Safety of zoster vaccine in adults from a large managed-care cohort: a Vaccine Safety Datalink study


Hung Fu. Tseng PhD, Department of Research and Evaluation, Southern California Permanente Medical Group, Kaiser Permanente, 100 S Los Robles Ave, 2nd Floor, Pasadena, CA 91101, USA. (fax: 626-564-3409; e-mail:


Abstract.  Tseng HF, Liu A, Sy L, Marcy SM, Fireman B, Weintraub E, Baggs J, Weinmann S, Baxter R, Nordin J, Daley MF, Jackson L, Jacobsen SJ, for the Vaccine Safety Datalink (VSD) Team (Kaiser Permanente, Pasadena, CA; Kaiser Permanente, Oakland, CA; Centers for Disease Control and Prevention, Atlanta, GA; Kaiser Permanente, Portland, OR; HealthPartners Research Foundation, Minneapolis, MN; Kaiser Permanente, Denver, CO; and Group Health Cooperative of Puget Sound, Seattle, WA; USA). Safety of zoster vaccine in adults from a large managed-care cohort: a Vaccine Safety Datalink study. J Intern Med 2012; 271: 510–520.

Objectives.  The aim of this study was to examine a large cohort of adults who received the zoster vaccine for evidence of an increased risk of prespecified adverse events requiring medical attention.

Design.  Two self-comparison approaches, including a case-centred approach and a self-controlled case series (SCCS) analysis were used.

Setting.  Eight managed-care organizations participating in the Vaccine Safety Datalink project in the United States.

Subjects.  A total of 193 083 adults aged 50 and older receiving a zoster vaccine from 1 January 2007 to 31 December 2008 were included.

Main outcome measures.  Prespecified adverse events were identified by aggregated International Classification of Diseases, Ninth Revision (ICD-9) codes in automated health plan datasets.

Results.  The risk of allergic reaction was significantly increased within 1–7 days of vaccination [relative risk = 2.13, 95% confidence interval (CI): 1.87–2.40 by case-centred method and relative rate = 2.32, 95% CI: 1.85–2.91 by SCCS]. No increased risk was found for the following adverse event groupings: cerebrovascular events; cardiovascular events; meningitis; encephalitis; and encephalopathy; and Ramsay-Hunt syndrome and Bell’s palsy.

Conclusions.  The results of this study support the findings from the prelicensure clinical trials, providing reassurance that the zoster vaccine is generally safe and well-tolerated with a small increased risk of allergic reactions in 1–7 days after vaccination.


Shingles or herpes zoster (HZ), is caused by reactivation of varicella zoster virus (VZV) that has remained latent in sensory ganglia since primary varicella infection. Incidence data and current population figures project estimates of up to 1 million new cases of HZ each year in the United States [1]. More than half of these cases occur in persons 60 years or older.

A zoster vaccine (Zostavax™, Merck, West Point, PA, USA) was licensed in 2006 for the prevention of HZ by the U.S. Food and Drug Administration (FDA) and subsequently recommended by the Advisory Committee for Immunization Practices (ACIP) for healthy persons of ages 60 years and older [2]. In March, 2011, the FDA also approved the use of Zostavax in individuals, 50–59 years of age. The vaccine is a lyophilized preparation of a live, attenuated strain of VZV, the same strain used in varicella vaccine, but with a viral titre at least 14-times greater. It also contains additional antigenic components including nonviable Oka/Merck VZV, porcine gelatin, fetal calf serum, MRC-5 tissue culture cells and neomycin.

