100,000 vaccine doses. Unvaccinated children with a previous history of varicella may have greater sensitivity to exogenous exposures (boosting) and a poorer cell-mediated response following primary infection relative to older age groups.
The varicella vaccine was approved by the Food and Drug Administration (FDA) and licensed for use in the United States in March 1995. Although success has been maintained with respect to the prevention of varicella [1-3], questions remain as to how herd protection and increasing impact of vaccination influence the closely related epidemiology of herpes zoster (HZ) .
As early as 1965, Dr. Hope-Simpson, in his qualitative model of the pathogenesis of herpes zoster, included exogenous exposures as a means by which an individual=s immunity is boosted to suppress reactivation of herpes zoster . In Japan, vaccine coverage remains sparse and incidence of wild-type (natural) varicella among children remains high after twenty years . Likewise, vaccine trials in the United States were conducted under conditions where the immune response was boosted due to individuals= periodic reexposures to wild-type varicella. Since 1995, aggressive vaccination programs in the United States, including mandates adopted in some states requiring varicella vaccination of children entering school, have resulted in a decline of varicella disease, providing opportunity to investigate whether exogenous exposures contribute heavily to boosting cell-mediated immunity (CMI) to suppress reactivation of HZ.
In September 1994, the Los Angeles County Department of Health Services entered into a cooperative agreement with the Centers for Disease Control and Prevention (CDC) to establish an active surveillance site for varicella among the 320,000 residents of the Antelope Valley Health Services District with a racial/ethnic composition as follows: 52% white/non-Hispanics, 12% African Americans, 29% Hispanics, and 7% Asians, American Indians, and others. In January 2000, HZ was added to the active surveillance.
This two year study of true HZ incidence represents approximately 147,884 person-years of observation time among individuals <20 years of age with prior varicella experience. It is the first to employ active surveillance of HZ among schools and healthcare providers. Goals of active surveillance are to establish a baseline epidemiological profile of disease incidence in different age groups, monitor varicella vaccine coverage levels, obtain a clinical description of HZ disease among children and adolescents, and detect the impact of increasing levels of vaccination on HZ epidemiology.
The Varicella Active Surveillance Project (VASP) collected case reports of HZ from approximately 300 reporting sites in the Antelope Valley population from January 1, 2000 through December 31, 2001. Reporting sites were comprised of nearly 100% of known public and private schools and day care centers with enrollments of 12 or more children; and approximately 90% of public health clinics, hospitals, private practice physicians, and health maintenance organization (HMO) offices. The largest HMO (Kaiser--serving an estimated 30% of the study population), as well as dermatologists, convalescent and nursing care homes, and other elderly care facilities in the study area were not included among the HZ surveillance, thus biasing the number of reported cases low. HZ cases were accepted on the basis of a healthcare provider diagnosis.
All sites submitted the cases encountered, including name, date of birth, address, telephone number, and race, on a Varicella and Zoster Surveillance Log to the VASP on a biweekly basis. With verbal permission from the parent/guardian, a structured telephone interview was conducted with the caregiver, usually within four weeks of the reported case, to collect detailed demographic, clinical, and health impact data on HZ cases aged <20 years. Data from each interview were entered into a computer database designed by project staff which automatically assigned a sequential case identification number. Completeness of reporting was estimated using two-source capture-recapture methods. This was achieved by tracking all reporting sites that refer to the same case. A Monthly Varivax Immunization Report was submitted on a monthly basis by all 57 providers currently administering the vaccine. Each report contained the number of doses of vaccine administered by age. With telephone and/or fax message prompting following each reporting period, site compliance in submitting reports to VASP was virtually 100%.
Ascertainment-corrected HZ incidence rates are computed by applying two-source (schools and healthcare providers) capture-recapture methods to the number of reported HZ cases [7-10]. The estimator N* of the total HZ cases is given by N*=[(b+1)(c+1)/(a+1)] - 1, where a is the number of HZ cases reported by both ascertainment sources, and b and c denote the number of HZ cases reported by the school and healthcare provider ascertainment sources, respectively. When a $7, there is 95% confidence that the theoretical bias is negligible; however, this does not account for any bias that might result from source dependencies or heterogeneity of the population within an ascertainment source [11, 12].
The distribution of the capture-recapture estimate is skewed in practice, so we have employed goodness-of-fit confidence intervals . The Antelope Valley population estimates are based on the 2000 Census from the U.S. Census Bureau.
Currently, in the Antelope Valley, the crude (population) incidence rate is confounded since it is influenced by the proportion of (a) children still susceptible to varicella and hence not candidates for reactivation, (b) vaccinees in which HZ incidence is considerably low (reportedly, 18 per 100,000 person-years ) and (c) children with a previous history of wild-type varicella. Therefore, in a community with moderate vaccination coverage, in which 50% to 60% of children <10 have been vaccinated, it is useful to stratify true HZ incidence rates (i.e., by excluding pre-varicella person-time) among (1) those unimmunized with prior wild-type (natural) varicella experience, and (2) those individuals immunized with varicella (Oka strain) vaccine. Reporting a single crude rate of HZ incidence would not be meaningful for a bimodal distribution and could potentially mask important trends and wide differences in rates among vaccinated and unvaccinated cohorts. Likewise, due to underreporting of HZ cases ascertained via active surveillance, failure to adjust for reporting completeness would yield uninterpretable HZ incidence rates.
The 290 surveillance units that participate in the active surveillance of HZ may be divided into two main ascertainment groups: schools comprised of 76 (26%) preschools/daycares, and 101 (35%) public and private elementary, middle, and high schools; and healthcare providers comprised of 81 (28%) private practice physicians, 18 (6%) HMO offices, 3 (1%) hospitals, and 11 (4%) public health clinics.
Of the 723 cases of HZ reported to VASP during 2000 and 2001, 61 (8.4%) were excluded since they referred to individuals residing outside the surveillance boundaries, 14 (1.9%) were ineligible since they were reported by miscellaneous sites (other than schools or healthcare providers), 9 (1.2%) involved children <10 years old that received varicella vaccination, and 639 (88.4%) were among individuals of all ages with a history of natural varicella.
