Postlicensure Safety Surveillance for Varicella Vaccine

Context Since its licensure in 1995, the extensive use of varicella vaccine and close surveillance of the associated anecdotal reports of suspected adverse effects provide the opportunity to detect potential risks not observed before licensure because of the relatively small sample size and other limitations of clinical trials.

Objectives To detect potential hazards, including rare events, associated with varicella vaccine, and to assess case reports for clinical and epidemiological implications.

Design and Setting Postlicensure case-series study of suspected vaccine adverse events reported to the US Vaccine Adverse Event Reporting System (VAERS) from March 17, 1995, through July 25, 1998.

Main Outcome Measures Numbers of reported adverse events, proportions, and reporting rates (reports per 100,000 doses distributed).

Results VAERS received 6574 case reports of adverse events in recipients of varicella vaccine, a rate of 67.5 reports per 100,000 doses sold. Approximately 4% of reports described serious adverse events, including 14 deaths. The most frequently reported adverse events were rashes, possible vaccine failures, and injection site reactions. Misinterpretation of varicella serology after vaccination appeared to account for 17% of reports of possible vaccine failures. Among 251 patients with herpes zoster, 14 had the vaccine strain of varicella zoster virus (VZV), while 12 had the wild-type virus. None of 30 anaphylaxis cases was fatal. An immunodeficient patient with pneumonia had the vaccine strain of VZV in a lung biopsy. Pregnant women occasionally received varicella vaccine through confusion with varicella zoster immunoglobulin. Although the role of varicella vaccine remained unproven in most serious adverse event reports, there were a few positive rechallenge reports and consistency of many cases with syndromes recognized as complications of natural varicella.

Conclusion Most of the reported adverse events associated with varicella vaccine are minor, and serious risks appear to be rare. We could not confirm a vaccine etiology for most of the reported serious events; several will require further study to clarify whether varicella vaccine plays a role. Education is needed to ensure appropriate use of varicella serologic assays and to eliminate confusion between varicella vaccine and varicella zoster immunoglobulin.

In March 1995, the Food and Drug Administration (FDA) licensed a live virus varicella vaccine (Varivax; Merck & Co, Inc) to prevent chickenpox. Authoritative recommendations^1-3 for nearly universal use of the new vaccine reflected the magnitude of morbidity and mortality due to varicella: 4000 to 9000 hospitalizations and 100 deaths per year from serious secondary infections or central nervous system and other complications.^4,5 The FDA based licensure on studies of safety and efficacy for varicella prevention in almost 9500 healthy children and more than 1500 adolescents and adults.⁶ Usually well tolerated,^7-12 varicella vaccine protects most recipients against primary chickenpox, and most breakthrough cases are mild.^13-15 Adverse events identified before licensure included injection site pain and erythema and a generalized varicella-like rash in 4% to 6% of vaccinees, usually appearing 5 to 26 days after the first dose.¹⁶

After licensure, with increasing use of a vaccine in larger and more diverse populations, rare risks that previously escaped detection can emerge.^17-20 The fraction of children aged 19 to 35 months vaccinated against varicella rose from 14% in the third quarter of 1996²¹ to 43% in all of 1998.²² Some 9.7 million doses had been sold through July 1998.²³ (This projection also used from unpublished Centers for Disease Prevention and Control [CDC] data for 1996-1997 [Robert Synder, MA, CDC National Immunization Program, oral and written communications, Fall 1998] and earlier information). Therefore, in this article, we review and summarize spontaneously reported adverse events from the first 3 years of postlicensure safety surveillance. By itself, such passive surveillance has important limitations, but we interpret the data within the context of previous studies and other information, including Merck's postmarketing study of acute safety^24,25 and the pregnancy registry for varicella vaccine (telephone: 800-986-8999), which seeks information about pregnancies with conception during the 3 months following receipt of varicella vaccine or with varicella vaccine administration during any trimester.²⁶

These analyses encompass reports received by the Vaccine Adverse Event Reporting System (VAERS) from March 17, 1995, through July 25, 1998. Jointly operated by the FDA and the CDC, VAERS consolidates voluntarily submitted reports of suspected vaccine adverse effects from manufacturers, health care workers, and patients for postlicensure vaccine safety surveillance.^20,27-29 Although the National Childhood Vaccine Injury Act of 1986 (Pub L No 99-660) obliges physicians to submit certain reports, VAERS data are typical of passive drug safety surveillance programs, with case counts that represent unknown but probably highly variable fractions of actual event numbers. Nonetheless, with its national scope and open-ended format, VAERS can reveal potential vaccine safety problems with new vaccines,³⁰ increased numbers for previously reported events,³¹ and potential associations between vaccines and entirely unanticipated events³² that might not have occurred in the relatively small prelicensure clinical trials. (The National Technical Information Service distributes VAERS data [http://www.ntis.gov], and the FDA provides additional information [http://www.fda.gov/cber/vaers/vaers.htm].)

