CI indicates confidence interval.
Hung Fu Tseng, Ning Smith, Rafael Harpaz, Stephanie R. Bialek, Lina S. Sy, Steven J. Jacobsen. Herpes Zoster Vaccine in Older Adults and the Risk of Subsequent Herpes Zoster Disease. JAMA. 2011;305(2):160–166. doi:10.1001/jama.2010.1983
Author Affiliations: Department of Research and Evaluation, Southern California Kaiser Permanente, Pasadena (Drs Tseng, Smith, and Jacobsen and Ms Sy); and Division of Viral Diseases, National Center for Immunization and Respiratory Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia (Drs Harpaz and Bialek).
Context Approximately 1 million episodes of herpes zoster occur annually in the United States. Although prelicensure data provided evidence that herpes zoster vaccine works in a select study population under idealized circumstances, the vaccine needs to be evaluated in field conditions.
Objective To evaluate risk of herpes zoster after receipt of herpes zoster vaccine among individuals in general practice settings.
Design, Setting, and Participants A retrospective cohort study from January 1, 2007, through December 31, 2009, of individuals enrolled in the Kaiser Permanente Southern California health plan. Participants were immunocompetent community-dwelling adults aged 60 years or older. The 75 761 members in the vaccinated cohort were age matched (1:3) to 227 283 unvaccinated members.
Main Outcome Measure Incidence of herpes zoster.
Results Herpes zoster vaccine recipients were more likely to be white, women, with more outpatient visits, and fewer chronic diseases. The number of herpes zoster cases among vaccinated individuals was 828 in 130 415 person-years (6.4 per 1000 person-years; 95% confidence interval [CI], 5.9-6.8), and for unvaccinated individuals it was 4606 in 355 659 person-years (13.0 per 1000 person-years; 95% CI, 12.6-13.3). In adjusted analysis, vaccination was associated with a reduced risk of herpes zoster (hazard ratio [HR], 0.45; 95% CI, 0.42-0.48); this reduction occurred in all age strata and among individuals with chronic diseases. Risk of herpes zoster differed by vaccination status to a greater magnitude than the risk of unrelated acute medical conditions, suggesting results for herpes zoster were not due to bias. Ophthalmic herpes zoster (HR, 0.37; 95% CI, 0.23-0.61) and hospitalizations coded as herpes zoster (HR, 0.35; 95% CI, 0.24-0.51) were less likely among vaccine recipients.
Conclusions Among immunocompetent community-dwelling adults aged 60 years or older, receipt of the herpes zoster vaccine was associated with a lower incidence of herpes zoster. The risk was reduced among all age strata and among individuals with chronic diseases.
Herpes zoster, commonly known as shingles, is a painful vesicular rash caused by reactivation of varicella zoster virus, persisting latently in dorsal root ganglia.1,2 The pain of herpes zoster is often disabling and can last for months or even years, a complication termed postherpetic neuralgia. Approximately 1 million episodes of herpes zoster occur in the United States annually, but aside from age and immunosuppression, risk factors for this condition are not known.3,4
The Shingles Prevention Study (SPS) is a clinical trial of a live-attenuated vaccine prepared from the Oka/Merck strain of varicella zoster virus.5 The SPS trial included 38 546 participants aged 60 years or older who had no history of herpes zoster, immunosuppression, or conditions that could interfere with participation. In SPS, herpes zoster vaccine reduced herpes zoster and postherpetic neuralgia incidence by 51% (P < .001) and 67% (P < .001), respectively; laboratory markers indicating protection were not identified.6 Consequently, Zostavax (Merck & Co Inc, Whitehouse Station, New Jersey) was licensed by the US Food and Drug Administration in 2006 and subsequently recommended by the Advisory Committee for Immunization Practices for healthy individuals aged 60 years or older.7
Although the SPS provided evidence that herpes zoster vaccine works under idealized conditions (ie, vaccine efficacy), confirmation of these results is needed in field conditions (ie, vaccine effectiveness) to show whether benefits of the vaccine can be generalized to conditions of clinical practice.8- 10 This is particularly important for herpes zoster vaccine, given the medical and physiological diversity in the elderly population for whom the vaccine is indicated1 and since there are stringent storage and handling requirements for this live-attenuated vaccine.7 Moreover, a large observational study also allows for exploration of vaccine benefits among important patient subgroups that would be unfeasible in most randomized controlled trials. We report the results of a large observational study among community-dwelling adults aged 60 years or older evaluating the risk of herpes zoster after herpes zoster vaccination in a managed care organization.