The 639 cases reported in unvaccinated individuals during two years have a bimodal distribution (not age-standardized) with approximately 21%, or 134 cases, occurring in children and adolescents <20 , and 24%, or 152 cases, occurring in adults 70+. (Table 1) The mean age is 48.5 years and the age range of the individuals is <1 to 96 years. Of the 639 cases of all ages, 501 (78%) indicated race/ethnicity as follows: 366 (73.1%) White, 89 (17.8%) Hispanic, 36 (7.2%) Black, and 10 (2.0%) Asian/American Indian. If we assume annual HZ incidence ranges from 250 to 300 per 100,000 population (typical rates for the U.S.), overall case ascertainment is approximately 32% to 39%.
Based on monthly reports collected during 2000 and 2001, summer months of July and August exhibited a combined peak of 127 (20.0%) cases. Other studies have suggested a similar trend [14, 15]. Active surveillance ascertained a mean of 27 cases (range 16 to 36) each month. (Table 1)
The assumption is rarely correct that active surveillance achieves 100% case ascertainment. Consider that schools reported 54 cases and healthcare providers reported 91 cases of HZ among unvaccinated children and adolescents aged five to nineteen years. Of these 145 case reports, 19 were duplicates. While all 54 cases that were reported by schools sought attention from healthcare providers participating in the active surveillance project, only 19 (35%) were reported by healthcare providers. We refine this analysis further by applying capture-recapture methods to derive a reporting completeness of 50% (95% CI 34% to 65%) among individuals aged 5 to 19 years. Since parents are encouraged by schools to submit a physician=s written notice to excuse or validate the students= absence, this positive dependence among school and healthcare provider ascertainment sources yields a capture-recapture estimate that serves as a likely lower bound .
Consistent with past studies of HZ, we first compute crude and unadjusted incidence rates of HZ in children using a numerator consisting of the actual cases reported during 2 years via active surveillance and a denominator consisting of figures from the 2000 population census. The crude HZ incidence rate is 63 per 100,000 person-years among children <10 (or 74 cases, i.e., 65 HZ cases in unvaccinated plus 9 HZ cases in vaccinated /116,548 person-years) and 57 per 100,000 person-years among those aged 10 to 19 years (or 69 cases/120,842 person-years). Since reporting completeness of HZ cases among children 5 to 19 is 50%, the ascertainment-corrected crude rates are approximately double the unadjusted rates.
HZ Incidence Among Children and Adolescents with a History of Natural Varicella
The age-specific 2000 population census for the Antelope Valley, indicates 58,274 children <10 in the Antelope Valley. To derive true incidence among children with a previous history of varicella, we must deduct those children in this cohort that have been vaccinated as well as those still susceptible to varicella. Deducting 33,918 (58.2%) vaccinated children and considering only the cohort <1, or 4,518 (7.8%), remains susceptible, we have remaining an estimated 19,838 (34.0%) with a previous history of varicella. The cumulative (2000 and 2001) ascertainment-adjusted HZ incidence rate is 307 per 100,000 person-years (122/39,676). This incidence rate represents a likely lower bound estimate given (a) the positive dependency of the school and healthcare ascertainment sources, (b) the assumption that only the cohort aged <1 remains susceptible, and (c) the active surveillance sites represent 100% of the available sources for ascertainment of HZ cases serving the entire Antelope Valley population. (Table 2) Only by investigating the realistic range of susceptibles in this cohort could we properly eliminate pre-varicella observation time from the denominator of the incidence rate calculation and finally derive an upper bound estimate of true HZ incidence among children <10.
Similarly, for individuals 10 to 19, the population census indicates 60,421 individuals. Deducting an estimated 3,296 (5.5%) reported as vaccinated and 3,021 (5.0%) estimated as susceptible, we have 54,104 (89.5%) with a previous history of varicella. The cumulative (2000 and 2001) ascertainment-adjusted HZ incidence rate is 138 per 100,000 person-years (149/108,208) among individuals 10 to 19. (Table 2).
HZ Incidence among Vaccinated Children <5
The crude (or unadjusted) HZ incidence rate among vaccinated healthy children aged <5 years is 11.2/100,000 person-years (4/35,673) and 7.9/100,000 person-years (3/38,208) in 2000 and 2001, respectively, for a cumulative (2000 and 2001) crude incidence rate of approximately 9.5/100,000 person-years. (Table 3) Again, active surveillance ascertained approximately 50% of the HZ cases in vaccinees if the true incidence rate is 18.8/100,000 person-years as reported in vaccinated children aged 12 months to 12 years . The crude number of HZ cases per 100,000 vaccine doses administered was 26 in 2000 and 18 in 2001, yielding a cumulative (2000 and 2001) rate of 22 per 100,000 vaccine doses administered.
Clinical Description of HZ in Unvaccinated Individuals <20
Distribution of the 134 HZ unvaccinated cases by age is given in Figure 1. All cases were physician diagnosed and 121 (90%) were available for interview. The mean age of interviewed cases was 11 years. Caregivers for 80 (74.1%) of 108 cases that recalled age of varicella, indicated onset of varicella before age five years. The average time between varicella and HZ was 8.0 years (mean age of zoster 10.9 years minus mean age of varicella 2.9 years).
Of the 134 cases <20 years of age, 124 (93%) indicated race/ethnicity as follows: 86 (64%) White, 29 (22%) Hispanic, 5 (4%) Black, and 4 (3%) Asian/American Indian. There was no significant difference between the distribution by race/ethnicity among reported cases of HZ in individuals <20 and that of the Antelope Valley population for that same age group (χ2 = 12.6, p<0.10).