The VAERS report form solicits information regarding the vaccinee (name, date of birth, current age, sex, and address), the adverse event or events (date of onset, therapy, and clinical course), and vaccine or vaccines administered (date, lot identifier, and dose sequence). VAERS adapted standardized coding terms to index each patient's reported events.³³ In selecting syndromes and other subsets for description here, we weighed medical severity, frequency, and potential preventability. Serious cases mainly encompass fatal or life-threatening events, hospitalizations, and permanent disabilities.

Reporting rates are numbers of reports divided by estimated varicella vaccine doses sold.²³ They cannot be interpreted as incidence rates because of potentially substantial underreporting and overreporting. In positive rechallenge cases, an adverse event recurred after a second dose of varicella vaccine. Occasional reports include information from a research laboratory's polymerase chain reaction (PCR) studies of viral DNA, which can confirm the presence of varicella zoster virus (VZV) in an isolate. Restriction fragment length polymorphisms can then often distinguish the Japanese-origin Oka vaccine strain from wild-type VZV circulating in the United States.³⁴ Vaccine failure refers to primary (breakthrough) wild-type VZV infection long enough after vaccination for most subjects to develop protective immunity. Because few patients had confirmatory PCR studies, we classified reports of generalized varicella at least 6 weeks after vaccination as possible vaccine failures.

VAERS received 6574 varicella vaccine reports between March 1995 and July 1998 (Figure 1), an overall rate of 67.5 reports per 100,000 doses distributed, with largest numbers of reports soon after licensure. There were 0.1 deaths, 2.8 other serious adverse events, and 64.5 nonserious adverse events per 100,000 doses. Number of reports by age, sex, and severity are presented in Table 1. Table 2 presents selected adverse events. Approximately 4% of cases (2.9/100,000) were serious, including 14 deaths (Box 1). Among all reports, 82% of patients received only varicella vaccine, 12% also received measles, mumps, and rubella vaccine (MMR), and other vaccine combinations (without MMR) accounted for the balance. For children, the sex ratio was close to 1, but females predominated among reports for teenagers and adults. Patient reports came from every state, and 9 reports originated abroad.

Patient A. Shortly before her eighth birthday, a girl with history of chronic severe asthma received a tuberculosis tine test and multiple vaccines (DT, MMR, OPV, and VV), which were followed by hives. Her brother had chicken pox 3 years earlier. The patient's medications included inhaled corticosteroids. She was hospitalized for asthma during the month after the vaccinations and again 19 months later, when she required intubation. She died with coagulopathy and Staphylococcus aureus sepsis. No rash had been recognized antemortem, but at autopsy 2 of several scattered papules showed "intraepidermal vesicles with viral inclusions and multinucleated cells." The autopsy disclosed fulminant hepatitis, esophagitis, and epiglottitis, with disseminated intravascular coagulation and positive varicella virus cultures from pharynx, lung, blood, and urine. Her liver was diffusely necrotic, and the few surviving hepatocytes showed viral cytopathic changes. The pathologist interpreted thymic depletion (weight 13.9 g vs 31.0 g expected at her age) and marked depletion of splenic white pulp as effects of corticosteroid therapy. Polymerase chain reaction studies identified wild-type VZV.

Patient B. A 16-month-old girl with history of upper respiratory tract infections was receiving an antibiotic for otitis media when she was immunized with HiBV and VV. Three days later, she began to vomit, developed progressive lethargy, and died the next day. At autopsy, she had meningitis, hepatitis, and otitis media. Chronic inflammatory cell infiltrates and multiple foci of isolated necrosis were found in the liver, but no viral inclusions were found. Viral culture of liver tissue was negative. Polymerase chain reaction identified wild-type VZV in brain tissue.