The study was conducted among members of Kaiser Permanente, Southern California (KPSC), and the protocol was reviewed and approved by the KPSC institutional review board, which waived requirement for informed consent. KPSC is an integrated health care system that provides comprehensive prepaid health services for its 3.2 million members. Members are socioeconomically diverse and more than 99% are community dwelling. Patient data regarding demographics, services, and diagnoses are tracked in the KPSC electronic health record databases from the outpatient, emergency department, and hospital settings. The databases provide a primary diagnosis for hospitalizations; secondary hospital diagnoses, as well as diagnoses from outpatient and emergency department visits, are not rank ordered. Data regarding care from non-KPSC providers are likely to be captured in KPSC databases because documentation is required for reimbursement of such services. Vaccinations received by members are tracked in the Kaiser Immunization Tracking System,11 regardless of whether administered in or outside of KPSC. Herpes zoster vaccine is provided to KPSC members at little or no charge.
This retrospective cohort study included data from January 1, 2007, through December 31, 2009. The vaccinated cohort consisted of KPSC members who received herpes zoster vaccine at age 60 years or older, from January 1, 2007, through June 30, 2009; the date of vaccination was termed the index date. The unvaccinated cohort consisted of randomly sampled members who were matched 3:1 to the vaccinated cohort based on birth date (±1 year). Unvaccinated individuals were assigned the same index dates as the matching vaccinated persons. Participants in both cohorts were limited to individuals with continuous membership and drug benefits for the 1 year before their index date, and were observed for first occurrence of herpes zoster, termination of KPSC membership, or study completion. Individuals with herpes zoster codes during the prior 6 months were excluded from analysis.
Herpes zoster vaccine is not recommended for immunocompromised patients.7 The study therefore excluded immunocompromised patients so that the vaccinated and unvaccinated cohorts would be comparable in terms of underlying herpes zoster risk and so that the reduced risk associated with herpes zoster vaccine could be interpreted in the context of national recommendations and the SPS.5,7 The study defined immunocompromised individuals as those with human immunodeficiency virus, leukemia, or lymphoma diagnoses within 1 year before the index date until the end of follow-up, or having immunosuppressing agents dispensed within 1 year before the index date (eBox).
Incident herpes zoster and ophthalmic herpes zoster were defined by International Classification of Diseases, Ninth Revision (ICD-9) codes (053.XX and 053.2X, respectively; in any position) from hospital, outpatient, and emergency department settings during the study period. We categorized individuals to assess factors associated with vaccination and risk of herpes zoster including sex, health care utilization, and comorbid chronic diseases. Race (self-reported) was also included in the analysis because herpes zoster vaccine coverage and risk of herpes zoster both likely differ by race.7,12 Health care utilization was defined as the number of hospitalizations or outpatient or emergency department visits within 1 year before the index date, and chronic diseases were defined as 1 or more diagnoses for diabetes or for heart, lung, kidney, or liver disease within 1 year before the index date. Patients could be assigned multiple conditions.
Incidence was calculated by dividing the number of herpes zoster cases by the total number of person-years. The 95% confidence intervals (CIs) were estimated assuming occurrence of herpes zoster follows a Poisson distribution. Hazard ratios (HRs) and 95% CIs were estimated for age, sex, race, and chronic comorbid diseases using Cox proportional hazards regression models. These models were also used to assess the association between vaccination and sex, race, health care utilization, and chronic comorbid diseases. Our sample size was sufficient to detect a relative risk of herpes zoster incidence between the vaccinated and unvaccinated cohorts of 0.85 during the study interval, with power of greater than 80% and type I error rate of .05. Significance was set at .05 based on a 2-sided test. We used SAS Enterprise Guide 4.2 (SAS Institute Inc, Cary, North Carolina) for all analyses.