Pain or paresthesia (burning, itching) was the initial symptom prior to the appearance of the HZ rash in 48 (40%) cases. The majority, 89 (74%) cases, involved thoracic dermatomes, followed by cervical then lumbar . Prescription pain medications were administered to 19 (16%) of the cases and 62 (51%) cases took an antiviral (acyclovir or Zovirax). Residual scarring was reported in 44 (36%) cases. Interviewees cited stress as a possible contributing factor (e.g., death of a household member, parental divorce, school, etc.) prior to HZ occurrence in 43 (36%) cases.
Nine cases of HZ were reported in vaccinated children aged less than ten years with an average interval of twenty-seven months (range 6 to 53 months) between vaccination and occurrence of HZ. No laboratory test was performed to confirm the diagnosis or to determine the virus strain associated with these cases. Vaccination dates were all confirmed by the healthcare providers administering the vaccine.
In the Antelope Valley, the true cumulative (2000 to 2001) incidence rate of HZ among children <10 with a history of varicella is approaching that of older adults. This result differs from historical studies.
An adolescent survey (unpublished VASP data) conducted in 2000 among 4,216 students aged 10 to 14 (37% of the public middle school population) in essentially the same Antelope Valley study population indicated that true cumulative (1987 to 1995) incidence was 145 per 100,000 person-years (95% CI, 86 to 228 per 100,000 person-years) (18 cases/12,457 person-years) among children <10 with a history of varicella in the prelicensure era. This compares with the true incidence rate of 133 per 100,000 person-years (95% CI, 98 to 176 per 100,000 person-years) among children <14 reported by a population-based study conducted in Boston, Massachusetts . Interestingly, these rates are similar to the rate of 138 per 100,000 person-years reported both in the medical practice of Hope-Simpson and the present Antelope Valley study based on active surveillance among individuals aged 10 to 19. The similarities between these true incidence rates suggest that despite differences in study methodology, the limitations inherent to these studies may have had no significant effect on true HZ incidence.
Given the corroborative studies above serve as surrogates for baseline incidence data in the prelicensure era, the cumulative (2000 to 2001) true incidence rate of HZ has undergone a conservative two-fold increase (307/100,000 person-years ) 147/100,000 person-years) among children <10 in the postlicensure period.
In 2001, approximately 60% of the children <10 in the Antelope Valley had received varicella vaccine. This was associated with a 71% decrease in varicella incidence from 1995 to 2000 . It is interesting to speculate that increasing vaccination levels reduced exogenous exposures to wild-type VZV, which in turn could have caused VZV immunity to wane among these children to a level at which VZV was more likely to reactivate, causing HZ. Children appear to have a poorer cell-mediated response following primary infection which predisposes them to early reactivation (as is the case with herpes simplex virus) . Children may also have greater sensitivity to boosting effects from exogenous exposures relative to older age groups.
The hypothesis that periodic exposures to wild-type varicella suppresses reactivation of HZ was first introduced in 1965, based on observations of varicella and HZ rates in a medical practice in Cirencester, England . Using the Royal College of General Practitioners (RCGP) data set in England, a notable decrease in zoster incidence was reported in children following a varicella epidemic in the winter of 1980 . Japanese pediatricians in their 50s and 60s who were re-exposed to VZV demonstrated HZ incidence rates 1/2 to 1/8 that of the general population . In 1983, Arvin et al noted a boost in CMI in 71% of adults who were exposed to varicella patients in the family . In a more recent study of HZ in physicians, pediatricians (who have a greater incidence of exposure to VZV) were reported to have lower rates of HZ than psychiatrists (who had the lowest VZV exposure rates) . Finally, in 2002, epidemiological evidence in England and Wales demonstrated higher incidence of HZ among adults living with children compared to those living without children  and similarly, in a case-control study in South London, UK, it was suggested that re-exposure to VZV via contact with children seems to protect latently infected individuals against zoster .
Interestingly, the cumulative (2000 and 2001) ascertainment-adjusted HZ incidence rate of 138 per 100,000 person-years among individuals aged 10 to 19 years determined via active surveillance is the same as that reported by Hope-Simpson in the same age group in the pre-licensure era . This may indicate that CMI to VZV has not as yet declined substantially to cause increased HZ reactivation in this cohort. However, only recently during the past two to three years (beginning in 1999) did this cohort experience a dramatic decline in frequency of re-exposures to wild-type varicella . Thus, this cohort is more likely to have already had boosts to immunity because of prior re-exposure.
Relative to the HZ incidence rate in the prelicensure era, the higher rate in the post-licensure period among children <10 with a history of natural varicella will become less of an issue as vaccinated children enter this age group; however, there is concern that HZ incidence will increase among adults who are subject to increased morbidity and mortality of disease due to waning of immune containment [26,27]. Increases in reactivation among adults due to reduction in exogenous exposures is implied by vaccine trials in which recipients demonstrate boosting of CMI to VZV .
Since the VASP relies on epidemiologic methods, it may not successfully control for confounding and bias in the analysis of true incidence of HZ by age. Factors such as increasing HMO enrollments, easier access to healthcare in general, increasing enrollments of children in daycares and preschools (causing exposures to varicella at younger ages), and greater diagnostic awareness of HZ (due to active surveillance) may all contribute to HZ incidence rates that are higher in the Antelope Valley compared to rates reported in historical studies. Additionally, increases in school-related stress and expanding use of immunosuppressive modalities may account, in part, for higher incidence rates of HZ in this study compared to studies conducted prior to the widespread development of these trends.
A limitation in this study, as in all previous studies of HZ incidence as well as in trials of antiviral drugs, is that HZ disease was not laboratory confirmed; however, the expected low rate among vaccinated children serves as a control indicating cases of HZ are not generally being misdiagnosed. We postulate that due to the evidence that asymptomatic reactivations occur more frequently in OKA-strain VZV compared to wild-type VZV , the HZ incidence rate among vaccinees is not as sensitive to decreased exogenous boosting as is the wild-type virus. Thus, the HZ incidence rate among vaccinees in the postlicensure period has thus far been unaffected by the fact that incidence of wild-type varicella is no longer high in the community.