Post-marketing safety information on the vaccine is limited. In clinical trials, it has been evaluated for safety in approximately 22 500 adults. In the Shingles Prevention Study (SPS) [3], subjects received either the zoster vaccine (n = 19 270) or placebo (n = 19 276), and were actively followed for safety outcomes through 42 days after vaccination and passively followed for safety after day 42. In the Adverse Event Monitoring Substudy of the SPS, 3345 enrolees received the zoster vaccine and 3271 received placebo; subjects used vaccination report cards to record adverse events occurring from days 0 to 42 after vaccination. Monthly surveillance for hospitalization for adverse events possibly related to vaccination was conducted through the end of the study, 2–5 years after vaccination. The authors concluded that the zoster vaccine is generally well-tolerated, causing minor local inoculation-site reactions, but no more serious adverse events than placebo [4].

While the SPS provided evidence that the vaccine is safe in a selected study population, it is important to evaluate safety of this vaccine in clinical practices with more physiologically heterogeneous populations, because such heterogeneity might potentially affect the vaccine safety profile. The SPS recruited study participants at each of 22 Veteran Affair centres via local advertising, enrolling motivated volunteers that included 40% females, 2% African American and just 1.4% Hispanic. Participants tended to be healthy and ambulatory. Participants were required to have a history of varicella, or at least 30 years of residence in the United States. Other SPS exclusions included chronic pain syndromes, cognitive impairment, severe hearing loss, major underlying illness and various other concurrent treatments, vaccinations or conditions [3]. These criteria clearly limit the generalizability of the study findings. Our study was undertaken to examine a large cohort of the general population who received the zoster vaccine for evidence of increased risks of prespecified adverse events.


Subjects and settings

All adults age ≥50 years who received a zoster vaccine from 1 January 2007 to 31 December 2008 were eligible for the study. Adults receiving the vaccine before the age of 60 were considered ‘off-label’ users. Vaccination data were identified from electronic medical records and collected from eight managed-care organizations participating in the Vaccine Safety Datalink (VSD) project, funded by the Centers for Disease Control and Prevention (CDC) [5]. The Institutional Review Boards of each organization approved this study.

Prespecified adverse events

We used computerized data to identify any adult with a prespecified event of interest or death. Events of interest were identified by diagnoses that were coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. Five major groups of events of interest included Group 1: Stroke and Cerebrovascular diseases, Group 2: Cardiovascular diseases, Group 3: Meningitis, encephalitis and encephalopathy, Group 4: Ramsay-Hunt Syndrome and Bell’s palsy, and Group 5: Medically attended reactions (reactions leading to a medical visit). These events of interest were prespecified based on safety data from clinical trials [3, 4] for events such as vascular disease and injection-site reaction, review of reports from the Vaccine Adverse Event Reporting System (VAERS) [6] for events such as myocardial infarction, stroke, facial palsy, facial pain, facial nerve disorder and serious local reactions, and biological plausibility for events such as neurological complications like Bell’s palsy [7, 8], meningitis and encephalitis [9–11] (Table 1).

Table 1. International Classification of Disease the 9th Edition codes, type of encounter and risk windows for prespecified adverse events following zoster vaccination, 2007–2008
Adverse event groupICD-9 CodesaCode descriptionType of encounterRisk window
  1. aFor three-digit codes, include all 4th and 5th digits with that code; for four digit codes, include all 5th digits with that code