Individuals who received herpes zoster vaccine in the study population may have differed from those who did not in their underlying risk for herpes zoster or in their ability and desire to access care for herpes zoster. To assess whether our results might have been distorted by such confounders, rate ratios for 13 acute symptomatic conditions in the vaccinated vs age-matched unvaccinated cohorts were estimated, as we did for herpes zoster (eTable 1). Because herpes zoster vaccine should not protect against these conditions, we reasoned that these values should cluster around 1.0 and that any deviation from 1.0 would alert us to biases that might be confounding results regarding herpes zoster. This strategy is analogous to approaches used to assess validity of estimates of vaccine effectiveness and other outcomes.13- 18
To conduct this analysis, conditions were selected that were likely to be comparable to herpes zoster in terms of access, health care seeking, and documentation, but with no plausible association with herpes zoster itself (eg, Bell palsy). We identified conditions that were acute and self-limited, with disturbing symptoms that would typically trigger efforts to seek care in outpatient primary care settings. We sought conditions with high incidence among elderly individuals but reasonably balanced across sex, race, and socioeconomic status; they were conditions that are easily diagnosed and specifically coded with 1 or few categories of ICD-9 codes. The conditions were selected without first estimating their incidence in the vaccinated and unvaccinated cohorts.
The incidence of the 13 indicator conditions was ascertained using the relevant ICD-9 codes (eTable 1) in any diagnostic position of the outpatient and emergency department files of the KPSC databases, from the index date through follow-up. Cases with codes for indicator conditions during the prior 6 months were excluded from analysis.
There were 75 761 vaccinated and 227 283 unvaccinated individuals included in the study. Compared with the unvaccinated cohort, individuals in the vaccinated cohort were more likely to be white, women, and to have had a larger number of outpatient visits, but fewer emergency department visits and hospitalizations in the 12 months prior to the index date (Table 1). The vaccinated cohort had a lower prevalence of chronic diseases.
There were 5434 herpes zoster cases identified in the study: 4593 (84.5%) in outpatient settings, 574 (10.6%) in emergency departments, and 267 (4.9%) hospitalized. Among the unvaccinated and vaccinated cohorts, follow-up averaged 1.56 and 1.72 years, yielding herpes zoster incidence of 13.0 (95% CI, 12.6-13.3) and 6.4 (95% CI, 5.9-6.8) per 1000 person-years, respectively, in univariate analysis (Table 2).
Among unvaccinated individuals, herpes zoster incidence varied by several factors. It increased with age (≥80 years vs 60-64 years: HR, 1.45; 95% CI, 1.30-1.63), was lower in men (HR, 0.75; 95% CI, 0.70-0.79), and was also lower in black individuals (HR, 0.69; 95% CI, 0.62-0.76). It varied by chronic disease, being higher in individuals with lung disease (HR, 1.34; 95% CI, 1.13-1.59) compared with those without. An increase in incidence among individuals with kidney disease (HR, 1.04; 95% CI, 0.95-1.13) and heart disease (HR, 1.06; 95% CI, 0.97-1.16) was not significantly significant. Differences in herpes zoster incidence remained when we adjusted for sex, race, chronic diseases, and health care utilization. In addition, the risk of herpes zoster was associated with number of outpatient visits during the year before the index dates, an indicator of health care–seeking behavior. The adjusted HRs were 1.94 (95% CI, 1.62-2.31) for 1 to 4 outpatient visits, 2.44 (95% CI, 2.05-2.91) for 5 to 10 visits, and 3.03 (95% CI, 2.55-3.61) for 11 or more outpatient visits. There was no association between hospitalizations or emergency department visits and herpes zoster.
The Kaplan-Meier plot showing cumulative risk of herpes zoster in the vaccinated and unvaccinated cohorts is shown in Figure 1. In the fully adjusted analysis, vaccination was associated with reduced risk of herpes zoster (HR, 0.45; 95% CI, 0.42-0.48); the reduction in risk did not vary by age at vaccination (Table 3), sex, race, or with presence of chronic diseases. Herpes zoster vaccine recipients had reduced risks of ophthalmic herpes zoster (HR, 0.37; 95% CI, 0.23-0.61) and hospitalizations coded as herpes zoster (HR, 0.35; 95% CI, 0.24-0.51). In a secondary analysis, definition of immunocompetent was further restricted by excluding individuals having documentation of at least 1 corticosteroid dispensed within 1 year before the index date, including those taken by mouth, injection, nasal spray, or inhalation. The HR in this fully adjusted analysis was 0.46 (95% CI, 0.43-0.49).
