Widespread use of varicella vaccine in the United States could enable detection of rare adverse events not identified previously. We reviewed data from 1995 to 2005 from the Vaccine Adverse Event Reporting System, including data from laboratory analyses, to distinguish adverse events associated with wild-type varicella-zoster virus (VZV) versus those associated with vaccine strain. Almost 48 million doses of varicella vaccine were distributed between 1995 and 2005. There were 25,306 adverse events reported (52.7/100,000 doses distributed); 5.0% were classified as serious (2.6/100,000 doses distributed). Adverse events associated with evidence of vaccine-strain VZV included meningitis in patients with concurrent herpes zoster. Patients with genetic predispositions may rarely have disease triggered by receipt of varicella vaccine. Overall, serious adverse events reported after varicella vaccination continue to be rare and must be considered relative to the substantial benefits of varicella vaccination. Ongoing safety surveillance and further studies may shed light on some of the hypothesized associations.
The United States initiated a routine childhood varicella vaccination program in 1995 [1–3]. Varicella immunization has reduced substantially the varicella disease burden in the United States [4–6]. No serious adverse event (SAE) associated with administration of varicella vaccine was documented in prelicensure trials [7–9]. A review of reported postlicensure data from 1995 to 1998 found that SAE reports after varicella vaccination were rare (2.9 reports/100,000 doses distributed) [10–12]. These SAEs included ophthalmic herpes zoster (HZ), pneumonia, encephalitis, thrombocytopenia, vasculitis, and hepatitis. One patient with ophthalmic HZ and a severely immunodeficient patient with pneumonia had laboratory confirmation of an association with vaccine-strain varicella-zoster virus (VZV). Subsequent case reports also have confirmed vaccine-strain VZV associated with hepatitis  and severe rash [14–16]; most of these children had serious preexisting medical conditions that were not recognized at the time of vaccination. In addition, less-serious adverse events confirmed to have been due to vaccine-strain VZV include postvaccination rash with pharyngitis, HZ, and secondary transmission to contacts of vaccinees [10, 11].
Use of varicella vaccine in the United States has increased 5-fold since 1998; thus, additional rare adverse events now could be detectable. We used data from the Vaccine Adverse Event Reporting System (VAERS) and data on vaccine-associated adverse events investigated at the National VZV Laboratory at the Centers for Disease Control and Prevention (CDC; Atlanta) to summarize the safety of varicella vaccine during the first decade of its use in the United States.
VAERS is a passive surveillance system managed by the CDC and the US Food and Drug Administration (FDA). It relies on reports from health care providers, vaccine manufacturers, and the public about adverse events temporally related to immunization . To avoid the double counting of follow-up reports, our analysis primarily used the initial submission to VAERS for each patient. For selected cases, we also integrated information from follow-up or other sources.
On the basis of the signs and symptoms of the adverse events, reports are coded by use of the FDA's Coding Symbols for Thesaurus of Adverse Reaction Terms (COSTART) . The Code of Federal Regulations defines SAEs as those resulting in hospitalization, death, life-threatening illness, permanent disability, or certain other medically important conditions . Reports of SAEs are followed up to obtain clinical information, including medical records.
We searched VAERS data for all reports associated with varicella vaccination in the United States from 23 May 1995 through 31 December 2005. Some COSTART codes were grouped (e.g., ataxia was combined with cerebellar ataxia and maculopapular, pustular, and vesicular rashes were combined as “rash”). We also searched text fields for “herpes zoster.” Reporting rates use the number of vaccine doses distributed in the United States (CDC Biologics Surveillance, 1995–2005, unpublished data). We reviewed the complete VAERS report of all deaths and the SAE reports of HZ and meningitis, including available laboratory-testing information from the Merck-Columbia University Varicella-Zoster Virus Identification Program (New York) that distinguished vaccine strains from wildtype strains in virus isolates.
More recently, the CDC National VZV Laboratory also conducted analyses of virus isolates. We analyzed data on specimens from patients with suspected varicella vaccine-associated adverse events that were tested by the CDC National VZV Laboratory between May 2001 and May 2006. At the CDC's laboratory, specimens were processed to isolate VZV DNA and were genotyped (wild-type versus vaccine strain) by use of realtime polymerase chain reaction (PCR) techniques (4 methods targeting vaccine markers in open reading frames 38, 54, and 62) . We used SAS (version 9.01; SAS Institute) and Microsoft Excel (version 2003) for the data analysis.