Group 1. Stroke, Cerebrovascular diseases First episode in 12 monthsInpatient and ED1–14 days
15–28 days
29–42 days
1–42 days
 433Occlusion and stenosis of precerebral arteries  
 434Occlusion of cerebral arteries  
 435Transient cerebral ischaemia  
 436Acute, ill-defined, cerebrovascular disease  
 437Other and ill-defined cerebrovascular disease  
Group 2. Cardiovascular events First episode in 12 monthsInpatient and ED1–14 days
15–28 days
29–42 days
1–42 days
Acute myocardial infarction410Acute myocardial infarction  
 411Other acute and subacute forms of ischaemic heart disease  
Acute pericarditis420Acute pericarditis  
Acute myocarditis422.0Acute myocarditis in diseases classified elsewhere  
 422.90Acute myocarditis NOS  
 422.91Idiopathic myocarditis  
 422.99Other and unspecified acute myocarditis  
 429.0Myocarditis unspecified  
Cardiomyopathy425.4Other primary cardio-myopathies NOS  
 425.8Cardiomyopathy in other diseases classified elsewhere  
 425.9Secondary cardiomyopathy unspecified  
 429.3Cardiomegaly – dilation, hypertrophy, ventricular dilation  
Heart failure (Defined as heart failure code plus cardiomyopathy code)428.0Congestive heart failure unspecified  
 428.1Left heart failure  
 428.2Systolic heart failure  
 428.4Combined systolic and diastolic heart failure  
 428.9Heart failure unspecified  
Group 3. Meningitis, encephalitis, and encephalopathy First episode in 12 monthsInpatient and ED1–14 days
15–28 days
29–42 days
1–42 days
 047.8Other specified viral meningitis  
 047.9Unspecified viral meningitis, Viral meningitis NOS, meningitis  
 049.9Unspecified nonarthropod-borne viral diseases of central nervous system, Viral encephalitis NOS  
 321.2; 321.8Meningitis due to viruses not elsewhere classified; Meningitis due to other nonbacterial organisms classified elsewhere  
 322.0; 322.1; 322.2; 322.9Meningitis with no organism specified as cause; Eosinophilic meningitis; Chronic meningitis; Meningitis, unspecified  
 323; 348.3Encephalitis, myelitis and encephalomyelitis; Encephalopathy, acute  
Group 4. Ramsay-Hunt Syndrome, Bell’s Palsy First episode in 12 monthsOutpatient and ED1–14 days
 053.11Geniculate HZ, Herpetic geniculate ganglionitis  
 351.0Bell’s palsy, Facial palsy  
Group 5: Medically attended Reactions First episode in 30 daysOutpatient and ED1–7 days
 (a) Cellulitis and infection682.3Cellulitis, upper arm and forearm  
 682.9Cellulitis, unspecified site  
 729.5Pain in limb  
 729.81Limb swelling  
 999.3Infection following infusion, infection or vaccination  
 999.9Complication of medical care  
 (b) Allergic reactions   1–7 days
 995.1Angioneurotic oedema  
 995.2Adverse effects of drug  
 995.3Allergy unspecified  
 708.0Allergic urticaria  
 708.1Idiopathic urticaria  
 708.9Urticaria, unspecified  
 999.5Serum reaction  
 995.0; 999.4Anaphylaxis 0–1 days

To avoid visits for follow-up of an existing condition, only the first event documented within a specified time interval for that individual was counted. For example, an incident event of stroke and cerebrovascular disease was defined as the ‘first episode in 12 months’, meaning that there was no such diagnosis in the 12 months prior to the current event. All eligible subjects with adverse events were required to have a minimum continuous membership prior to their first event in the study period, so that we could determine if the event of interest was the first episode within a specific time interval. For medically attended reactions, the membership requirement was 30 days. The membership requirement was 12 months for all other events. Depending on the outcome, only certain visit types (e.g. outpatient, inpatient and emergency department) were included in an attempt to more narrowly define outcomes to restrict to the most severe outcomes, which may be less prone to misclassification (Table 1). Death was identified from the mortality file prepared by the VSD sites.

The length of the risk window(s) for the potential adverse event varied from 1 to 42 days depending on the event of interest (Table 1). As the temporal patterns of risk were uncertain, a variety of time periods were studied. Shorter windows were chosen for events with more immediate onset. All events (except for anaphylaxis) occurring on the day of vaccination (day 0) were excluded to eliminate events with onset prior to vaccination or that may have initiated the visit at which the vaccine was given. For anaphylaxis, we identified all cases with ICD-9 codes for anaphylaxis on Days 0–1, and reviewed the medical records to confirm the diagnosis and determine if the event occurred after receiving the zoster vaccine.