The poor reporting of HZ cases by healthcare providers under active surveillance highlights the importance of having another source (schools) for case ascertainment and demonstrates why application of capture-recapture methods are critical for incidence estimation [8, 9]. The largest HMO in the study population (Kaiser) did not participate in the bimonthly reporting of HZ cases to the VASP. Based on a computer search of medical records for cases of HZ over a six month period from January through June 2002, it is estimated that this HMO would have contributed some 60 cases among individuals <20 and 400 cases in adults during the two year study period. These additional cases suggest the estimated reporting completeness of HZ cases is reasonable and the capture-recapture assumptions of source independence and homogeneity of the source populations are plausible.
Physicians and healthcare providers are encouraged to report any suspected adverse events that occur after varicella vaccination to the Vaccine Adverse Events Reporting System (VAERS) established by the U. S. Department of Health and Human Services. However, since immunized individuals are not the same ones that experience the deleterious effects (in terms of increased incidence of HZ), VAERS is unsuitable for reporting such adverse events concerning interaction of varicella and HZ epidemiology. Furthermore, there is no legal precedent requiring a vaccine manufacturer to perform studies on individuals that have not been immunized with their product. Thus, postmarketing surveillance for such deleterious effects remains an important tool in assessing varicella vaccine safety.
This study highlights the importance of continued HZ monitoring. The results of this study will serve as a useful baseline for further studies of HZ in the Antelope Valley. It may also encourage other investigators to examine HZ rates on a longitudinal basis, not only among vaccine recipients, but also among those who have not received vaccine . If a clear vaccine-associated increase in HZ is demonstrated by longitudinal data and is confirmed in further studies in broader populations, this should be considered by public health authorities in evaluating vaccine use strategies.
We wish to thank the healthcare providers, school nurses, child care center directors, and other professionals and members of the Antelope Valley community who reported cases of HZ in 2000 and 2001. The Varicella Active Surveillance Project under the Los Angeles County Department of Health Services (LACDHS), Acute Communicable Disease Control unit, is supported by a grant provided by the Centers for Disease Control and Prevention (CDC, Atlanta Georgia).
 Seward JF, Watson BM, Peterson CL, et al. Varicella Disease After Introduction of Varicella Vaccine in the United States, 1995-2000. JAMA 2002; 287(5):606-611.
 Wise RP, Salive ME, Braun MM, et al. Postlicensure Safety Surveillance for Varicella Vaccine. JAMA 2000 Sept.; 284(10):1271-1279.
 Vazquez M, LaRussa PS, Gershon AA, Steinberg SP, Freudigman K, Shapiro ED. The effectiveness of the varicella vaccine in clinical practice. N Engl J Med 2001; 344:955-960.
 Spingarn RW, Benjamin JA. Universal vaccination against varicella. N Engl J Med 1998 March (Letter); 338(10):683.
 Hope-Simpson RE. The nature of herpes zoster: A long-term study and a new hypothesis. Proc R. Soc. Med. 1965. 58:9-20.
 Asano Y. Varicella vaccine: the Japanese experience. J Infect Dis 1996; 174(Suppl 3):S310-S313.
 Regal RR, Hook EB. Goodness-of-fit based confidence intervals for estimates of the size of a closed population. Stat Med. 1984;3:287-291.
 Hook EB, Regal RR. The value of capture-recapture methods even for apparent exhaustive surveys: the need for adjustment for source of ascertainment intersection in attempted complete prevalence studies. Am J Epidemiol 1992; 135:1060-1067.
 McCarty DJ, Tull ES, Moy CS, Kwoh CK, LaPorte RE. Ascertainment corrected rates: applications of capture-recapture methods. Int J Epidemiol 1993; 22:559-565.
 Hook EB, Regal RR. Effect of Variation in Probability of Ascertainment by Sources (AVariable Catchability@) upon ACapture-Recapture@ Estimates of Prevalence. Am J Epidemiol 1993; 137:1148-1166.
 Seber GAF. Closed Population: Single Mark Release (Chapter 3). In: The Estimation of Animal Abundance and Related Parameters. London, England: Charles Griffin & Company Limited; 1973:59-129.
 Robson DS, Regier HA. Sample Size in Petersen Mark-Recapture Experiments. Transactions of the American Fisheries Society, 1964; 93(3):215-226.
 Vaccination and Immunization Education: a teleconference series. Strategic Institute for Continuing Health Care Education. Strategic Implications International, 1998:45.
 Glynn C, Crockford G, Gavaghan D, Cardno P, Price D, Miller J. Epidemiology of shingles. J. R. Soc. Med., 1990; 83:617-619.
 Hellgren L, Hersle K. Statistical and clinical study of herpes zoster. Geront. Clin., 1966; 8:70-76.
 Hook EB, Regal RR. Capture-recapture methods (Letter). Lancet 1992; 339:742.
 Guess HA, Broughton DD, Melton LJ, Kurland LT. Epidemiology of herpes zoster in children and adolescents: a population-based study. Pediatrics 1985; 76:512-518.
 Donahue JG, Choo PW, Manson JE, Platt R. The Incidence of Herpes Zoster. Arch Intern Med 1995 August; 155:1608.
 Miller E, Marshall R, Vurdien J. Epidemiology, outcome and control of varicella-zoster infection. Reviews in Medical Microbiology, 1993; 4:222-230.
 Joseph CA, Noah ND. Epidemiology of chickenpox in England and Wales, 1967-1985. Br Med J 1988; 296:673-676.
 Terada K, Hirago U, Kawano S, Kataoka N. Incidence of herpes zoster in pediatricians and history of reexposure to varicella-zoster virus in patients with herpes zoster. Kansenshogaku Zasshi 1995; 69(8):908-912.
 Arvin A, Korpchak C, Wittek A, Immunologic evidence of reinfection with varicella-zoster virus. J. Infect. Disease 1983; 148:200-205.