Through 31 December 2005, 47,733,950 doses of varicella vaccine were distributed in the United States, and 25,306 adverse events after varicella vaccination were reported to VAERS (reporting rate, 52.7/100,000 doses distributed). The reporting rate declined over time, from 245.0/100,000 doses in 1995 to 20.8/100,000 doses in 2005. Of all the reports, 1276 (5.0%) were classified as serious (2.6/100,000 doses). The reporting rate for the SAE subset also declined, from 5.8/100,000 doses in 1995 to 1.4/100,000 doses in 2005 (figure 1).
More than one-half of all reports (14,780, or 58.4%) described administration of varicella vaccine alone. Among children 12–23 months of age, 7308 reports (66.6%) followed varicella vaccine administered in combination with other vaccines; 5627 (77.0%) of these reported vaccination in combination with measles-mumps-rubella (MMR) vaccine. Except among adolescents 10–17 years of age, higher proportions of reports related to varicella vaccine administered in combination with other vaccines were classified as serious, compared with the proportions of reports related to varicella vaccine administered alone (table 1).
The most commonly reported adverse event was rash (8262 reports; 17.3/100,000 doses), accounting for 32.6% of all reports; 5288 (64.0%) of these were related to varicella vaccine administered alone (table 2). Fever (5451 reports; 11.4/100,000 doses) and injection-site reactions (3291 reports; 6.9/100,000 doses) accounted for 21.5% and 13.0% of reports, respectively. Two-thirds of reported fevers began within 14 days (peak 0–6 days) after vaccination, and 60.0% of reported injection-site reactions appeared within 2 days after vaccination. There were 1047 reports of urticaria (2.2/100,000 doses; 70.2% of cases within 3 days after vaccination), and two-thirds of these urticaria cases occurred after receipt of varicella vaccine in combination with other vaccines.
Of 852 reports of convulsions (1.8/100,000 doses), febrile seizures accounted for 38.2%. Overall, 79.9% of reported convulsions occurred in children 12–23 months of age, and 80.7% of these followed varicella vaccine given in combination with other vaccines. Cellulitis, otitis media, diarrhea, and thrombocytopenia were reported to have occurred more commonly among children 12–23 months of age who had received varicella vaccine in combination with other vaccines, whereas arthralgia was more common in older age groups, with 70.0% of reported cases following receipt of varicella vaccine alone.
Deaths. There were 60 deaths reported to VAERS during the study period (reporting rate, 0.1/100,000 doses); 23 deaths (38.3%) followed administration of varicella vaccine alone (table 3). The median interval between vaccination and death was 9 days (range, <1 day to 6.8 years). The most common fatal events reported were septicemia and multiorgan failure (11 [18.3%] of 60 deaths), all in individuals with severe congenital anomalies or with disorders that affect the immune system, and 10 deaths (16.7%) were reported as “crib deaths.” Six of the 60 deaths have been described elsewhere [10, 14]. Vaccine-strain VZV was described as a contributing factor in the death of a child with NK-cell deficiency . In addition to the 2 deaths associated with wild-type VZV in vaccinated children, described by Wise et al. , there was 1 death associated with wild-type VZV in a 9-year-old child who had been vaccinated at 12 months of age and then developed glomerulonephritis 3-4 years later. One week before the onset of rash, the child had received high doses of steroids. Complications of severe breakthrough varicella disease included hepatitis and hepatic necrosis. Two healthy children 3 and 8 years of age developed ataxia, encephalopathy, and elevated plasma ammonia levels, with the onset of the first symptom 14 and 30 days after vaccination, respectively; ornithine transcarbamylase deficiency was diagnosed later. One of these children died 1 month after the diagnosis, and the other recovered from the initial episode but died 6 years later from complications related to ornithine transcarbamylase deficiency. There also was a child who died 2 months after receiving a diagnosis of familial hemophagocytic lymphohistiocytosis (HLH) after varicella vaccination. An autopsy was not performed. The child's older sibling had died 1 year previously after HLH had been diagnosed after varicella vaccination.
Meningitis. Of 39 reports of meningitis (reporting rate, 0.1/100,000 doses), 33 (84.6%) occurred in children ⩽12 years of age. Of these 33 children, 10 (30.3%) had confirmed bacterial meningitis due to Neisseria meningitidis (4 children) or Streptococcus pneumoniae (6 children); 10 (30.3%) had concomitant HZ, 9 of whom did not have any underlying conditions; and 4 (12.1%) were reported in association with clinical illness consistent with breakthrough varicella, 2 of whom had cerebral spinal fluid (CSF) positive for VZV by PCR (1 described elsewhere ).