As statistical power may not be adequate to detect an adverse outcome from individual codes, we examined aggregated codes representing groupings of clinically related individual ICD-9 codes, except for cardiovascular events where subgroup analyses were performed. These aggregated codes were determined in advance by the authors and derived in part from grouped codes used in previous studies [12–14]. The risk windows were prespecified and not further adjusted after data were collected and analysed.

Statistical analysis

As vaccinated and unvaccinated persons differ in ways that are difficult to measure and control, we focused on within-person comparisons using the case-centred [15–17] and the self-controlled case series (SCCS) methods [18, 19] in persons with the events of interest. We presented results analysed by two different approaches in this study to demonstrate the consistency of findings.

Case-centered design.  The case-centred approach was detailed elsewhere by VSD investigators [15–17]. In practice, the method begins by identifying the event of interest among the vaccinated population. We determine how many of them were vaccinated during the period that we prespecified as the risk window – for example, during the 42 days prior to the event. Although SCCS would evaluate whether the event happened during the 42 days after vaccination, the case-centred approach evaluates whether vaccination happened during the 42 days before the event. Thus, the SCCS window is anchored by the vaccination date; the case-centered window is anchored by the date of adverse event.

We illustrate the design of the method in Fig. 1 (a). For each individual event of interest, the vaccination could fall into either the recent follow-back period (i.e. the prespecified risk window) or the remote follow-back period. The expected probability that a vaccination falls in the prespecified risk window assuming the vaccine is safe can be estimated by the proportion of doses of the entire vaccinated population falling within that time interval. In theory, the method compares the ‘observed’ and ‘expected’ probability of receiving vaccination in the prespecified risk window. For example, suppose we observed 10 cases of an event of interest, and on average 10% of the vaccine doses were given in the prespecified risk window, and 90% were in the remote follow-back period. Assuming the vaccine is safe, we would expect that one case had vaccination in the prespecified risk window and nine cases had vaccination in the remote follow-back period. The expected ‘case-split’ in this scenario would be 1/9. If we observed a different case-split (e.g. 3/7), we could use the case-centred method to evaluate whether this was unusual. The method compares the ‘observed’ case-split (i.e. 3/7) to the ‘expected’ case-split under the null hypothesis that the vaccine is safe (i.e. 1/9). The method yields a relative risk (RR) estimate that maximizes the likelihood of the observed case-split and a test of the null hypothesis.

Figure 1.

 Diagrams of Case-Centred design and self-controlled case series design. (a) The case-centred window is anchored by the date of the adverse event. It looks backward to identify whether the vaccination date falls in the recent follow-back period (the prespecified risk window). (b) The self-controlled case series window is anchored by the time of vaccination. It looks forward to identify the adverse event of interest after vaccination.

Analytically, a logistic regression model was used to fit data with one record for each event for which the individual received a vaccine within 1 year prior to the event date. The 1-year period was specified to limit the possible bias resulting from time-varying covariates. The dependent variable indicates whether the case was vaccinated inside the prespecified risk window (coded as 1) versus outside the window (coded as 0). A binomial probability that the vaccine was given inside versus outside the prespecified risk window is the basis for an offset term for each case. For each case, the offset is the log of the odds of the entire VSD population receiving vaccination in the prespecified risk window. By specifying the offset term, the intercept term of the logistic regression model becomes an estimate of the log of the RR.

SCCS design.  We also conducted SCCS analyses on all events of interest except death. Only cases of an event of interest occurring after vaccination were included in the analysis. The incidence rate of an event of interest in the prespecified risk window was compared with that in a control window by conditional Poisson regression. The control window was defined as the same length of time after a 30 day ‘wash-out’ period following the risk window (Fig. 1b). We assumed age was not a time-varying confounder in this case as the observational period was relatively short and the risk of events of interest was not likely to vary significantly in such a period among adults.

