Methods

SettingThe DCC operated on weekdays from 6:30 a.m to 5:30 p.m Over half of the children attended ⩾40 h per week, and 17 children attended a part-time program before and after school

Case definitionsA case of varicella was defined as a pruritic maculopapulovesicular rash with no other apparent cause in a child attending the DCC, with onset between 24 December 1999 and 9 April 2000. Recognizing that vaccinees may have an atypical rash with few or no vesicular lesions, we distinguished between varicella cases among unvaccinated children and breakthrough cases among children vaccinated >42 days before rash onset [9]. Diagnosis relied on physician or parental report of illness, without laboratory confirmation. Varicella cases were classified as mild if the child had <50 lesions without systemic complaints, such as fever or malaise; as moderate if there were 51–500 lesions, with or without systemic complaints; or as severe if there were >500 lesions or serious complications, such as varicella pneumonitis, encephalitis, hemorrhagic complications, or bacterial superinfection

Outbreak investigationTo control the outbreak, letters were sent to parents or guardians, advising them to have any susceptible child vaccinated against varicella. A self-administered questionnaire was distributed to the parent or guardian of each child to learn date of birth, sex, attendance (hours per week), health conditions, medications, history of varicella or varicella vaccination, receipt of other vaccines concurrently with varicella vaccine, exposure to varicella-zoster virus in the previous 7 months, and information on susceptibility of other household members. Parents of children who had developed a varicella-like rash since 1 October 1999 were asked to provide detailed information about the illness, including dates of onset and resolution, rash description, resultant complications, and exposures to varicella-zoster virus outside the DCC. If parents consented, medical and vaccination history were verified with their child’s health care providers. To calculate the time since vaccination, the midpoint of the outbreak was used for the age of children without varicella. Data were entered into Epi Info 6.04 (CDC) and were analyzed with SAS Institute software (release 8.0). Differences between proportions were tested by χ² statistics unless any cell had an expected value <5, in which case we used Fisher’s exact test; differences between medians were tested by the Wilcoxon&amp;rank sum test. All P values are 2-sided

VE and 95% confidence intervals (CIs) were calculated via the cohort method [10]. We calculated the attack rates in unvaccinated (ARU) and vaccinated (ARV) children ⩾12 months old who lacked a history of varicella as follows: VE%=[(ARU-ARV)/ARU]×100. VE was calculated for all forms of varicella and for moderate or severe varicella. Children who attended the DCC for <1 week during the study period were excluded, as was one child born after the start of the outbreak

Some children were vaccinated during the outbreak period. Vaccination status was classified according to the child’s status during most of the outbreak. Children were considered to be unvaccinated for 14 days after vaccination and were considered to be vaccinated for >42 days after vaccination. We excluded from analysis 4 children vaccinated early in the outbreak (on 2, 16, 28, and 29 December) who could not be reliably classified because they could have developed immunity at any time 2–6 weeks after vaccination. To account for the possibility that a child who developed rash within one incubation of a sibling’s rash represented household transmission, we recalculated VE on the assumption that infection occurred outside the DCC

Results

DCCIn all, 139 children attended the DCC ⩾1 week during the study period. Although children were divided into 7 classes, they shared a common bathroom, played together outside, and occupied classrooms connected by open doors or doorways. All 21 staff members at the DCC reported a history of varicella

Epidemiologic investigationParents returned 135 (97%) of 139 questionnaires. Health care provider records were obtained for 133 children (99%). Four children (3%) vaccinated early in the outbreak were excluded, because their immunity status for most of the outbreak could not be determined. The median age of the 131 eligible children was 3.7 years (range, 3.2 months–11 years); 77 (59%) were boys. Sixteen infants (12%) <12 months old were too young to be vaccinated (none had a history of varicella). Of the 115 children ⩾12 months old, 20 (17%) were not vaccinated and had no history of varicella, 15 (13%) had a history of varicella, and 80 (70%) had received varicella vaccine. Vaccination coverage among susceptible children ⩾12 months old at the start of the outbreak was 80%

Varicella casesForty-one cases of varicella were reported. The index case patient was an unvaccinated 4.5-year-old child with rash onset on 24 December 1999, and the last case patient developed rash on 23 February 2000 (figure 1). The median age of case patients was 4.2 years (range, 3.9 months–10.3 years); 25 (61%) were boys. Seventeen cases (41%) occurred in unvaccinated children, 14 (34%) in vaccinated children, 9 (22%) in infants, and 1 (2%) in a child with a history of varicella. The median interval between vaccination and rash onset was 3.2 years (range, 1.5–4.3 years)

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Dates of rash onset among varicella case patients by vaccination and prior disease status, varicella outbreak, Reading, PA, December 1999–February 2000

Varicella severityAmong case patients, vaccinated children had fewer lesions, missed fewer days of day care, and were more likely to be described as “not very sick” than unvaccinated children. Vaccinated children were 4.5 (95% CI, 1.5–13.7) times more likely to develop mild varicella than unvaccinated children (table 1); however, 21% of vaccinated case patients had varicella as severe as that of the typical unvaccinated child. No child required hospitalization