Patient C. A healthy 18-month-old boy had no history of allergy or any prior postvaccinal adverse event when he received multiple vaccinations (DTaP, OPV, VV) 2 to 3 weeks after a viral syndrome. He was admitted to the intensive care unit 4 days later with a low platelet count. He began to bleed from the mouth, had an abnormal computed axial tomography scan of the head, and died 2 days later from cerebral hemorrhage after an emergency left frontal lobectomy. His autopsy showed profound thrombocytopenia with changes compatible with idiopathic thrombocytopenic purpura, including hypercellular bone marrow with abundant megakaryocytes. Polymerase chain reaction studies detected neither wild-type nor Oka-strain VZV. Because of the short interval from vaccinations until recognition of thrombocytopenia, the prevaccination viral syndrome is thought to be a more plausible trigger for idiopathic thrombocytopenic purpura than the varicella or poliomyelitis vaccine strains.

Patient D. A 15-year-old boy's history included microcephaly, cerebral palsy, quadriplegia, mental retardation, multiple episodes of aspiration pneumonia, permanent tracheostomy, and thrombocytopenia thought to be secondary to an anticonvulsant drug. One month after vaccination against varicella, he developed adult respiratory distress syndrome with severe varicella pneumonia, disseminated varicella sepsis, and renal failure. He died 10 days later; no autopsy or PCR studies were performed.

Patient E. A 12-month-old boy received VV at a well-baby visit. He had no significant history beyond colds and ear infections, but his father and grandfather had apparently viral enteritis. The patient vomited once and became irritable 3 days after vaccination. He received acetaminophen the next day for irritability and fever (102° F [38.9 °C]). A rash developed 1 week after vaccination, which a physician diagnosed as roseola, describing it as erythematous, maculopapular, and extending from torso to groin. An apparently minor recent head trauma from a fall was also noted. Five days later the patient seemed well, but he became irritable and received acetaminophen. Some 6 hours later, his mother heard a "shrill" cry and found the patient supine and convulsing. Emergency personnel found the patient apneic and pulseless. At autopsy, he had no histologic signs of VZV. "Morphologic findings to suggest varicella as an etiology are not seen. The findings are interpreted to be most suggestive of early changes of viral meningoencephalitis. . . ."

DT indicates diphtheria and tetanus toxoids, pediatric strength; MMR, measles, mumps, and rubella vaccine; OPV, oral polio vaccine; VV, varicella vaccine; HiBV, Haemophilus influenzae type b conjugate vaccine; VZV, varicella zoster virus; DTaP, diphtheria and tetanus toxoids with acellular pertussis vaccine; and PCR, polymerase chain reaction; Box describes only individual case reports cited in text. Details of the 9 other reported fatalities are available on request.

Rash and Possible Vaccine Failure. Rashes, usually vesicular, accounted for more than half of all reports (3640 cases, 37.4/100,000 doses) (Table 2). Polymerase chain reaction studies found VZV in 70 adequate rash specimens. Wild-type virus appeared in 43 patients (61%), who had symptoms at a median of 1 week postimmunization, while the Oka strain was identified in 22 patients (31%) whose symptoms began a median of 4 weeks after vaccination. The VZV type could not be distinguished in 5 cases. Of the 1260 reports of possible vaccine failure (12.9/100,000), 51% described rash; 9 patients had complications, particularly secondary bacterial infections of vesicles. Negative serologic tests, rather than varicella infections, seemed to prompt 17% of the reported possible vaccine failures.

Injection Site Reactions. Among 575 patients with reported injection site reactions (5.9/100,000), the majority of reactions followed varicella vaccine administration by less than a week, and 77% of patients received no other vaccine. Eight reports described positive rechallenge. Four patients developed a shingles-like rash in the immediate vicinity of the injection site from 2 to 16 weeks after vaccination. Most of 10 serious cases involved the injection site only incidentally or as a possible portal of initial infection.

Herpes Zoster. The characteristic rash of herpes zoster (HZ) (asymmetric, dermatomal, and vesicular) facilitates specific diagnosis, but the reactivated virus may be of wild-type or Oka strain. Polymerase chain reaction studies verified VZV in specimens from 26 of 251 patients with HZ; 12 patients had the wild type at a median of 3 weeks after varicella vaccine vaccination, while 14 had the Oka strain a median of 19 weeks after vaccination.