Of the 10 patients with HZ and meningitis, 2 were confirmed to have vaccine-strain VZV in CSF: (1) A 4-year-old child undergoing chemotherapy for acute lymphocytic leukemia developed HZ affecting the C6-C7 dermatome and then meningitis (interval between HZ and meningitis not stated). The child had received varicella vaccine 19 months previously, while healthy. Vaccine-strain VZV also was detected in a specimen from the HZ lesions. After receiving acyclovir intravenously, the child made a full recovery. (2) A healthy 4-year-old child received varicella vaccine 2 years and 8 months before developing an HZ rash on the right arm, followed by meningitis 3 days later. DNA extracted from CSF was sent from an external laboratory to the CDC for strain typing, and vaccine-strain VZV was confirmed. A specimen from the HZ lesions was confirmed for VZV by PCR, but genotyping was not done. The child made a full recovery after receiving acyclovir intravenously.
Of the other 8 patients with HZ and meningitis, 4 had VZV confirmed in both CSF and skin lesions (but genotyping was not done); 2 had vaccine-strain VZV in skin lesions and no VZV in CSF (but 1 had enterovirus in CSF); and 1 had Tzanck-positive skin lesions but testing of CSF was not done. Laboratory test results were not available for 1 patient. Onset intervals between HZ rash and meningitis ranged between 2 and 10 days.
Of the 6 adults with meningitis, 2 developed rashes (one rash described as “blotchy” and the other as vesicular and papular) 4 and 13 days after vaccination, respectively. Another report described aseptic meningitis after the first and second doses of varicella vaccine, administered ∼5 years apart, at intervals of 3 months and 4–5 weeks after vaccination, respectively.
HZ. Of 981 reports coded for HZ (reporting rate, 2.1/ 100,000 doses), 52 (5.3%) were classified as SAEs, and 47 patients were hospitalized, including the 10 patients with meningitis described above. The median age of hospitalized patients with HZ was 2.5 years (range, 12 months to 12 years), and the median interval between vaccination and onset of rash was 7.3 months (range, 3 days to 4.3 years). Seven (13%) of the 52 cases of serious HZ were laboratory confirmed to be due to wild-type VZV; 10 (19%) were confirmed to be due to vaccine-strain VZV (including the 2 meningitis cases due to vaccine strain VZV described above); 11 (21%) were laboratory confirmed to be due to VZV, but genotyping was not done; 1 was diagnosed as herpes simplex virus (HSV) infection; and 1 was due to an allergic reaction. For the remaining cases, the available laboratory data were incomplete. Among hospitalized patients with HZ, the most frequently reported rash location was the face (information on rash location was available for 21 of 43 patients). Of these patients, 2 had a diagnosis of Ramsay Hunt syndrome, with lesions confirmed to be due to VZV by PCR (no genotyping results), and 10 were reported to have ophthalmic HZ. The lesions of 2 patients with ophthalmic HZ were identified to be due to vaccine-strain VZV, and those of 2 patients were reported to be due to wild-type VZV; specimens from the other patients either were not genotyped or were inadequate for testing.
Data from the National VZV Laboratory
From May 2001 through May 2006, the National VZV Laboratory at the CDC investigated 338 specimens from patients with suspected adverse events after varicella vaccination (table 4). Among the 209 submitted specimens obtained from varicella-like rashes, wild-type VZV was detected in 55 specimens (26.3%), and vaccine-strain VZV was detected in 37 specimens (17.7%); the remaining specimens tested negative for VZV or were inadequate for testing. The median age of patients with adverse events confirmed to be associated with vaccine-strain VZV was 1 year (range, 1–53 years), and the median interval between vaccination and the onset of rash was 23 days (range, 6–43 days). Three cases of vaccine-strain-associated varicella disease resulted from secondary transmission; 2 were reported elsewhere . The other case was in a healthy father who developed a rash (dates of onset not provided) caused by vaccine- strain VZV after exposure to his 1-year-old child with severe immunodeficiency who, after varicella vaccination, developed disseminated varicella confirmed to be due to vaccine-strain VZV.
Among the 118 specimens from patients with HZ rashes, wild-type VZV was detected in 24 specimens (20.3%), and vaccine-strain VZV was detected in 48 specimens (40.7%; including those from 1 of the 2 children with meningitis described above); the remaining specimens tested negative or were inadequate for testing. The median age of patients with confirmed vaccine-strain-associated HZ was 5 years (range, 1–11 years), and the median interval between vaccination and the onset of rash was 3 years (range, 4 months to 8.5 years).
