Cost-Effectiveness of a Vaccine to Prevent Herpes Zoster and Postherpetic Neuralgia in Older Adults
Michael B. Rothberg
Kenneth J. Smith
Division of General Medicine, University of Pittsburgh
Tufts University School of Medicine
Division of General Medicine and Geriatrics, Department of Medicine, Baystate Medical Center
Background. A vaccine to prevent herpes zoster was recently approved by the United States Food and Drug Administration. We sought to determine the cost-effectiveness of this vaccine for different age groups. Methods. We constructed a cost-effectiveness model, based on the Shingles Prevention Study, to compare varicella zoster vaccination with usual care for healthy adults aged 160 years. Outcomes included cost in 2005 US dollars and quality-adjusted life expectancy. Costs and natural history data were drawn from the published literature; vaccine efficacy was assumed to persist for 10 years. Results. For the base case analysis, compared with usual care, vaccination increased quality-adjusted life expectancy by 0.0007-0.0024 quality-adjusted life years per person, depending on age at vaccination and sex. These increases came almost exclusively as a result of prevention of acute pain associated with herpes zoster and postherpetic neuralgia. Vaccination also increased costs by $94-$135 per person, compared with no vaccination. The incremental cost-effectiveness ranged from $44,000 per quality-adjusted life year saved for a 70-year-old woman to $191,000 per quality-adjusted life year saved for an 80-year-old man. For the sensitivity analysis, the decision was most sensitive to vaccine cost. At a cost of $46 per dose, vaccination cost !$50,000 per quality-adjusted life year saved for all adults 160 years of age. Other variables related to the vaccine (duration, efficacy, and adverse effects), postherpetic neuralgia (incidence, duration, and utility), herpes zoster (incidence and severity), and the discount rate all affected the cost-effectiveness ratio by 120%. Conclusions. The cost-effectiveness of the varicella zoster vaccine varies substantially with patient age and often exceeds $100,000 per quality-adjusted life year saved. Age should be considered in vaccine recommendations. Herpes zoster (shingles) is a dermatomal neurocutaneous disease caused by reactivation of the varicella zoster virus. The lifetime risk of herpes zoster ranges from 10% to 30% [1, 2], and the severity of both the disease and its sequelae increase with age . Because it causes acute burning pain, herpes zoster can seriously impair physical, social, and emotional functioning . Acute herpes zoster has been estimated to cost 1$400 per case in physician's fees, antiviral therapy, and lost wages ; the total cost of herpes zoster-related hospitalizations in 1 year exceeded $8 million in Connecticut alone . In addition, 10%-70% of patients
with herpes zoster develop postherpetic neuralgia
(PHN), which persists for 11 year in up to one-half of
patients 155 years of age [4, 7], at a cost exceeding
$500 per patient .
Antiviral therapy is cost-effective for decreasing the
symptoms of herpes zoster , and it has been shown
to reduce the duration , but not the incidence ,
of PHN. Unfortunately, topical analgesics, opioids,
tricyclic antidepressants, nerve blocks, neuroactive drugs,
and steroids all have limited success in treating the
chronic and often debilitating pain associated with the
disease [1, 11].
The Shingles Prevention Study (SPS) 
demonstrated that live attenuated Oka/Merck varicella zoster
vaccine protected against herpes zoster and PHN and
reduced the symptoms of acute herpes zoster. Although
the vaccine is effective, its cost-effectiveness has been
debated. Because herpes zoster incidence varies by age
and sex, the cost-effectiveness of the vaccine should not
be uniform for all patients 160 years of age (the SPSs
target population). Moreover, the SPS lasted only 4
years, so the vaccines long-term efficacya potentially
important determinant of cost-effectivenesswill not be known
for some time. We examined the pharmacoeconomics of the
varicella zoster vaccine using a decision analytic model to
explore a variety of patient and vaccine scenarios.
We constructed a Markov model  using standard computer
software (Decision Maker software, version 7.07; Pratt Medical
Group) to compare the cost-effectiveness of vaccination with
the varicella zoster vaccine with no vaccination for healthy
immunocompetent adults 160 years of age. The study was
conducted from a societal perspective. Costs (expressed in 2005
US dollars) and utilities (expressed in quality-adjusted life years
[QALYs]) were both discounted at 3% annually.