The incidence of 13 acute symptomatic indicator conditions was compared in the vaccinated and age-matched unvaccinated cohorts (Figure 2 and eTable 2). The adjusted rate ratios for the 13 conditions ranged from 0.76 to 1.38 (mean, 1.05; SD, 0.19), being 1.0 or greater for 7 of the 13 conditions. None of the conditions had adjusted rate ratios as low as 0.45, the rate ratio for herpes zoster.
Our data complement the results of the original clinical trial of herpes zoster vaccine,5 indicating that the vaccine was associated with a reduced risk of herpes zoster in a community setting with its mixed population and routine clinical practices. In addition, our data indicate that the vaccine was associated with reduced risks of ophthalmic herpes zoster and hospitalizations potentially attributable to herpes zoster. The results also suggest that these potential benefits extended to individuals of all ages for whom herpes zoster vaccine is recommended, and to individuals with chronic diseases.
Overall, we found that herpes zoster vaccine was associated with a 55% reduction in incidence of herpes zoster, which is consistent with the 51% vaccine efficacy reported from the SPS. These results would be consistent with an absolute reduction in herpes zoster risk of 1.4% at 30 months of follow-up; alternatively, 1 episode of herpes zoster would be averted for every 71 individuals receiving the vaccine. In contrast to the SPS, however, we found that the lower herpes zoster risk associated with the vaccine was retained across all age strata (P = .62).5,19 Thus, our results support recommendations to offer herpes zoster vaccine to eligible patients of all ages including the oldest population. Not only might these patients experience a reduction in their relative risk of herpes zoster, but for the oldest group, this could translate into a very large absolute reduction in disease because they bear the greatest burden of herpes zoster and postherpetic neuralgia and are also especially vulnerable to these disabling conditions.1,20,21
The efficacy of herpes zoster vaccine at preventing ophthalmic herpes zoster was not assessed in the SPS. Our finding that the vaccine recipients had a reduced risk of these episodes was therefore particularly important. Ophthalmic involvement is a common manifestation of herpes zoster and it can lead to serious vision-threatening sequelae.2,4,22- 24
Our study provides new information regarding the risk of herpes zoster in key vaccinated and unvaccinated subgroups. Herpes zoster vaccine was associated with a similar reduction across sexes and racial groups. Although we confirm results of other studies showing that black persons are at lower risk of herpes zoster than white persons,3,25,26 incidence among black individuals remains considerable and prevention in this population is important because racial disparities in access to care may hinder the complicated management of herpes zoster and postherpetic neuralgia. Our data demonstrating disparities in herpes zoster vaccine uptake, consistent with other national data,12,27,28 highlight the need to offer this vaccine to all racial and ethnic groups. We also found that individuals with certain chronic diseases were at increased underlying risk of herpes zoster, but that herpes zoster vaccine was associated with a reduced risk in these individuals too. This is reassuring information because it was plausible that these diseases might have interfered with functional immunity and vaccine effectiveness.29 Control of pain from herpes zoster and postherpetic neuralgia is complicated in these patients because of their underlying conditions and the medications they must take. Herpes zoster vaccine provides an opportunity to prevent herpes zoster and the clinical challenge of managing its symptoms.
Observational studies can be subject to biases because patients who seek the vaccine may differ in their underlying risk of disease or in their ability and desire to access care for the condition. Such biases have, for instance, confounded interpretation of results from studies of influenza vaccine in elderly individuals because those at greatest risk of hospitalization or death from influenza may be least likely to be offered the vaccine.8- 10 The potential for bias in studies of herpes zoster vaccine is particularly large because the risk factors that distinguish immunocompetent individuals who develop herpes zoster and who receive herpes zoster vaccine are not known.3 The marked heterogeneity in physiology, fitness, and health care seeking that exists in elderly individuals targeted for herpes zoster vaccine could potentially hide these important yet unrecognized risk factors. We assessed the association between vaccination and 13 indicator conditions in the vaccinated and unvaccinated study cohorts, just as we had for herpes zoster. The adjusted rate ratios for all 13 conditions were considerably greater than that for herpes zoster; in fact, 7 adjusted rate ratios were significantly greater than 1.0. A possible explanation for this excess risk is that individuals receiving herpes zoster vaccine are generally more inclined or better able to obtain medical care for acute conditions. This interpretation is consistent with our observation that herpes zoster vaccine recipients had more outpatient encounters. If correct, herpes zoster vaccine recipients might be more likely to seek care for episodes of herpes zoster as well, making our estimates of reduced risk associated with herpes zoster vaccine conservative.