Extensive postlicensure experience after distribution of nearly 48 million doses of varicella vaccine continues to show an excellent overall safety profile. Use of varicella vaccine has resulted in substantial reductions in varicella-associated mortality, morbidity, and health care costs [4–6]. Consistent with the results of earlier studies, the majority (95%) of adverse events reported were not serious [10, 11]: two-thirds of the reports described fever, rash, or injection-site reactions. Our expanded review of postlicensure data has documented laboratory-confirmed vaccine-strain VZV from the CSF of 2 patients with meningitis and concurrent HZ, 1 of whom was a healthy child. Despite a >80% decline in the national incidence of varicella, wild-type VZV still circulates in the United States and causes a substantial proportion of laboratory-confirmed VZV-associated SAEs.
Meningitis can be a rare manifestation of VZV reactivation and was thought to occur mainly in persons with impaired cellular immunity , although it also has been described to occur in healthy children and adults [24–28]. The significance of wild-type or vaccine-strain VZV in CSF after VZV reactivation, especially when cranial nerves are involved, is unclear. However, the 2 patients with confirmed vaccine-strain-associated meningitis had sufficient neurological symptoms and signs to warrant diagnostic evaluation of CSF. Chiappini and de Martino  have speculated that early acquisition of (wildtype) VZV may predispose to subsequent neurologic complications in association with VZV reactivation. Acquisition of VZV infection in the first year of life is a known risk factor for HZ in childhood . Although available data suggest a lower risk of HZ for vaccinated children, compared with those with previous varicella [13, 30, 31], longer follow-up of vaccine recipients is needed, especially of children vaccinated at 12 months of age. The episodes of meningitis reported after each dose of varicella vaccine administered to an adult may represent positive rechallenge—that is, the occurrence of the same adverse event after a subsequent dose of vaccine—even though neither episode was laboratory confirmed to be due to VZV. However, in this instance, the first episode of meningitis occurred 3 months after vaccination, an interval that does not appear to be biologically plausible unless there was VZV reactivation.
Our data extend previous observations that vaccine-strain VZV is capable of reactivation, causing HZ in healthy vaccine recipients [10–11, 32]. We describe a broader range of disease severity and complications, including hospitalization, from reactivation of vaccine-strain VZV. We could not evaluate whether HZ due to wild-type VZV occurred after breakthrough varicella disease. Ramsay Hunt syndrome complicating HZ caused by wild-type VZV has been documented elsewhere [33, 34]. Genotyping for VZV was not done for either of the 2 vaccinated patients with HZ who also had a diagnosis of Ramsay Hunt syndrome.
Convulsions, cellulitis, otitis media, diarrhea, vomiting/nausea, and pharyngitis all have been described as potentially related to varicella vaccine in pre- and postlicensure reports [8–11]. However, with the possible exception of cellulitis complicating varicella vaccine vesicles, the high background prevalence of these conditions impedes assessment of possible vaccine etiologies, in the absence of laboratory confirmation. It nonetheless remains plausible that varicella vaccine could contribute to some of these illnesses. Thrombocytopenia is a well recognized complication of measles-containing vaccine [35, 36] and also has been reported with varicella vaccine [10, 11, 37]. Allergic reactions, particularly urticaria, were reported relatively often and may reflect a reaction to the gelatin component of the vaccine , although 1 case of recurrent papular urticaria was reported to be linked to an immunologic response to the varicella vaccine . Anaphylaxis has been reported in association with varicella vaccine [10, 11], and the development of a gelatin-free varicella vaccine has been associated with a marked reduction in reported anaphylactic reactions in Japan .
Interpretation of the reported deaths after varicella vaccination requires caution, because most were unlikely to be related to varicella vaccination (except temporally) or had insufficient information provided to evaluate possible association. The potential for vaccine-strain VZV to cause death appears to be remote, and the benefits provided by the vaccine far outweigh such concerns. Vaccine-strain VZV was confirmed as a possible contributing factor in only 1 death, that of a child with NK-cell deficiency . Before the varicella vaccination program in the United States, ∼100 children and adults, most of whom were healthy, died each year from varicella . In contrast, since implementation of the varicella vaccination program, deaths due to varicella have declined substantially. By 2001, varicella as an underlying cause of death had declined by 92% among children 1–4 years of age and by ⩾74% among all age groups <50 years .