Pharyngitis. Pharyngitis (172 reports) frequently accompanied a rash (62% of reports), sore throat, malaise, fever, or other upper respiratory symptoms. Two of 3 rash specimens from patients with pharyngitis were positive for Oka-strain VZV and 1 for wild type. Nine reports of stomatitis with no specific pharyngeal involvement included 1 positive rechallenge.

Cellulitis. Among 38 patients with cellulitis, 1 had a positive rechallenge, and 8 cases were serious. Almost half (17) of the reports involved the varicella vaccine injection site 4 to 20 days after injection. Several patients with infected remote vesicles of breakthrough varicella or HZ developed cellulitis but usually without clear evidence for a role of varicella vaccine. An exception was a 5-year-old girl with a zosteriform eruption 3 weeks after vaccination; unilateral facial lesions involved her eye. She required hospitalization for group A β-hemolytic streptococcal infection complicating HZ with culture-proven VZV and PCR confirmation of Oka strain.

Hepatic Pathology. Twenty-five reports described hepatitis, elevated enzyme levels, or other signs of hepatic pathology. Three reports specified encephalopathy, raising the question of possible vaccine-related Reye syndrome. However, 1 of the 3 patients had a wild-type VZV infection with extensive hepatic necrosis 20 months subsequent to vaccination (Box 1, Patient A). Encephalopathy developed in the other 2 patients about a week after vaccination, without evident cause, but neither patient had Reye syndrome, based on cerebrospinal fluid (CSF) in one and magnetic resonance imaging in the other. A patient with vomiting (another potential sign of Reye syndrome) and lethargy died with meningitis and chronic hepatic inflammation; brain tissue yielded wild-type VZV (Box 1, Patient B). Two patients had vomiting without encephalopathy. One had hepatitis A, and the other probably had Gilbert syndrome.

Pneumonia. Nineteen patients with pneumonia included 5 with predisposing immunodeficiencies from acquired immunodeficiency syndrome (AIDS), corticosteroids, or other conditions. In addition, 2 mothers developed severe varicella pneumonia during pregnancy. One received "shots" that may have included varicella vaccine at 5 months' gestation after her children developed chickenpox. The vaccinated child of the other mother developed a mild rash 2 weeks after vaccination, with the mother's varicella and pneumonia ensuing after another 2 weeks. Neither mother had PCR confirmation of VZV strain. A severely immunodeficient 13-month-old human immunodeficiency virus (HIV)–positive boy with pneumonia did have direct evidence of vaccine involvement. Approximately 10 weeks after vaccination, while he was hospitalized for gram-negative pneumonia, serology and a bronchoalveolar lavage specimen had positive results for VZV, and PCR studies from the lavage and a lung biopsy both identified the Oka strain.

Erythema Multiforme and Stevens-Johnson Syndrome. Erythema multiforme (EM) or its more severe form, Stevens-Johnson syndrome (SJS), or both developed in 46 patients. Among 42 cases reported as EM, 16 patients described symptom onset within 1 week after vaccination, and disease developed in 23 patients 1 to 5 weeks postvaccination; 3 reports lacked dates. One individual progressed to SJS; and 4 other patients (not reported as having EM) developed SJS. Three of the 5 SJS cases began within 1 week after vaccination. Except for 3 adults with EM, all patients with EM or SJS were younger than 9 years.

Arthropathy. Adults accounted for about half of the 45 reports of arthropathy. Thirteen patients developed arthritis, and 32 reported only arthralgias. Specified joints in arthritis reports included knee, ankle, metacarpal, metatarsal, or hip, but 3 patients described neck stiffness or pain. Although arthritis usually affected multiple joints, 3 patients had monoarticular presentations, and 2 had negative bacterial cultures. Arthritis developed in 8 patients within 2 days after vaccination, including 4 patients with EM-associated symptoms. Five patients had first symptoms of probable reactive arthritis 1 to 4 weeks after vaccination.

Thrombocytopenia. Fifteen children in the second year of life accounted for almost half of 31 thrombocytopenia reports. Two patients died. One received varicella vaccine shortly after a viral illness and then developed fulminant idiopathic thrombocytopenic purpura (Box 1, Patient C). The other patient had a history of drug-related thrombocytopenia (Box 1, Patient D). Symptoms in most cases (24/31) began 4 to 28 days after vaccination. Five patients had borderline platelet count depressions (120-147×10⁹/L), but 22 had counts of 52×10⁹/L or less, including 10 below 10 ×10⁹/L. A 14-year-old boy with probable positive rechallenge noted petechiae on his extremities about 1 week after receiving varicella vaccine and tetanus-diphtheria toxoid. Ten days after his second dose of varicella vaccine, his platelet count fell to 11,000.