Our study had limitations. It included a fully insured population in 1 region of the country; therefore, the results need to be generalized carefully. It was not designed to assess severity or duration of symptoms among herpes zoster cases or to assess the effectiveness of herpes zoster vaccine at preventing postherpetic neuralgia. In the SPS, herpes zoster vaccine was associated with less pain and discomfort among individuals in whom herpes zoster developed and reduced the incidence of postherpetic neuralgia by 66.5%.5 Average length of follow-up in our study was short and it was not designed to capture any decline in protection that is likely to occur with time.30 Misclassification was possible because we ascertained study outcomes and variables using passive follow-up through health care encounters and using electronic health records. Such data have proven fairly reliable for detecting herpes zoster, although their role in detecting ophthalmic involvement has not been adequately explored.4,31- 33 It would, however, seem that misclassification of vaccination or herpes zoster status is most likely to result in an underestimate of the reduction in risk associated with herpes zoster vaccine. The prevalence of chronic diseases and number of hospitalizations prior to the index date were higher in the unvaccinated cohort; we adjusted for these factors and the magnitude of the difference was small, so they should not affect our conclusions substantially.
Regarding chronic disease status, our results were unchanged when we expanded our definition to include individuals with at least 1 relevant diagnosis at any time during follow-up (data not shown). Although our conclusion that herpes zoster vaccine was associated with fewer herpes zoster–coded hospitalizations is important, it should be interpreted cautiously because herpes zoster is often incidental to herpes zoster–coded hospitalization rather than its cause.7,34 Furthermore, although we controlled for age and excluded immunosuppressed individuals from our study, the risk of hospitalization due to herpes zoster is so strongly associated with these factors that residual confounding is possible.4,7,20 A source of potential confounding we could not address in this study was family history of herpes zoster, which could predispose family members to herpes zoster, encourage them to seek vaccination, or both.35 Such a bias would result in an underestimate of reduced risk associated with the vaccine.
In conclusion, we found that individuals aged 60 years or older who received herpes zoster vaccine had a reduced risk of herpes zoster regardless of age, race, and presence of chronic diseases. Herpes zoster vaccine was licensed recently, which means the durability of its protection needs to be assessed in future studies. Meanwhile, however, this vaccine has the potential to annually prevent tens of thousands of cases of herpes zoster and postherpetic neuralgia nationally.7 To date, herpes zoster vaccine uptake has been poor due to weaknesses in the adult vaccine infrastructure and also due to serious barriers to the vaccine among clinicians and patients.12,36 Solutions to these challenges need to be found so that individuals seeking to receive herpes zoster vaccine will be able to reduce their risk of experiencing this serious condition.
Corresponding Author: Hung Fu Tseng, PhD, MPH, Department of Research and Evaluation, Kaiser Permanente, Southern California Medical Group, 100 S Los Robles Ave, Second Floor, Pasadena, CA 91101 (Hung-Fu.x.Tseng@kp.org).
Author Contributions: Dr 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.
Study concept and design: Tseng, Smith, Harpaz, Bialek, Jacobsen.
Acquisition of data: Tseng, Smith.
Analysis and interpretation of data: Smith, Harpaz, Bialek, Sy, Jacobsen.
Drafting of the manuscript: Tseng, Harpaz, Bialek.
Critical revision of the manuscript for important intellectual content: Smith, Harpaz, Bialek, Sy, Jacobsen.
Statistical analysis: Tseng, Smith, Harpaz.
Obtained funding: Jacobsen.
Administrative, technical, or material support: Tseng, Bialek, Jacobsen.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Drs Tseng, Smith, and Jacobsen and Ms Sy report having received research funding from Merck for other vaccine studies. Dr Jacobsen reports having served as an unpaid consultant for Merck Research Laboratories. Drs Harpaz and Bialek report no disclosures.
Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention, US Department of Health and Human Services.