 Solmon BA, Kaporis AG, Glass AT, Simon SI, Baldwin HE. Lasting immunity to varicella in doctors study (L.I.V.I.D. study). Am Acad Dermatol 1998; 38:763-765.
 Brisson M, Gay NJ, Edmunds WJ, Andrews NJ. Exposure to varicella boosts immunity to herpes-zoster: implications for mass vaccination against chickenpox. Vaccine 2002; 20(19-20):2500-2507.
 Thomas SL, Wheeler JG, Hall AJ. Contacts with varicella or with children and protection against herpes zoster in adults: a case-control study. The Lancet, July 2, 2002; http://image.thelancet.com/extras/01art6088web.pdf.
 Levin MJ, Muray M, Zerbe GO, White CJ, Hayward AR. Immune responses of elderly persons 4 years after receiving a live attenuated varicella vaccine. J Infect Dis 1994; 170:522-528.
 Straus SE, Ostrove JM, Inchauspe G. Varicella-zoster virus infections: biology, natural history, treatment, and prevention. Ann Intern Med 1988; 108:221-237.
 Krause PR, Klinman DM. Varicella vaccination: Evidence for frequent reactivation of the vaccine strain in healthy children. Nature Medicine 2000 April; 6(4):451-454.
 Krause PR, Klinman DM. Efficacy, immunogenicity, safety, and use of live attenuated chickenpox vaccine; The Journal of Pediatrics 1995; 127:518-525.
Figure 1. Age Distribution of Unvaccinated Cases of HZ Reported among individuals <20 (n=134), Antelope Valley, 2000 to 2001.
Table 1. Unvaccinated Cases of HZ Reported by Schools and Healthcare Providers by Month and Age Category in the Antelope Valley, 2000 and 2001.
Table 2. Likely Lower Bound Cumulative (2000 to 2001) Incidence Rate Estimates of Herpes Zoster Among Individuals with a Previous History of Natural Varicella in the Antelope Valley.
1Includes 1 recurrent HZ case.
2Total doses reported by providers was 27,134 based on the mean of 23,754 and 30,514 in vaccinated children <10 in year 2000 and 2001, respectively. This figure is approximately 80% of the total doses reported as shipped by the vaccine manufacturer. A later report by vaccine manufacturer indicated total doses reported represented 72% of doses shipped. We used the more conservative report.
3This figure is based on the mean of 2,530 and 4,061 vaccinated individuals 10 to 19 in year 2000 and 2001, respectively.
4Based on census population for those individuals <1 year old (which should not be vaccinated); therefore the estimated number of susceptibles is biased low since it does not consider susceptible children 1 to 9 years old.
5An adolescent survey conducted among all 13 public middle schools in the Antelope Valley indicated 10% susceptible among 10- and 11-year-olds declining to 7% among 14 year-olds. Based on this trend we estimated 5% as the percentage susceptible among 11- to 19-year-olds.
6Based on an ascertainment-corrected number of 122 (95% CI, 90 to 205) HZ cases (a=11, b=31, c=45, completeness 53%) and observation time associated with population with a history of natural varicella. The ascertainment corrected incidence rates were 266 and 339 per 100,000 person-years in 2000 and 2001, respectively.
7Based on an ascertainment-corrected number of 149 (95% CI, 104 to 287) HZ cases (a=8, b=24, c=53, completeness 46%) and observation time associated with population with a history of natural varicella. The ascertainment-corrected incidence rates were 146 and 120 per 100,000 person-years in 2000 and 2001, respectively.
Table 3. Prevalence of Vaccinated Children <5 in 2000 and 2001 in the Antelope Valley.
1Assumes all annual doses were administered at the beginning of the year.
Using Capture-Recapture Methods to Assess Varicella Incidence
in a Community Under Active Surveillance
The Varicella (chickenpox) Active Surveillance Project has been conducting active surveillance since January 1, 1995 in the high desert community known as Antelope Valley, CA (population 300,000) among 300 public and private schools, daycares, and healthcare providers. Capture-recapture methods were applied to estimate reporting completeness for 1995 varicella incidence data and these were compared with the national average incidence rates by age reported by the National Health Interview Survey (NHIS). Varicella cases reported among individuals aged <20 years reflect under-reporting in excess of 50%. Despite limitations on accuracy, capture-recapture estimates are a reasonably accurate, quick, and inexpensive approach in epidemiologic studies.
Capture-recapture methods, derived from techniques developed for studies of animal abundance, estimate the true population size by evaluating the degree of overlap among incomplete lists of cases from existing data sources. Capture-recapture involves consideration of three basic assumptions: (1) the population is closed, i.e., no significant changes occur in the population under study during the investigation (e.g., due to migration or death); (2) there is no loss of tags so that individuals can be matched from capture to recapture; and (3) each individual has the same probability of being caught in the second sample so that capture in the first sample does not affect capture in the second sample, i.e., the samples are independent. While assumptions (1) and (2) are often met, failure to meet assumption (3) can lead to inaccurate and sometimes misleading resultsBsince in epidemiological investigations, sources often display dependence and heterogeneity of capture probabilities .
This is the first paper to apply capture-recapture to varicella, a common communicable disease that displays seasonality. Capture-recapture methods have been used in the 1980s and 1990s in studies to estimate the completeness and prevalence of many infectious and chronic diseases, including, pertussis , tetanus , sexually transmitted disease , AIDS [5,6], measles[7,8], meningococcal disease (which is also a highly seasonal infection) [9,10], HIV-1 , and cryptococcosis .
The major question individuals have regarding capture-recapture is AWill capture-recapture give you the >truth=, i.e., an extremely accurate estimate of the incidence of disease?@ Simply answered, no--it will not. The estimates presented in most epidemiologic studies are extremely poor, missing 10% to 90% of the cases, with a high degree of variation [13-16]. Therefore, the options are (a) not to use capture-recapture and report varicella cases from which incidence rates are almost uninterpretable, (b) try to count every case of varicella which is horrendously expensive and slow, or (c) utilize capture-recapture which, depending on the degree to which the assumptions are satisfied, can be a compromise, reasonably accurate, quick, and inexpensive approach.