The entire cohort begins in the unvaccinated state, with
individuals undergoing vaccination at the beginning of the first
year. For each year, an age- and sex-specific underlying attack
rate is applied to calculate the number of cases, major
complications, and deaths due to herpes zoster. Vaccinated
individuals experience a decrease in cases, complications, burden
of illness, PHN, and death proportional to vaccine efficacy,
which first wanes and then ceases at the maximal duration.
Patients experiencing herpes zoster or its complications incur
both costs and utility decrements.
In addition to the decision model, we analyzed the
costeffectiveness of the vaccine during the SPS, assuming that all
costs and benefits accrued within the study period (with the
exception of PHN, which could extend for up to 10 years). We
calculated the cost of vaccination and then subtracted the
savings from avoiding herpes zoster and PHN. We also calculated
the difference in quality-adjusted life expectancy by assigning
utility values to avoided cases of herpes zoster and PHN on
the basis of our model inputs.
Data and Assumptions
Baseline estimates and ranges for sensitivity analyses are
provided in table 1.
Zoster incidence and complications. The model
incorporates age- and sex-specific incidence from the largest
observational cohort study . Rates were similar to those
published 10 years earlier  and to rates observed in the SPS
for subjects 70 years of age, but they were lower than those
observed for subjects !70 years of age . Age-specific burden
of illness in the first 6 months following infection and the rate
of PHN (defined as pain persisting 3 months after the onset
of illness) were based on the SPS . Duration of PHN [7,
22] and age-specific rates for complications not reported in the
SPS were drawn from observational studies [15, 36] and
government databases [6, 20, 21]. Complications included death,
hospitalization for disseminated herpes zoster or herpes zoster
meningitis, ocular involvement and monocular blindness, and
herpes zoster oticus and monaural deafness.
Vaccine efficacy and adverse effects. The SPS reported 3
efficacy measures, stratified by age: herpes zoster incidence,
burden of illness (BOI), and PHN incidence . BOI and
PHN efficacy were reported for the entire population, not just
for those who developed herpes zoster; thus, these measures
incorporated the decreasing herpes zoster incidencein effect,
double counting. To see if there was efficacy beyond the
decrease in herpes zoster incidence, we calculated the efficacy per
case of herpes zoster by dividing the cases of PHN and the
burden of illness scores by the number of cases of herpes zoster
Efficacy in decreasing herpes zoster incidence varied with
age, from 65% among subjects aged 5964 years to 8% in
subjects aged 85 years. Vaccination also appeared to prevent
PHN during the first year following vaccination, but it had no
statistically significant additional efficacy in subsequent years
(table 2). Moreover, this apparent efficacy was due to placebo
recipients experiencing PHN at a significantly higher rate
during the year following vaccination, relative to subsequent years,
rather than to any reduction in PHN among vaccinees. We
therefore assumed no reduction in PHN beyond that afforded
by reducing the incidence of herpes zoster illness. In contrast,
the vaccine demonstrated efficacy in reducing the severity per
case in patients aged 70 years (BOI score, 225 vs. 160 in the
placebo and vaccination groups), but not in patients aged !70
years (BOI score, 134 vs. 128 in the placebo and vaccination
In their US Food and Drug Administration briefing, the
manufacturer of the vaccine provided data for each of the 3
efficacy measures, stratified by years after vaccination .
Efficacy waned in the first year after vaccination and then
appeared to level off, but there were only 4 data points (1 for
each year of the trial). We fit a declining logarithmic function
for efficacy against zoster incidence (y p 0.1018 ln(x) +
0.5903; R 2 p .56) and then calculated age-specific efficacy for
each year using age-specific ORs derived from US Food and
Drug Administration data. In sensitivity analysis, we tested
other functions, ranging from no decrease after year 4 to
complete lack of efficacy following the study period. We assumed
no decrease in efficacy for decreasing severity of illness per case.
The most common adverse effect was arm soreness, but 0.7%
of vaccinees experienced a serious reaction not experienced
by participants who received placebo (95% CI, 0.1%1.3%).