Anaphylaxis. All 30 patients with anaphylaxis survived (9 reports specified anaphylaxis, and we classified another 21 as probable cases, based on compatible clinical features, including respiratory and skin symptoms within 4 hours after vaccination). In half of 22 detailed reports, symptoms developed within 15 minutes after vaccination. Five of the 30 patients had significant past medical histories of surgery for congenital cardiac anomalies or spina bifida, and several others had asthma. Four patients had food sensitivities, including 1 with a history of egg allergy and a similar reaction to MMR. Three patients had medication allergies to antibiotics (2 patients) or to atropine and an unspecified ophthalmic solution (1 patient).

Vasculitis. Among 15 vasculitis reports, 3 children aged 1 to 3 years appeared to have Kawasaki syndrome, and 10 patients developed Henoch-Schönlein purpura within 7 weeks of vaccination. Eight of these 10 patients were younger than 8 years; one also had EM, and 4 had associated joint pain, swelling, or both.

Aplastic Anemia. Two boys, 1 and 6 years old, developed aplastic anemia 2 months after administration of varicella vaccine alone or with MMR. Both children required bone marrow transplantation. Four patients had milder degrees of cytopenia in conjunction with arthritis (3 cases) or SJS (1 case).

Neuropathies. Reports for 193 patients include a wide variety of central and/or peripheral neuropathic signs or symptoms at a median of 4 days after vaccination, with 90% of symptoms reported as beginning within 5 weeks. In 50 of 193 cases, the primary pathology may have been associated encephalopathy, ataxia, meningitis, or seizures.

Fifteen patients developed Bell palsy, including 6 teenagers or adults, at intervals from less than 24 hours to almost 1 month after vaccination.

Another 15 patients, aged 1 to 38 years, developed demyelinating syndromes 6 to 128 days after vaccination: transverse myelitis (5 patients), optic neuritis (4 patients), acute demyelinating or disseminated encephalomyelitis (3 patients), Guillain-Barré syndrome (3 patients), and multiple sclerosis (1 patient). One patient had optic neuritis and transverse myelitis together. An unusual 16th report described a mother who developed transverse myelitis 1 month after her infant daughter's immunization (no live vaccine product); transverse myelitis recurred 1 year later, 3 days after the daughter's vaccination against varicella.

Among the remaining neuropathy reports, main adverse events included hypokinesia (33 cases), paresthesia (22 cases), and hypotonia (19 cases). A patient with positive rechallenge had paresthesias after his first dose of varicella vaccine, followed by resolution prior to exacerbation 2 weeks after his second dose.

Convulsions. Most of 163 reported seizures involved children aged 12 to 23 months. Febrile seizures accounted for about one half of reports and usually occurred after multiple immunizations. A larger proportion of reported seizures after administration of MMR with varicella vaccine occurred in the second week postvaccination than did reported seizures without preceding MMR (31/79 [39%] vs 9/82 [11%]).

In 25 reports, patients with no prior seizure history received only varicella vaccine and had no evident pathology to account for convulsions. Children younger than 5 years accounted for 12 of 13 febrile seizures in this subset, and 10 of these 13 had convulsions within 4 days after vaccination.

Ataxia. Forty-three reports of ataxia included 22 patients younger than 2 years. In 39 patients with interval information, symptoms began 7 to 42 days after vaccination in 21 patients; symptoms developed during the first week in 12 cases and more than 6 weeks after vaccination in 6 patients. Many cases appeared consistent with transient cerebellar ataxia, although 7 included encephalopathic features. Unrelated etiologies emerged in 2 patients: a possible brain tumor in one, and a congenital metabolic defect in the other.

Encephalopathy. Among 32 reports of encephalopathy (including encephalitis), 8 patients had evidence for etiologies independent of varicella vaccine, such as brain tumors in 3 patients. The primary pathology in 6 reports seemed to be aseptic meningitis. Three patients were diagnosed as having acute demyelinating or disseminated encephalomyelitis. The remaining 15 had a variety of symptoms, usually 1 to 4 weeks after vaccination. In one of the better documented cases, a 15-month-old girl developed hemiparesis 18 days after receiving diphtheria and tetanus toxoids with acellular pertussis vaccine, Haemophilus influenzae type b conjugate vaccine, and varicella vaccine. Her CSF was normal, with no VZV found in PCR studies. Magnetic resonance imaging showed edema of the left basal ganglia.