Two patients developed hyperammonemia associated with ornithine transcarbamylase deficiency, after vaccination. Viral infections may play a role in triggering ornithine transcarbamylase deficiency, although this relationship has not been reported with wild-type VZV infection [41, 42]. Hyperammonemia can manifest as encephalopathy and ataxia, which are well established, rare complications of varicella disease ; these conditions also have been reported after varicella vaccination [10, 11, 44]. The case of hemophagocytic lymphohistiocytosis also could be associated with vaccine-strain VZV. A variety of infectious agents can provoke secondary hemophagocytic lymphohistiocytosis and can trigger a familial form of the disorder, including Epstein-Barr virus, cytomegalovirus, VZV, HSV, the adenoviruses, and parvovirus B19 [45–47]. Further investigations are needed to confirm the role of vaccine-strain VZV in the activation of diseases linked with genetic proclivity. These diseases are rare, however, and would be difficult to detect during prevaccination screening. Nevertheless, health care providers should pay close attention to varicella vaccination contraindications, including “family history of congenital or hereditary immunodeficiency, unless the immune competence of the potential vaccine recipient is demonstrated” [48, p. 6].
Passive surveillance systems such as VAERS have important inherent limitations, including underreporting [17, 49]. VAERS is more likely to capture events that follow soon after vaccination rather than later events . Overreporting also is a factor, since many reports may describe events related in time but likely to be caused by confounding factors, especially diseases and medications. The majority of reported adverse events occurred after coadministration of varicella vaccine with other vaccines, primarily in the 12–23-month age group. Thus, the potential association of varicella vaccine with the reported adverse event often is difficult to evaluate. In VAERS, most reports are not verified, lack consistent diagnostic criteria, and provide limited clinical and laboratory information. An earlier study of VAERS data reviewed many reports individually, including follow-up information . Because of the current accumulation of substantially larger numbers, we did not evaluate most individual case reports, and we tabulated most frequencies on the basis of initial submission coding. One consequence is that misclassifications of HZ or other events may affect our data (e.g., invalid codes and missed codes), but we do not believe that such misclassifications would influence the basic conclusions. Although the earlier study provided details on a large number of types of reported adverse events, we focused on selected SAEs. Overall reporting rates and those for the SAE subset were similar to or lower than those described by Wise et al. . This decline may reflect a waning Weber effect (greater reporting enthusiasm for recently introduced products ). Other limitations of our study include the fact that data on the numbers of doses distributed are not available by age and sex, for calculation of age- and sex-specific reporting rates. We also lack denominator data on the administration of multiple versus single vaccines, for comparison of reporting rates of adverse events after the receipt of varicella vaccine alone versus when given in combination with other vaccines.
With the extensive use of varicella vaccine in the United States for more than a decade, surveillance of vaccine safety continues to portray the occurrence of SAEs after varicella vaccination as rare; these risks must be considered relative to the substantial benefits of varicella vaccination [4–6]. Physicians are reminded to screen potential vaccine recipients for immunocompromising conditions, to report suspected adverse events to VAERS, and to send virus specimens to the CDC National VZV Laboratory (NationalVZVlab@cdc.gov) or the Merck-Columbia University Varicella-Zoster Virus Identification Program, for laboratory testing and genotyping. Careful surveillance and analytical studies are needed to investigate the hypothesized risks of varicella vaccination. In September 2005, the FDA licensed a quadrivalent MMR-varicella vaccine for use with children 12 months to 12 years of age . The titer of the VZV component in this new vaccine for the prevention of chickenpox is higher than that in the original varicella vaccine. Continued surveillance will be important in monitoring the safety of both products.
We are thankful for the contribution of the following: Susanne Pickering, Elaine Miller, Beth Hibbs, Scott Campbell, Claudia Vellozzi, John Iskander, and Robert Davis (Immunization Safety Office, Centers for Disease Control and Prevention, Atlanta); Martha Bryant-Genevier, Dale R. Burwen, Soju Chang, Miles Braun, AnnW. McMahon, JaneWoo, Philip Perucci, Manette Niu, Sukhminder Sandhu, and Robert Ball (Center for Biologics Evaluation and Research, US Food and Drug Administration, Rockville, MD); and Robert Weibel, Vito Caserta, and Geoffrey Evans (Health Resources and Services Administration, Rockville, MD). We also thank the staff of Constella Group, Inc. (Durham, NC).M
Supplement sponsorship. This article was published as part of a supplement entitled “Varicella Vaccine in the United States: A Decade of Prevention and the Way Forward,” sponsored by the Research Foundation for Microbial Diseases of Osaka University, GlaxoSmithKline Biologicals, the Sabin Vaccine Institute, the Centers for Disease Control and Prevention, and the March of Dimes.