The Varicella Active Surveillance Project (VASP) collected case reports of varicella from approximately 300 reporting sites representing nearly 100% sampling of the Antelope Valley population from January 1, 1995 through December 31, 1995. A case of varicella is defined as illness with acute onset of a diffuse papulovesicular rash without other known cause. Reporting sites were comprised of two main ascertainment sources: schoolsBconsisting of all public and private schools and day care centers with enrollments of 12 or more children; and healthcare providersBconsisting of public health clinics, hospitals, private practice physicians, and health maintenance organization (HMO) offices.
All sites submitted the cases encountered, including name, date of birth, address, telephone, and race on a Varicella Surveillance Log to the VASP on a biweekly basis. With verbal permission from the parent/guardian, a structured telephone interview was conducted with the caregiver usually within four weeks of a reported case to collect detailed demographic, clinical, and health impact data. Cases were considered probable if the caregiver could not be contacted after five attempts or as the result of a disconnected phone number or no forwarding change of address. Data from each interview were entered into a computer database designed by project staff which automatically assigned a sequential case identification number. Completeness of reporting was estimated using two-source capture-recapture methods. This was achieved by tracking all reporting sites that refer to the same case. With telephone and/or fax message prompting following each reporting period, site compliance in submitting reports to VASP was 100%.
It is possible, using log-linear capture-recapture methods [17, 18], to evaluate the possible dependency of the lists when three or more ascertainment sources are utilized; however, this is not possible when there are only two ascertainment sourcesBsuch as schools and healthcare providers. In this latter case we must resort to comparing the capture-recapture estimate with a criterion (“gold”) standard. Thus, annual varicella incidence in 1995 in the Antelope Valley based on cases ascertained via active surveillance is compared with the national incidence reported by the National Health Interview Survey (NHIS) for 1990 to 1994. The NHIS data are obtained through personal interviews with household members. Interviews are conducted each week throughout the year by a permanent staff of interviewers employed by the U.S. Bureau of the Census. The NHIS is a criterion standard for comparison since data collected from approximately 49,000 households including about 132,000 persons in a calendar year forms the basis of annual varicella incidence estimates in the civilian noninstitutionalized population of the United States . While varicella incidence rates via active surveillance were available in subsequent years, 1996 through 2001, the NHIS national varicella incidence rates were unavailable for comparison and would have been affected by variation in the use of varicella vaccine.
The population estimates for 1995 were projected from the 1990 MARS files (Modified Age, Race, and Sex), produced by the U.S. Census Bureau and modified by local death rates, migration rates, and fertility rates within age, sex, and racial/ethnic groups.
The distribution of the value of the capture-recapture estimate is skewed in practice and so as to avoid misleading results associated with standard error, goodness-of-fit based confidence intervals were utilized . Let a be the number of cases reported by both ascertainment sources, and b and c be the number of cases reported by healthcare providers only and schools only, respectively. Then the numerically unbiased estimate of the number of cases missed by both ascertainment sources, dnue, is given by dnue = (bc)/(a+1) and the numerically unbiased estimate of the total cases, pnue, using the Lincoln-Peterson estimate is given by
Pnue = [(a + b + 1)(a + c + 1)/(a+1)] - 1.
Reporting completeness of varicella cases via active surveillance using two ascertainment sources was 46% with an estimated 4,498 (95% CI=4,122 to 4,962) cases in the 1 to 19 age group. (Table 1) The NHIS annual varicella incidence in the U.S. of 53.2 cases/1,000 in this same age group is 4.2% higher than the ascertainment-corrected incidence in the Antelope Valley of 50.9 cases/1,000 (4,498/88,379). (Table 2)
Cases among individuals aged 1 to 19 years were stratified by quarter to evaluate the effects of seasonality on the capture-recapture estimate. The sum of the estimates for each of quarters 1 through 4 yielded an estimated 4,500 cases, or an incidence of 50.9 cases/1,000 (4,500/88,379), thereby differing from the NHIS incidence by -4.2%. (Table 3)
Finally, to test the assumption of uniform (or homogenous) reporting probabilities, verified cases were stratified by both severity of disease index and age. A severity of disease index of 1 indicates no complications. A severity of disease index of 2 or greater is assigned to a case of varicella with an associated complication ranging from bacterial infection with a prescription of antibiotics, or fever greater than 40°C for one reading, to lower respiratory infection, hospitalization, and significant disability or death. The capture-recapture estimates corresponding to the 1 to 4, 5 to 9, 10 to 14, and 15 to 19 age categories differed from the NHIS incidence by -8.5%, -0.9%, -11.1%, and +15.4%, respectively. The sum of the capture-recapture estimates yielded 4,453 cases (or 50.4 cases/1,000), differing from NHIS average annual U.S. incidence by -5.2%. (Table 4)
As another check of homogenous reporting, the verified cases were stratified by grading the number of lesions: less than average (<50 lesions), average (50 to 500 lesions), and greater than average (>500 lesions). The sum of capture-recapture estimates yields 4,196 cases, or 47.5 cases/1,000, differing from the NHIS average annual U.S. incidence by -10.7%. (Table 5)
After varicella vaccine was introduced into the childhood immunization schedule, NHIS national incidence data were unavailable. The reporting completeness for combined years 1995 to 2001 based on verified varicella cases in individuals aged 5 to 18 years is 64% with an annual range of 60% to 69%. Interestingly, reporting completeness using capture-recapture methods in 1995, a year of high varicella incidence, closely compares to that of 1999 and 2000 where reported varicella cases of all ages were dramatically reduced by approximately 80% and 70%, respectively. (Table 6)
The racial/ethnic breakdown based on the estimated 1995 census for 1 to 19 year olds in the Antelope Valley of 59.2% White/Non-Hispanic, 7.4% African-American, 28.4% Hispanic, and 5.0% for Asian/Other compared favorably with that of reported varicella cases with 57.1% White/Non-Hispanic, 11.6% African-American, 26.6% Hispanic, and 4.8% Asian/Other. The California Basic Education Demographics Survey (CBEDS) suggests a higher percentage of African-Americans than indicated by the 1995 census.