Meningitis. None of 11 meningitis reports, including 2 deaths (Box 1, Patients B and E), had PCR evidence of Oka-strain VZV in CSF or other specimens. Two patients had bacterial infections (Neisseria meningitidis and Borrelia burgdorferi). Brain tissue yielded wild-type VZV in a 16-month-old girl with vomiting, lethargy, and hepatitis (Box 1, Patient B). Another patient developed a high fever almost 24 hours after vaccination and progressed to aseptic meningitis. In a third case, a 33-month-old girl developed right facial HZ and viral meningitis. She had received varicella vaccine at age 1 year when her sibling had fresh chickenpox vesicles. Polymerase chain reaction studies identified wild-type VZV from HZ lesions in the trigeminal distribution and no VZV in the CSF. The remaining 2 adults and 3 children all developed meningitis approximately 2 to 3 weeks after vaccination. One woman developed a broadly distributed vesicular rash with 75 to 100 lesions 13 days after receiving varicella vaccine, followed after another 9 days by aseptic meningitis. Neither her CSF (by viral culture) nor rash (by PCR) had evidence of VZV, but her 2 children developed varicella 2 and 4 weeks after the appearance of their mother's postvaccinal rash.

Reports of unintentional exposures to varicella vaccine include 145 possible secondary transmissions from vaccinees, usually without PCR confirmation of vaccine strain VZV; 43 administrations to infants younger than 12 months (Table 1); 19 vaccinations of pregnant women given varicella vaccine by mistake (instead of varicella zoster immune globulin) after chickenpox exposures; 6 ocular contacts (from splashing during administration), usually followed by irritation and redness, with no case of HZ ophthalmicus; 5 unintended double doses (at the same visit or months apart), associated with injection site reactions, irritability, and anxiety; and 2 cutaneous exposures leading to localized vesicles.

VAERS received reports of 87 women who received varicella vaccine prior to or during pregnancy. None depicts characteristic features of congenital varicella infection in the exposed offspring. Several reports describe gestational varicella vaccine exposure followed by malformations (eg, Down syndrome and tetralogy of Fallot).

Growing use of varicella vaccine in the United States promises substantial control of chickenpox and its serious complications. Consistent with experiences in clinical trials, our review of 6574 spontaneous reports of suspected adverse effects from varicella vaccine during the first 3 years after licensure found that the vast majority of reported cases were not serious. Symptoms like fever, rash, and injection site reactions can be expected from this live virus vaccine, while the vaccine's role in most of the infrequent reports of serious adverse events remains unconfirmed.

VAERS data, subject to the inherent limitations of passive safety surveillance, merit cautious interpretation. Most reports cannot prove whether vaccination caused the subsequent symptoms. Not all adverse events that occur after vaccination are reported, and many reports describe events that may have been caused by confounding factors, including medications and diseases. Chickenpox remains prevalent, and the wild-type virus accounts for many reported events, including some serious cases. Follow-up disclosed tumors or other causes unrelated to vaccines in other cases of serious adverse events. Larger numbers of reports soon after licensure probably reflect the "Weber effect" of greater adverse event reporting for new drugs.^35,36 The quality of reported information varies widely, and simultaneous administration with other vaccines (especially MMR) confounds attribution. In addition, crude sales data preclude calculation of age-specific reporting rates. However, surveillance data can stimulate hypotheses for systematic evaluation through controlled studies.³⁷

This study extends safety data from clinical trials and postlicensure studies. Serious adverse events had not been seen before licensure, and a systematic search afterward in hospitalization and other records of almost 90,000 vaccinated members in a health maintenance organization found no case of an acute, serious adverse event.^24,25 In clear contrast, VAERS received multiple reports of anaphylaxis. All of these patients survived without complications. The offending allergen may be a gelatin stabilizer in varicella vaccines, MMR, and other products.^38-40