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Characteristics of varicella case patients in a day-care center outbreak, Reading, PA, December 1999–February 2000

Risk factors for breakthrough diseaseAlthough it did not achieve statistical significance, younger age at vaccination was associated with an increased risk of breakthrough disease. Children were vaccinated at ages 359–1848 days (median, 385 days). Two children were vaccinated during the week before their first birthday. Children vaccinated at <14 months of age were 3 times as likely as children vaccinated at ⩾14 months of age to develop breakthrough disease (11/44 vs. 3/36; relative risk, 3.0; 95% CI, 0.9–9.9). No other risk factors for vaccine failure were identified, including time since vaccination, vaccination provider, vaccine lot number, underlying diseases, or medications

VEAttack rates among vaccinated and unvaccinated susceptible attendees ⩾12 months old were 18% (14/80) and 85% (17/20), respectively. The attack rate in infants <12 months old was 56% (9/16). VE against all varicella among DCC attendees ⩾12 months old during the outbreak was 79% (95% CI, 66%–88%) and against moderate or severe varicella was 95% (95% CI, 84%–98%). Five unvaccinated children (1 infant and 4 children >12 months) developed moderate or severe varicella 14–16 days after a sibling’s rash onset and were thus more likely to have contracted varicella at home than in the DCC outbreak. Their exclusion did not alter the VE estimates

Discussion

Despite a vaccination coverage rate of 80% in vaccine-eligible children, a 2-month-long outbreak of 41 cases of varicella occurred in a DCC with ∼140 children, with 42% of cases occurring among susceptible vaccine-eligible children and 22% among infants too young to be vaccinated. One-third of cases occurred among vaccinated children, resulting in a VE estimate of 79% against all varicella and 95% against moderate or severe varicella, similar to pre- and postlicensure estimates [3 –8]. Cases among vaccinated children were milder and resulted in less absenteeism than cases among unvaccinated children, although 21% of vaccinated children developed varicella of the same severity as that of the average unvaccinated child

We found a trend toward increased risk of breakthrough disease with younger age at vaccination. Children vaccinated at <14 months of age had a 3 times higher risk of breakthrough than children vaccinated at ⩾14 months of age. In part because of small numbers, this finding did not achieve statistical significance. Data from a prelicensure trial of varicella vaccine indicated that infants and young children may have maternal antibody to varicella after their first birthday. [11] White et al. [11] reported that 47% of children at age 12 months had detectable antibody; by age 16 months, only 4% of children had detectable antibodies

The optimal timing of vaccination depends on the earliest age at which high rates of seroconversion can be achieved, while ensuring that children are adequately protected before they are likely to be exposed to and contract disease. With measles vaccination, the most significant risk factor for vaccine failure early in the vaccination program was found to be the presence of maternal antibody. Prelicensure data suggesting that maternal measles antibodies rarely persist beyond 7 months of age [12] led to a recommendation for routine vaccination at age 9 months when measles vaccine was licensed in 1963. Because of vaccine failures, the recommended age for vaccination was raised to 12 months in 1965 and to 15 months in 1976. The recommended age was only lowered to 12–15 months in 1994 [12], 31 years after licensure. By this time, most US women giving birth had either vaccine-derived immunity or natural immunity that had not been repeatedly boosted by exposure to wild-type disease, resulting in lower levels of measles antibody transferred to the infant [12]

The relative importance of maternally acquired varicella antibody and the infant’s evolving cell-mediated immunity remains to be fully elucidated. Epidemiologic data support the notion that maternal antibody provides protection against varicella early in life. During the first year of life, maternally acquired varicella antibodies wane, and the infant’s cell-mediated immune system matures. A higher rate of vaccine failure in children vaccinated soon after their first birthday might be a consequence of either or both factors

This investigation had several limitations. We relied on parental history of varicella for diagnosis, because medical providers did not evaluate most children reported to have varicella. Given the high awareness of the outbreak among parents and widespread concern about vaccine failure, some nonvaricella rashes may have been misclassified as breakthrough illness. Conversely, some mild cases of rash may not have been detected and reported. Finally, we were unable to determine the risk of vaccine failure stratified by age at vaccination

This investigation demonstrated that sizeable and disruptive outbreaks can occur in moderately to highly vaccinated populations. Among children ⩾12 months old, natural and vaccine-derived immunity of 83% in the DCC proved to be insufficient to prevent the outbreak from occurring and propagating. Since infants <12 months old are ineligible for vaccination, they can only be protected from contracting disease by preventing transmission from older children through vaccination. Considerably higher levels of immunity and improved VE among older children may be required to protect infants indirectly. In addition, our investigation raises the possibility that children vaccinated near their first birthday may not be as adequately protected as children vaccinated at age ⩾14 months. Further study is needed to confirm these findings and to define the optimum age for routine childhood immunization against varicella