In the absence of any other details, we assumed that serious
reactions would be equivalent to spending 3 days in the
Utilities. SPS participants used the worst-pain component
of the herpes zoster brief pain inventory to record their pain
daily for up to 182 days, generating an area under the pain
Value at baseline
Herpes zoster illness
Annual incidence per 1000 persons
Any ophthalmic complications
Monocular blindness, given ophthalmic complications
Monaural deafness, given herpes oticus
PHN, given herpes zoster, by age
Proportion of PHN lasting 12 months, by age
Hospitalization, given herpes zoster
Death due to herpes zoster,
death per 100,000 persons
Duration of hospitalization, mean days
Duration of PHN, yearsc
Efficacy against herpes zoster illness
Likelihood ratio for vaccine efficacy, by age
Efficacy in decreasing burden of illness, by age
Serious adverse effect
PHN after 6 months
Short term morbidities, QALYs
Acute herpes zoster, by age
Costs, US dollars
Acute herpes zoster
Indirect costs, by age or condition
[7, 12, 19]
[7, 12, 19]
Value at baseline
NOTE. Data are probabilities, unless otherwise indicated. PHN, postherpetic neuralgia; QALY, quality-adjusted life year.
a Baseline value varies with age.
b For parameters that vary by patient age or year since vaccination, we conducted sensitivity analysis by multiplying the value for each
patient by a constant factor. Probabilities were converted into odds, then multiplied by the factor and converted back into probabilities.
c If 112 months.
d 0.59030.1018 ln(vaccine year). Efficacy also varies by age. Each year, this result is converted to odds, multiplied by the age-specific
likelihood ratio, and then converted back to efficacy.
e Utilities range from 1.0, signifying perfect health, to 0, signifying death. Utilities in the model are adjusted for age by multiplying the
utility in the table by the mean utility for a patient of that age (e.g., a 70-year-old patient has a baseline utility of 0.836. With monocular
blindness, the utility would be 0.92 0.836 p 0.769).
f Cost per dose, assuming the purchase of a 10-pack of vaccine.
curve (BOI score range, 01820). We transformed BOI scores
into utilities on the basis of results of another study  that
compared the area under the curve of the worst-pain
component of the herpes zoster brief pain inventory and the
EuroQOL-5D, a validated measure of patient utilities . In
that study, herpes zoster brief pain inventory scores over 35
days were linearly correlated with utilities (utility p 0.1001
BOI score + 95.767, R 2 p .95). We multiplied average BOI
scores from the vaccine trial by this formula and then multiplied
by 365/35 to calculate the number of QALYs lost. The utility for
mild, moderate, and severe PHN was based on a survey of PHN
involving 385 elderly patients . The proportion of patients
experiencing mild, moderate, and severe symptoms came from
the only population-based study reporting severity at 1 year .
All utilities were adjusted for age .
Costs. At least 7 economic analyses have measured costs
associated with herpes zoster or PHN [5, 8, 3134, 39]. We
chose our direct cost estimate for acute herpes zoster illness
from the largest of these  but tested the results of all trials
in sensitivity analysis. Age-specific indirect costs, primarily
patient and caregiver lost wages, came from a prospective
population-based study . The costs of PHN [8, 39] and ocular
complications of herpes ophthalmicus  were based on
antiviral trials. The cost of treating herpes oticus included an
initial consult, audiometry, treatment with prednisone, and a
follow-up visit. Hospitalization costs came from the Healthcare
Utilization Project Web site , using national cost-to-charge
ratios. Vaccine costs assumed the purchase price for a 10-pack
of vaccine , plus 5 min of a nurses time for vaccination
. We did not assign a cost to adverse effects of the vaccine.
Costs were updated to 2005 US dollars .
Base Case Analysis
On the basis of the results of the SPS, without modeling vaccine
efficacy beyond the study period or discounting, the cost to
vaccinate the entire study population against herpes zoster was
$2.9 million, which prevented 327 cases of herpes zoster and
53 cases of PHN and extended quality-adjusted life expectancy
PHN per case
of herpes zoster
6069 years old (n p 10,370) 1,543,367
70 years old (n p 8884)
by 22.4 years (71% of which came from preventing PHN that
pitalizations and mortality accounted for !5% of the vaccine
lasted 16 months) (table 3). Savings attributable to preventing
benefit, even in the oldest age group.
herpes zoster and PHN offset 12% of the cost of vaccination.
The cost per QALY gained by vaccination ranged from $201,000
for patients aged 6069 years to $75,000 for patients aged
In the decision model, we assumed that vaccine efficacy
persisted beyond the study period and we considered healthy men
and women of 3 different ages (table 4). At all ages, vaccination
resulted in the greatest quality-adjusted life expectancy. Overall
benefits (range, 0.00070.0024 QALYs per person) came almost
exclusively from improvements in quality of life. Depending
on patient age, 50%65% of the difference in quality-adjusted
life expectancy was attributable to preventing PHN, whereas
35%50% came from preventing acute herpes zoster. The
vaccine was less effective in reducing PHN in our analysis than in
Vaccination increased total health expenditures in all groups
by $94$135 per person, compared with no vaccination.