Medical record review was conducted to validate presumptive cases if significant increased risks were found or if considered necessary by authors based on clinical judgment. We reviewed all the medical records for the conditions that were severe or a sample of medical records to confirm the diagnoses and describe the nature of the events if the total number of events was large. If a significant increased risk was found, we also repeated the analysis stratified by age groups (50–59, 60–69, 70–79, ≥80 years) to further examine whether the increased risk differs by age or not. All analyses were conducted using sas version 9.1 (SAS Institute Inc., Cary, NC, USA).


A total of 193 083 adults aged 50 and older received zoster vaccinations in the VSD cohort during the study period. Among them, 6816 (3.5%) were 50–59 years old, 103 410 (53.6%) were 60–69 years old, 65 730 (34.0%) were 70–79 years old and 17 127 (8.9%) were 80 years and older. Women comprised 58.7% (113 386) of the cohort.

For Group 1, no increased risk was found for stroke or cerebrovascular events in the 1–14, 15–28, 29–42 or 1–42 days post-vaccination. For Group 2, an increased risk of acute myocarditis was found when events were identified by diagnostic codes. Before medical record review, two cases of acute myocarditis were identified by ICD-9 codes from 15 to 28 days post-vaccination and three cases from 29 to 42 days. The RRs estimated by case-centred methods were 8.98 [95% confidence interval (CI): 1.67–46.36] and 17.18 (95% CI: 3.71–79.67) for the two risk windows respectively. Medical record review of these five cases revealed that all were miscoded and were not incident cases of acute myocarditis. No increased risk was found for cardiovascular events in the 1–14, 15–28, 29–42 or 1–42 days post-vaccination. For Group 3, the risk of meningitis, encephalitis and encephalopathy in the prespecified risk periods was not elevated. For Group 4, the risk of Ramsay-Hunt syndrome and Bell’s palsy did not increase within 14 days following vaccination. For Group 5, a small but significant increased risk of cellulitis and infection was found 1–7 days post vaccination by the case-centred approach (RR = 1.30, 95% CI: 1.18–1.44) but not with the SCCS approach. The risk of allergic reaction significantly increased in 1–7 days of vaccination (RR = 2.13, 95% CI: 1.87–2.40 by case-centred method and 2.32, 95% CI: 1.85–2.91 by SCCS). Due to this increased risk in the outpatient and emergency department setting, we further examined the inpatient data and found no same day allergic reactions in the emergency and inpatient settings. For the mortality outcome, the all-cause mortality rate was significantly reduced in the 42 days after vaccination (Table 2).

Table 2.   Relative risk (RR) and 95% confidence interval (CI) of prespecified adverse events within predefined risk windows following vaccination with a zoster vaccine in eight managed-care organizations in the United States, 2007–2008
 Day 1–14Day 15–28Day 29–42Day 1–42
  Case-splitRR (95% CI)Case-splitRR (95% CI)Case-splitRR (95% CI)Case-splitRR (95% CI)
  1. aCase-centred method, case-split: The number of cases vaccinated inside the risk window versus the number of cases vaccinated outside the risk window.

  2. bSelf-controlled case series method, case-split: The number of cases in the risk window versus the number of cases in the control window.

  3. cAfter medical record review.

  4. dFour cases had events in both risk and control windows.