Since the Antelope Valley is a geographically isolated community in which few individuals seek healthcare or attend school outside the region, the study population is effectively closed. Cases reported to active surveillance with addresses outside the surveillance area of Antelope Valley were excluded.
Failure to identify the same individual in both the school and healthcare ascertainment sources will lead to an overestimation of varicella cases based on the assumptions of capture-recapture methods. Since demographic information for each case included the name, street address, city, zip code, phone, gender, race/ethnicity, and date of birth, this information helped to minimize this problem. A computer program designed by project staff first investigated potential duplicates by performing an exact comparison of the date of birth and the first five digits of the street address. Next, surnames were inspected for minor differences in spelling and syllables and the demographic data of any close matches were then compared.
Misdiagnosis of varicella disease occurs when individuals are classified as having varicella when they do not (false positive) or when individuals that have varicella go undetected (false negative). If both types of misdiagnoses occur equally, the capture-recapture result from two ascertainment sources is unaffected. A preponderance of false positive diagnoses will, however, cause an overestimation of the number of varicella cases, and therefore bias varicella incidence higher than the true incidence. Because varicella is easily diagnosed by the lay public, information on disease incidence can be collected from household-based surveysBespecially if the recall period is one year or less [21-23].
The year-round school schedule in the Antelope Valley contributes to the homogeneity of capture probabilities of cases reported by the school ascertainment source. This year-round schedule may also contribute more varicella cases compared to regions with traditional school schedules where a lengthy summer break tends to interrupt the transmission of varicella outbreaks. On the other hand, in 1995, the Antelope Valley experienced five consecutive months with a high ambient air temperate in excess of 84°F which tended to reduce transmission of varicella disease relative to other cooler geographic regions of the U.S. Accepting the fact that we might expect some local variation in varicella incidence, we assume incidence rates in the Antelope Valley are reasonably approximated by those national rates reported by NHIS for the U.S.
We did not attempt to compare varicella incidence among children <1 year old nor adults 20 years and older since cases in these age categories are reported by only one ascertainment source--healthcare providers. The ascertainment-corrected incidence in the Antelope Valley for age categories 1 to 4, 5 to 9, and 10 to 14 were all lower than the average incidence reported by NHIS. This may be due to the fact that schools often require a healthcare provider=s written notice to excuse an absent student, thereby creating a positive dependence of school and healthcare provider ascertainment sources. Under this scenario, the capture-recapture estimate represents a likely minimum .
The ascertainment-corrected incidence in the Antelope Valley for the 15 to 19 year olds was higher than that given by NHIS. This may be due to the fact that many students ages 18 and 19 have graduated high school and are no longer found in the school ascertainment sourceBthus creating heterogeneity of probabilities. Alternatively, varicella incidence in the Antelope Valley could have been actually higher than the average annual incidence rates reported by NHIS.
The close agreement between the estimated incidence when stratified and unstratified suggests the assumption of uniform reporting probabilities is plausible. It is likely that the high percentage of schools in the Antelope Valley community on a year-round schedule contributed to reduced seasonal variation that may have otherwise been more prevalent in communities with traditional school years (and summer break).
Cases reported via active surveillance aged 1 to 19 years reflect under-reporting in excess of 50%. Therefore, application of capture-recapture, despite limitations on accuracy, is reasonable on the basis that (1) the ascertainment-corrected incidence of 50.9 cases/1,000 obtained via Active Surveillance in the Antelope Valley in 1995 differed from NHIS average annual U.S. incidence by -4.2%, (2) the capture-recapture estimate based solely on verified varicella cases stratified by severity of disease and by grading of lesions yielded incidence rates that differed from NHIS by -5.2% and -10.7%, respectively, and (3) some local variation in annual varicella incidence is expected from the NHIS average annual U.S. figure.
The capture-recapture methods assume the population is closed (i.e. no migration) and that there was 100% data linkage in matching dual reported cases. Additionally, it is assumed that the two ascertainment sources were generally independent of one another and all individuals had the same probability of being included within the list. It is impossible to perfectly satisfy all these assumptions as well as quantify the effect of the dependence of the two ascertainment sources. Because it is the policy of schools to require a written notice from a physician to excuse a child=s absence, it is suspected that schools and healthcare providers are positively dependent, with the effect of increasing the number of case matches and thus leading to an underestimate of varicella incidence . This positive dependence is also suggested by the fact that the capture-recapture estimates of annual incidence were generally lower than that reported by NHIS. Only with the availability of a third ascertainment source could dependencies begin to be reliably and quantitatively assessed and corrected.
The reported varicella cases were not verified by laboratory or serologic testing and therefore some false positives may have occurred due to differential diagnoses. Likewise, mild cases that did not come to the attention of the ascertainment sources would contribute to false negatives and therefore would not be included in the capture-recapture estimate.
Capture-recapture analysis is a useful tool to evaluate reporting completeness of cases ascertained via active surveillance. It can be used in sequential years of surveillance to distinguish whether trends in varicella annual incidence are decreasing due to varicella vaccination, or whether decreasing case reports simply reflect poorer reporting compliance of the surveillance sites. The close agreement between the capture-recapture estimate and the NHIS estimate imply that despite limitations on accuracy, ascertainment-corrected incidence rates can be utilized for common communicable diseases such as varicella. Capture-recapture methods will be utilized more widely in the future as it is seen that the inherent assumptions are met.
We wish to thank each of the 300 surveillance sites, including school principals and nurses, as well as physicians and other healthcare professionals in the Antelope Valley community for their faithful and continued reporting of varicella cases to this project. The Varicella Active Surveillance Project under the Los Angeles County Department of Health Services (LACDHS), Acute Communicable Disease Control unit, is supported by a grant provided by the Centers for Disease Control and Prevention (CDC, Atlanta Georgia).