Incremental costs increased with age, because productivity gains
associated with preventing acute herpes zoster partially offset
the cost of the vaccine for younger patients. The incremental
cost to extend life by 1 QALY ranged from $44,000 for a
70year-old woman to $191,000 for an 80-year-old man. Disease
incidence increased with age, but vaccine efficacy fell, producing
a U-shaped curve for cost-effectiveness (figure 1). As a result,
vaccination was most cost-effective for patients aged 70 years
and, at every age, vaccination was more cost-effective for
women than it was for men.
the SPS, because we did not model a reduction in PHN beyond
We varied all the parameters through the ranges presented in
the reduction in herpes zoster incidence. Reductions in
Female 70 Years 80 Years Male
Total cost, US$
per QALY saved
NOTE. QALE, quality-adjusted life expectancy; QALY, quality-adjusted life years.
cination by 20% included variables related to the vaccine
(duration, cost, efficacy, and side effects), PHN (incidence,
duration, and utility), herpes zoster (incidence and severity), and
the discount rate.
Vaccine variables. The model was most sensitive to vaccine
cost (figure 2). At a cost of $46 per dose, vaccination had an
incremental cost-effectiveness ratio !$50,000 per QALY for all
adults aged 6080 years. Duration of immunity was important
for younger patients, who were unlikely to develop herpes
zoster immediately but who had a long life expectancy. If efficacy
lasted only 6 years, the cost to vaccinate a 60-year-old patient
was 1$200,000 per QALY saved (figure 3), but it improved out
to 25 years. For patients aged 70 years, cost-effectiveness
curves were flat beyond 10 years. Variation in vaccine efficacy
within the 95% CI had modest effects (10%) on the
costeffectiveness. If, however, the vaccine was able to prevent an
additional 33% of cases of herpes zoster in patients 169 years
of age, the cost-effectiveness ratio for 80-year-olds would be
reduced by one-half.
Disease variables. Because preventing PHN comprises the
majority of the vaccines benefit, the model was sensitive to
increases in the incidence, duration, or severity of PHN.
Vaccination cost !$50,000 per QALY for 70-year-old men if the
utility of PHN was !0.72 and it cost !$50,000 per QALY for
60-year-old women if the utility was !0.64 and the probability
exceeded 8.5% (figure 4). For 80-year-old patients, no
combination of utility and probability produced a cost-effectiveness
ratio below that threshold. Other sequelae were too rare to have
a substantial impact.
This cost-effectiveness analysis demonstrates that, although the
varicella zoster vaccine improved quality-adjusted life
expectancy for all adults 60 years of age, the benefits of vaccination
were not evenly distributed. Although the vaccine is most
effective for younger patients, most cases of herpes zoster and
PHN occur among older patients. Consequently, vaccination
is most cost-effective for patients aged 70 years. At any age, it
is more cost-effective to vaccinate women than men, because
women experience a disproportionate amount of herpes zoster
and PHN. Although the cost-effectiveness for 70-year-old
women is attractive, compared with other commonly accepted
interventions, the cost-effectiveness in other groups often
exceeds $100,000 per QALY, an amount that would be considered
expensive. At $150 per dose, the vaccine does not appear to
be cost-saving under any assumptions.
Two formal analyses have explored the cost-effectiveness of
varicella zoster vaccine. Because both were conducted before
vaccine pricing, neither estimates the actual cost-effectiveness
of the vaccine. The first analysis, performed by Edmunds et al.
 before the SPS, concluded that the vaccine would be
costeffective for patients aged 65 years at a cost of $148 if the
vaccine was 60% effective and lasted at least 10 years. The
second, conducted by Hornberger and Robertus  before
vaccine licensure, concluded that, at a price of $50, the vaccine
would have a cost-effectiveness of !$50,000 per QALY gained
and, at a price of $200, would cost 1$100,000 per QALY gained.
Both of these conclusions are compatible with our findings.