The majority of patients for whom serious adverse events were reported, including pneumonia, encephalitis, ataxia, thrombocytopenia, SJS, arthritis, vasculitis, and hepatitis, lacked varicella strain testing. Wild-type VZV and other mechanisms could also cause these syndromes, and alternative etiologies were confirmed for some patients through follow-up. Nonetheless, these and other diseases described in multiple serious reports are plausible as potential effects of varicella vaccine. Some are commonly recognized complications of natural chickenpox, particularly ataxia, cellulitis, and encephalitis.⁴¹ Others have also been described with wild-type VZV infections, including arthropathy, EM and SJS, aplastic anemia, pneumonia (usually in adults),⁷ thrombocytopenia,^42-47 Henoch-Schönlein purpura or vasculitis of the central nervous system,^48-50 and central and peripheral neuropathies.^51-56 Positive rechallenge reports (eg, for idiopathic thrombocytopenic purpura and paresthesias) bolster suspicion of relationships with varicella vaccine. Continued safety surveillance and epidemiologic evaluations may clarify whether these and other rarely reported adverse events can be attributed to varicella vaccine. Since many adverse events may be caused by wild-type VZV, physicians should obtain appropriate specimens for laboratory evaluation, including strain identification. While commercial laboratories do not yet have this capability, physicians can consult with CDC's National Varicella Reference Laboratory (Scott Schmid, PhD, telephone: 404-639-0066, e-mail: dss1@cdc.gov).

Reported seizures after administration of varicella vaccine and MMR clustered in the second week after vaccination to a much greater extent than did reported seizures after receipt of varicella vaccine alone. These patterns support a role of MMR in postvaccination febrile seizures, but further research is needed.

Where autopsy and other follow-up data were available, investigations of reported deaths often disclosed clear causes unrelated to vaccination, including malignancies, wild-type VZV, respiratory syncytial virus, and echovirus. An immunocompromised patient died with clinical diagnoses of varicella sepsis and pneumonia but without laboratory studies to confirm VZV or distinguish the strain. Another patient, severely asthmatic, died with wild-type VZV documented 21 months after vaccination. She may have had a primary varicella infection prior to vaccination, with subsequent disseminated HZ under the influence of corticosteroid therapy. Alternatively, corticosteroids, other asthma medications, or both might have attenuated her response to vaccination or later depressed her immune defenses.

The most commonly reported adverse events included rash, possible vaccine failures, and injection site reactions. Many cases likely represent unrelated infectious diseases. Wild-type VZV accounts for much of the postvaccinal varicella-like rash. If caused by VZV, rash within 7 days after vaccination is almost certain to be wild-type virus, while cases occurring 1 to 6 weeks after vaccination may be either Oka-strain or wild-type virus; almost all disseminated varicella rash beyond 6 weeks after vaccination is caused by wild-type virus and represents partial or complete vaccine failure. Most suspected secondary transmissions occurred between 7 and 42 days after vaccination and would require laboratory studies to distinguish between Oka-strain and wild-type VZV.

Vaccine-strain HZ may occur among vaccinated immunocompromised patients at a lower rate than in similar patients after natural varicella.⁵⁷ Our data verify that Oka-strain HZ can also occur in immunocompetent vaccinees. However, half of the reported HZ cases with adequate laboratory specimens had wild-type virus, which is evidence of natural VZV infection before immunization. The short intervals after vaccination until HZ occurrence in several patients seem consistent with the intriguing hypothesis that varicella vaccine might, in rare cases, provoke reactivation of latent wild-type VZV.⁵⁸

Involvement of the Oka strain in an immunocompromised patient with pneumonia was confirmed with PCR studies. Immune deficits (eg, congenital or resulting from AIDS or drug-induced immunosuppression) may contraindicate live virus vaccination. In addition, a postlicensure field study found lower vaccine effectiveness among children with asthma.¹⁵ Further studies should examine immune responses to vaccination in asthmatics receiving various treatment regimens. For patients who require varicella vaccine prior to planned immunosuppression, risks may be minimized by first allowing some weeks for the acute vaccine-induced VZV infection to resolve. (However, this precaution may not be necessary with relatively low-dose therapy.¹) Vaccinees with potentially impaired defenses should be closely monitored for the possible need to treat complications with acyclovir.

VAERS reports confirm that secondary transmission of the vaccine virus can occur, probably rarely. The risk of person-to-person passage of Oka-strain VZV is quite small^6,59,60 and probably limited to patients with a rash.⁷ Only 3 cases of secondary transmission have been confirmed in immunocompetent persons.³ Public health authorities recommend that family members and other close contacts of immunocompromised persons should receive varicella vaccine, in view of the threat otherwise faced by these patients from natural chickenpox and its complications.^1,2 If a vaccinee develops a rash, isolation from the immunocompromised person can reduce the transmission risk, but even with contact, vaccine-strain disease is unlikely and usually mild.