 Stephen C. Capture-recapture methods in epidemiological studies. Infect Control Hosp Epidemiol 1996; 17(4):262-266.
 Sutter RW, Cochi SL. Pertussis hospitalizations and mortality in the United States, 1985-1988. JAMA, 1992; 267:386-391.
 Sutter RW, Cochi SL, Brink EW, Sirotkin BI. Assessment of vital statistics and surveillance data for monitoring tetanus mortality, United States, 1979-1984. Am J Epidemiology, 1990; 131:132-142.
 Rubin G, Umbach D, Shyu SF, Castillo-Chavez C. Using mark-recapture methodology to estimate the size of a population at risk for sexually transmitted diseases. Statistics in Medicine, 1992; 11:1533-1549.
 Modesitt SK, Julman S, Fleming D. Evaluation of active versus passive AIDS surveillance in Oregon. Am J Public Health, 1990; 80:463-464.
 Hardy AM, Starcher ET, Morgan WM, et. al. Review of death certificates to assess completeness of AIDS case reporting. Public Health Rep, 1987; 102:386-391.
 Davis SF, Strebel PM, Atkinson WL, Markowitz LE, Sutter RW, Scanlon KS, Friedman S, Jadler SC. Reporting efficiency during a measles outbreak in New York City, 1991. Am J Public Health, 1993; 83:1011-1015.
 McGilchrist CA, McDonnell LF, Jorm LR, Patel MS. Loglinear models using capture-recapture methods to estimate the size of a measles epidemic. Journal of Clinical Epidemiology 1996; 49(3):293-296.
 Hubert B, Descenclos JC. Evaluation of the exhaustivness and representativeness of a surveillance system using the capture-recapture method. Application of the surveillance of meningococcal infections in France in 1989 and 1990. Rev Epidmiol Sante Publique 1993; 41(3):241-249.
 Ackman DM, Birkhead G, Flynn M. Assessment of surveillance for meningococcal disease in New York State, 1991. AJE 1996; 144(1):78-82.
 Abeni DD, Brancato G, Perucci CA. Capture-recapture to estimate the size of the population with human immunodeficiency virus type 1 infection. Epidemiology 1994; 5(4):410-414.
 Dromer F, Matholulin S. Dupont B, Laporte A. Epidemiology of cryptococcosis in France: a 9-year survey (1985-1993). Clinical Infectious Disease 1996; 23(1):82-90.
 Deming WE. Out of crisis. 1991 MIT Center for Advanced Engineering Study. Cambridge, MA.
 Thacker, SB, Berkelman RL. Public health surveillance in the United States. Epidemiol Rev 1988; 10:164-1988.
 McCarty DJ, Tull ES, Moy CS, Kwoh CK, LaPorte RE. Ascertainment Corrected Rates: Applications of Capture-Recapture Methods. International J of Epidemiology, 1993; 22(3):559-565.
 Hook EB, Regal RR. The value of capture-recapture methods even for apparent exhaustive surveys. The need for adjustment for source of ascertainment intersection in attempted complete prevalence studies. Am J Epidemiol 1992; 135:1060-1067.
 Cormack RM. Log-linear models for capture-recapture. Biometrics 1989; 45:395-413.
 Hilsenbeck SG, Kurucz C, Duncan RC. Estimation of Completeness and Adjustment of Age-Specific and Age-Standardized Incidence Rates. Biometrics 1992; 48:1249-1262.
 Massey JT, Moore TF, Parsons VL, Tadros W. Design and Estimation for the National Health Interview Survey, 1985-94. National Center for Health Statistics. Vital Health Stat 2(110). 1989.
 Regal RR, Hook EB. Goodness-of-fit based confidence intervals for estimates of the size of a closed population; Stat. Med. 1984; 3:287-291.
 Guess HA, Broughton DD, Melton Jr. L. III, et al. Population-based studies of varicella complications; Pediatrics, 1986; 78:S723-7.
 Finger R, Hughes JP, Meade BJ, et al. Age-specific incidence of chickenpox. Public Health Rep., 1994; 109(6):750-5.
 Yawn PB, Yawn RA, Lydick E. Community impact of childhood varicella infections. J. Pediatr., 1997; 130:759-65.
 Hook EB, Regal RR. Capture-recapture methods (Letter). The Lancet 1992; 339:742.
Table 1. Verified and Probable Varicella Cases Stratified by Age Category reported via Active Surveillance
195% confidence intervals are based on goodness-of-fit.
2Difference between the capture-recapture estimate based on ages 1 to 19 unstratified vs. stratified is 59 cases (4498 - 4439), or 1.3%.
Table 2. Comparison of Annual Varicella Incidence (Cases/1,000) Stratified by Age Category
1Based on 1995 Revised Estimates projected from 1990 Census
2NHIS 1990-1994 for the U.S.
3Based on verified and probable cases reported by healthcare providers and daycares, preschools, and schools.
Table 3. Capture-Recapture Estimates Stratified by Quarter for Individuals Aged 1 to 19 Years
1Or 50.9 cases/1,000; differs from NHIS by -4.2%
Table 4. Capture-Recapture Estimates Stratified by Disease Severity Index and Age based on Verified Varicella Cases Only
Table 5. Capture-Recapture Estimates Stratified by Lesion Grading based on Verified Varicella Cases Only
1Or 47.5 cases/1,000; differs from NHIS by -10.7%
Table 6. Capture-Recapture Estimates based on Verified Varicella Cases in 5- to 18-Year-olds by Year, 1995 to 2001.
1Sum of estimated cases stratified by year is 9,071 differs from estimated cumulative 9,119 cases by 48 (0.5%).
The incidence of zoster after immunization with live
attenuated varicella vaccine. A study in children with leukemia. Varicella
Vaccine Collaborative Study Group