Our results diverge regarding the effects of age. Edmunds et
al.  accurately modeled increasing incidence and morbidity
of herpes zoster with age but assumed equal vaccine efficacy
at all ages. Consequently, they found vaccination to be most
cost-effective at age 80 years. In contrast, Hornberger and
Robertus  correctly modeled the differential efficacy by age,
but they assumed risk of PHN to be the same at all ages and
thus found vaccination to be most cost-effective for patients
aged 60 years. Incorporating age-related changes in both
parameters, we observed a U-shaped curve for cost-effectiveness.
Both private insurers and Medicare have grappled with
paying for varicella zoster vaccine . There are 61 million
Americans aged 60 years. At the bulk price of $145.45 per
vaccination, vaccinating all eligible Americans would cost $9
Figure 3. Cost-effectiveness of vaccination against herpes zoster in
women as a function of duration of vaccine efficacy. Curves for men (not
shown) are similar, but they shift upwards. QALY, quality-adjusted life
billion. There is no doubt that the vaccine is effective, but would
this enormous investment represent good value? The answer is
yes and no. The vaccine is much more cost-effective for some
recipients than for others. Vaccinating a 70-year-old woman
costs the same as vaccinating a 60-year-old man, but it offers
3 times the health benefit. Using a threshold of $50,000 per
QALY, only 70-year-old women would be eligible for
vaccination. At a more liberal threshold of $100,000 per QALY, men
aged 6575 years and women aged 6075 years would be eligible
to receive the vaccine. By comparison, influenza virus and
pneumococcus vaccination are generally cost-saving in the
elderly population [43, 44], and varicella vaccine, which costs
$66, appears to be cost-saving in children . Less
cost-effective vaccines, such as meningococcal  and pneumococcal
 conjugate vaccines, have been accepted into the childhood
vaccination schedule, but not all states universally cover their
costs. Notably, a reduction in the cost of the varicella zoster
vaccine would have a dramatic effect on the value it represents.
At a price of $46, it could be universally recommended as being
cost-effective and sometimes cost-saving.
Our analysis has several limitations. First, although many of
our inputs were based on large studies that provided stable
estimates, a few key values are based on limited information
or vary widely among studies. A single study of 127 patients
provided both the percentage of patients who have symptoms
of PHN for 11 year and the proportion of long-standing
symptoms that are mild, moderate, and severe . We explored the
effects of varying these inputs, but larger, long-term studies of
both the probability and severity of PHN at 1 year are necessary
to reliably determine the cost-effectiveness of the vaccine.
Second, it is not known how long vaccine efficacy will last, and
it will be at least 5 years before we know whether our 10-year
assumption was correct. Measures of cell-mediated immunity
show the vaccines protection to have a half-life of 56 months
, leaving little efficacy after 1015 years. Third, the
incidence of herpes zoster observed for subjects aged 6069 years
in the SPS was higher than that observed by Insinga et al. ,
although both are contemporaneous, multicenter studies. The
discrepancy could represent underdiagnosis in the
observational cohort or a recent increase in cases among younger age
groups attributable to lack of boosting that followed widespread
childhood vaccination against primary varicella infection .
Although increased herpes zoster incidence improves the
costeffectiveness of vaccination, even given a 40% increase among
60-year-old individuals, vaccination would still cost $84,000 per
QALY. Finally, our estimates for the cost of PHN are based on
data that is 110 years old, albeit updated to 2005 US dollars.
Newer, more expensive drugs and treatments may have
increased the cost and decreased the morbidity associated with
PHN. Additional studies would be welcome.
Having successfully conquered many of the major infectious
diseases afflicting Western countries, vaccine manufacturers
have turned to rarer or less harmful diseases. As the potential
number of childhood and adult vaccines mushrooms, both
policy makers and payers struggle with decisions about which
ones to recommend or cover. The Institute of Medicine
(Washington, DC) has recommended that the Advisory Committee
on Immunization Practices incorporate societal benefits and
costs into its vaccine coverage decisions , and
cost-effectiveness analysis offers one tool with which to do this .
Trying to pay for all available interventions, regardless of cost,
strains budgets and causes inequity, because the rising cost of
care forces some patients out of the health care system entirely.
Considering both costs and benefits, the varicella zoster vaccine
does not seem to represent a good value for all patients 160
years of age, although it may be reasonable for patients 70 years
old, depending on the cost-effectiveness criterion used.
Lowering the cost to that of the varicella vaccine would make it
competitive with other generally accepted interventions.
Potential conflicts of interest.
Doris Duke Clinical Scientist Development Award All authors: no conflicts.