No vaccine has perfect efficacy, but varicella vaccine nearly always protects against severe varicella.¹⁵ Three postlicensure studies have demonstrated vaccine effectiveness in the range of 85% to 90% for prevention of all disease and 100% for prevention of severe disease.^15,61,62 Currently, approximately 1 in 10 vaccinated children may develop mild breakthrough disease following exposure to chickenpox. As exposures to natural varicella decline with increasing vaccine coverage, numbers of breakthrough cases should also fall.

More than 200 reports of possible vaccine failure apparently stemmed from misinterpretation of negative postvaccination serologic results as failures to seroconvert. Commercial assays are not sufficiently sensitive to detect all protective antibody responses following vaccination.¹⁶ Physicians and pertinent laboratory personnel should recognize this limitation in available tests.

Merck and the CDC jointly monitor potential fetal risks from gestational exposures through a pregnancy registry,²⁶ which, like VAERS, has received no report of congenital varicella syndrome.⁶³ However, reported pregnancy exposures highlight the need for physicians to ask women of childbearing age, before administering varicella vaccine, about the possibilities of current pregnancy or planned or potential conception in the next month. Gestational exposures to varicella vaccine through confusion with varicella zoster immune globulin delay protection against VZV⁶⁴ and expose the developing embryo to a potentially teratogenic virus.⁶⁵ Continued reporting of this error despite initial publicity⁶⁶ indicates a need for additional educational interventions.

Chickenpox can be serious and even deadly, but varicella vaccine can now prevent serious varicella infections with a high degree of reliability.^1-3 Safety surveillance through VAERS confirms that most of the vaccine's adverse effects are minor. Although reports to VAERS provide either tentative or clear evidence for a variety of serious vaccine risks, all appear to be rare, and the majority, while plausible, lack confirmation of causation by Oka-strain VZV. The manufacturer, in consultation with the FDA, has revised safety labeling for varicella vaccine (Table 3) based on continuous assessment of postmarketing spontaneous reports. Additional studies⁶ and ongoing investigations within the CDC's Vaccine Safety Datalink (VSD) Project³⁷ will evaluate several of the hypothesized vaccine risks, including ataxia, aplastic anemia, encephalopathy, seizures, and thrombocytopenia. Our analysis also suggests that further educational measures might help assure appropriate use and interpretation of varicella serologic assays and eliminate inadvertent substitutions of varicella vaccine for varicella zoster immune globulin, particularly in pregnancy. With continuing improvements in varicella vaccine coverage, evidence for control of varicella will emerge as declines in disease incidence, varicella-related hospitalizations, and mortality.⁶⁷

A summary of safety reports received since preparation of this article is presented in Box 2.

Vaccine safety surveillance is an ongoing process. VAERS received over 3000 additional varicella vaccine case reports in the 18 months after July 1998, with most patterns remaining stable as reports continue to accrue. However, 3 reports from this period subsequent to our primary analysis have positive rechallenge or other special information value.

Patient A. A 13-month-old boy with severe combined immunodeficiency developed hepatitis and respiratory distress 2 weeks after receiving varicella vaccine. Liver biopsy showed VZV infection, and PCR testing of the biopsy supernatant and of a rash specimen 6 weeks after vaccination both identified Oka-strain VZV. This case verifies vaccine virus involvement in hepatic pathology. In addition, as in the HIV-positive patient who developed pneumonia, it demonstrates persistence of vaccine virus activity for at least 6 weeks in an immunocompromised host.

Patient B. A 16-year-old boy without previous convulsions had an absence seizure 3 days after varicella vaccine. One month later, 2 generalized tonic-clonic seizures followed his second vaccine dose at the same interval. This patient's positive rechallenge for seizure activity increases suspicion that varicella vaccine may be more than a coincidental factor in observations of postvaccinal convulsions.

Patient C. Two weeks after receiving varicella vaccine, a 4-year-old girl developed hemiparesis with evidence from magnetic resonance imaging for cerebral infarctions in the putamen and internal capsule. Her apparent cerebrovascular accident assumes particular importance after recent description of a significant statistical association between natural chickenpox and subsequent ischemic strokes in children.⁶⁸