Stroke, cerebrovascular diseases(1)a88/16421.03 (0.83–1.28)77/16530.92 (0.73–1.16)86/16441.06 (0.85–1.31)251/14791.00 (0.87–1.15)
(2)b81/860.94 (0.70–1.28)74/731.03 (0.74–1.42)83/860.97 (0.71–1.30)233/2350.99 (0.83–1.19)
Acute myocardial infarction(1)a75/12571.17 (0.92–1.48)66/12661.04 (0.81–1.34)61/12710.97 (0.75–1.26)202/11301.07 (0.92–1.26)
(2)b71/581.22 (0.87–1.73)63/511.24 (0.85–1.79)57/590.97 (0.67–1.39)189/1801.05 (0.86–1.29)
Acute pericarditis(1)a0/200/201/191.04 (0.13–8.05)1/190.27 (0.03–2.22)
(2)b0/30/31/11.00 (0.06–15.99)1/20.50 (0.05–5.51)
Acute Myocarditis(1)a0/72/58.98 (1.67–46.36)3/417.18 (3.71–79.67)5/219.44 (3.58–105.68)
(2)b0/12/03/13.00 (0.31–28.84)5/15.00 (0.58–42.80)
Cardiomyopathy(1)a34/9210.73 (0.51–1.03)50/9051.11 (0.83–1.48)45/9101.00 (0.74–1.36)129/8260.94 (0.77–1.14)
(2)b33/470.70 (0.45–1.10)45/431.05 (0.69–1.59)45/490.86 (0.57–1.29)119/1270.94 (0.73–1.20)
Heart failure(1)a17/5610.76 (0.46–1.24)23/5551.08 (0.70–1.65)20/5580.95 (0.60–1.49)60/5180.91 (0.68–1.21)
(2)b17/220.77 (0.41–1.46)22/240.92 (0.51–1.63)18/280.64 (0.36–1.16)56/640.88 (0.61–1.25)
Meningitis, encephalitis and encephalopathy(1)a4/1490.54 (0.19–1.52)6/1470.90 (0.40–2.05)4/1490.62 (0.23–1.69)14/1390.66 (0.37–1.16)
(2)b4/50.80 (0.21–2.98)6/70.86 (0.29–2.55)4/50.80 (0.21–2.98)14/180.78 (0.39–1.56)
Ramsey-Hunt syndromes and Bell’s palsy(1)a7/1810.63 (0.29–1.38)      
(2)b7/90.78 (0.29–2.09)      
  Day 1–7       
Cellulitis and infection(1)a416/144491.30 (1.18–1.44)      
(2)b409/373d1.10 (0.95–1.26)      
Allergic Reaction(1)a257/40802.13 (1.87–2.40)      
(2)b253/1092.32 (1.85–2.91)      
Mortality(1)a      55/11710.31 (0.23–0.40)

As a result of the increased risk of the allergic reaction category, we abstracted the medical records of 118 patients who were reported as having an allergic reaction (Group 5b, Table 1) in days 1–7 of receiving zoster vaccine from one site, and they were reviewed by one of the authors (SMM) to establish the nature of these reactions and their relationship to the administration of zoster vaccine. Thirty-one cases were clearly attributable to unrelated causes. The history and physical findings described for 16 cases were vague and inadequate to determine timing of the onset of reaction; to establish whether a reported reaction was related to zoster vaccine or another vaccine given concomitantly, often adjacent to the zoster vaccine; to establish whether due to a new medication started at or shortly after the immunization; or to determine if there was even a reaction at all. Of the 71 patients whose medical visit was determined to be the result of a reaction to the zoster vaccine, most (n = 59, 83%) complained of a localized inflammatory response with varying degrees and combinations of redness, swelling and/or tenderness at the site of the injection. Eleven (15%) presumably allergic, pruritic, urticarial, macular or papular rashes were described: nine were diffuse, one consisted of an urticarial eruption surrounding the site of injection and one was manifest as unilateral eyelid swelling. A single patient was described as having a zosteriform rash ‘left T4...a few hours after getting the shingles vaccine’.

There were nine vaccinated patients with diagnosis codes indicating anaphylaxis on the day of vaccination (Day 0). No cases were found on Day 1. After reviewing the medical records, none of them were confirmed as anaphylaxis following vaccination. For seven of the nine cases, the anaphylaxis codes were given on the day of vaccination as a follow-up for conditions with an onset before vaccination. For the other two cases, the codes were given while no anaphylaxis diagnosis was mentioned.

Table 3 presents the age-specific RR of allergic reactions 1–7 days following vaccination. For the 50–59 year-old age group, there was a 10-fold increased risk (RR = 11.07, 95% CI: 6.68–18.33 by case-centred method or RR = 10.00, 95% CI: 2.34–42.78 by SCCS). The risk was also significantly increased in the 60–69 and 70–79 year-old groups (RR = 2.71, 95% CI: 2.28–3.21 and RR = 1.52, 95% CI: 1.20–1.93, respectively, by case-centred method or RR = 3.17, 95% CI: 2.29–4.38 and RR = 1.59, 95% CI: 1.10–2.30, respectively, by SCCS). No increased risk was found among patients 80 years and older.

Table 3.   Age-specific relative risk (RR) of allergic reaction 1–7 days following vaccination with a zoster vaccine in eight managed-care organizations in the United States, 2007–2008
Age groupCase-centred, case-splitaRR (95% confidence interval)SCCS, case-splitbRelative rate (95% confidence interval)
  1. aCase-centered method, case-split: The number of cases vaccinated inside the risk window versus the number of cases vaccinated outside the risk window.

  2. bSelf-controlled case series method, case-split: The number of cases in the risk window versus the number of cases in the control window.

50–5921/9411.07 (6.68–18.33)20/210.00 (2.34–42.78)
60–69155/20252.71 (2.28–3.21)152/483.17 (2.29–4.38)
70–7973/15411.52 (1.20–1.93)73/461.59 (1.10–2.30)
≥808/4200.51 (0.24–1.08)8/130.62 (0.26–1.49)


In this study designed to detect an increase in adverse events after administering zoster vaccine to adults aged ≥50 years, we found that the risk of a severe adverse event in the 42 days following vaccination was no more than what is expected based on the risk in the comparison period.

The only significantly increased risk, identified by both analytic methods, was allergic reactions in the 1–7 days window in the outpatient and emergency department settings. After further review of the medical records, we found that more than 80% of the events involved a localized inflammatory response with varying degrees and combinations of redness, swelling and/or tenderness at the site of the injection. This reflects the behaviours of coding these localized inflammatory responses using allergic-related codes, either by care providers or by professional coders. A small increased risk of cellulitis, 1–7 days following vaccination found by case-centred method may well represent inflammatory or allergic reactions rather than true infectious cellulitis. This finding is consistent with the SPS safety study, in which injection-site reactions were more common among vaccine recipients, although most were mild or moderate in severity [4].

In addition, we found that inflammatory or allergic reactions were particularly common among persons aged 50–59 years old. The risk was also increased more than 2.5-fold in persons aged 60–69 years and 1.5-fold in persons aged 70–79 years. This finding is comparable to the SPS safety study, in which the proportion of vaccinated persons with at least one inoculation-site adverse event was significantly higher in persons aged 60–69 years than in ≥70 years at 56.6% and 39.2% respectively [4]. In the SPS, erythema (41.6% vs. 29.4%), swelling (32.4% vs. 19.5%) and pain/tenderness (43.0% vs. 25.3%) were significantly more common in persons aged 60–69 years than in those ages ≥70 years [4]. The authors of the SPS study suggested that the local side effects of the zoster vaccine may have been mediated by immune responses to the attenuated vaccine virus, which may have been more vigorous in younger than older participants [4, 20]. Similar factors may have been responsible for the increased risk found in the 50–59 year-olds in our study, although it might also have been due to random variation, health seeking behaviours or some unknown factors associated with off-label users.

We found no significant increased risk among categories of cerebrovascular or cardiovascular events after medical record review of myocarditis coded cases within 42 days after vaccination. With a proposed mechanism that an increased risk for acute vascular pathology may be due to inflammatory mediators generated during the immune response to the live virus vaccine [21], Simberkoff et al. [4] conducted two post hoc analyses to address concerns about any possible vaccine-associated increase in the risk for vascular events in older adults. Rates of serious adverse events in the 42 days after inoculation did not significantly differ in vaccine and placebo recipients ages 80 years or older, potentially the most vulnerable trial participants. In addition, no significant or clinically meaningful differences in the occurrence of cardiovascular and cerebrovascular events classified on the basis of inferred pathophysiology were found between vaccine and placebo recipients [4].

Based on the conditions associated with naturally acquired varicella or HZ, cerebellar ataxia and vaccine-associated rash have been proposed as potential adverse events associated with immunization. However, because dizziness and rash are nonspecific findings that would require intensive medical record review for detection and verification, they were not included in this study. Other possible vaccine-associated adverse events not investigated include worsening of long-standing chronic conditions and transmission of vaccine virus to susceptible close contacts. Although vaccinated children with varicella vaccine-associated rash have been demonstrated to transmit vaccine virus to adult contacts [22], this has not to date been reported among recipients of the zoster vaccine. Also, vaccine-strain VZV may potentially lead to adverse events related to VZV reactivation. Vaccine-strain VZV may reactivate and cause HZ, Ramsay-Hunt syndrome or Bell’s palsy, as has been seen with reactivation of wild-type VZV. As it may take several years before vaccine-strain VZV reactivates, this will require more long-term surveillance and would be best examined with a different type of study design.

The incidence of mortality appears to be lower in the 42-day period after vaccination than at other times. This is consistent with the observation that persons who are at the end stage of their life are less likely to receive the vaccine [23]. People usually receive vaccination when they feel they are healthy or when their physicians recommend it based on their health status. The bias resulting from this phenomenon is likely to be minimal when the events under investigation have an acute onset and are difficult to predict. For example, the probability of receiving a vaccine at any given time is unlikely to be affected by the subsequent onset of stroke or acute myocardial infarction, which is hard to foresee. While it cannot be ruled out completely, underestimation of the risk of serious adverse events other than death during the risk period following vaccination due to a healthy vaccine effect is probably negligible.

Ideally, events considered for this analysis would be associated with primary contact with VZV or an immune-mediated inflammatory response to the vaccine virus, because we are examining a short-term risk. However, because the safety of the zoster vaccine in large populations is still under surveillance, some of the prespecified events investigated in this study were selected based on the biological possibility of reactivation of latent VZV (i.e. not complications of acute vaccine virus infection). For example, meningitis was selected based on the fact that VZV can spread to the central nervous system, producing a wide variety of syndromes ranging from diffuse small-vessel vasculopathy to meningitis and ventriculitis in patients with depressed cellular immunity [9–11]. However, whether inoculation of an Oka virus vaccine would increase the risk of such event and what constitutes a reasonable risk window remains unknown.

A limitation of the study is the potential for misclassification of event status. Some of the events with insidious onsets (e.g. cardiomegaly) are so poorly ascertained that even 12 months of prior observation time could be insufficient to determine and verify their occurrence. Such misclassification is likely to be nondifferential with respect to window periods. It is also possible that using ICD-9 codes would miss adverse events that were not coded for.

National data have shown that early uptake of the vaccine has been low [24]. While there is no evidence to suggest that safety concerns contribute importantly to that low uptake, the data from this study are likely to be reassuring. The results of this study support the findings from the prelicensure clinical trials, providing reassurance that the zoster vaccine is generally safe and well-tolerated with a low increased risk of inflammatory or allergic reactions. The causes of an elevated risk of these reactions among the 50–59 year age group warrant further investigation.

Conflict of interest statement

No conflict of interest to declare.


This study was funded through a subcontract with America’s Health Insurance Plans (AHIP) under contract 200-2002-00732 from the Centers for Disease Control and Prevention (CDC). The authors thank David McClure and Lei Qian for their comments on statistical methods, Sungching Glenn for her assistance with programming, and Theresa Im, Felicia Bixler, Sarah Fisher and Zendi Solano for their assistance with medical record reviews. Dr. Hung Fu Tseng had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.


The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.