American Journal of Obstetrics and Gynecology (2005) 193, 12749

www.ajog.org

Acyclovir prophylaxis for pregnant women with

a known history of herpes simplex virus:
A cost-effectiveness analysis
Sarah E. Little, AB, Aaron B. Caughey, MD, MPP, MPH
Department of Obstetrics, Gynecology, and Reproductive Sciences, University of California, San Francisco, CA
Received for publication March 1, 2005; revised April 29, 2005; accepted May 9, 2005

KEY WORDS
Neonatal herpes
simplex virus
infection
Acyclovir
Antiviral therapy
Decision analysis
Cost-effectiveness
analysis

Objective: Previous literature has shown acyclovir to be cost-effective as prophylaxis for women
with genital symptomatic herpes simplex virus infection recurrence during pregnancy. We extend
this analysis by adding quality-adjusted life year measurements and considering women with a
diagnosed history of herpes simplex virus infection but without recurrence in pregnancy.
Study design: A decision analytic model was designed that compared acyclovir prophylaxis
versus no acyclovir for women with a history of diagnosed genital herpes simplex virus infection
but without recurrence in pregnancy. Sensitivity analysis and Monte Carlo simulations were
performed to test for robustness.
Results: We found that 22,286 women must be treated to prevent 1 neonatal death, 8985 women
to prevent 1 affected child, and 177 women to prevent 1 cesarean delivery. As compared with no
acyclovir, acyclovir prophylaxis at 36 weeks of gestation saves approximately $20 per person and
increases total quality-adjusted life years by 0.01. In univariate sensitivity analysis, this result was
robust to all reasonable probability and quality-adjusted life year estimates. Monte Carlo
simulation demonstrated acyclovir to be cost-effective 100% of the time and cost saving O99% of
the time.
Conclusion: Acyclovir prophylaxis versus no treatment for pregnant women with a diagnosed
history of genital herpes simplex virus infection but without recurrence during pregnancy is costeffective over a wide range of assumptions.
2005 Mosby, Inc. All rights reserved.

Neonatal herpes simplex virus (HSV) infection is one

of the most devastating diseases acquired during the perinatal period. In the United States, the current incidence of
approximately 11 to 33 cases per 100,000 live births has
Supported by the National Institute of Child Health and Human
Development, grant HD01262 (A.B.C.).
Presented at the 25th Annual Meeting of the Society for Maternal
Fetal Medicine, February 7-12, 2005, Reno, Nevada.
Reprints not available from the authors.
0002-9378/$ - see front matter 2005 Mosby, Inc. All rights reserved.
doi:10.1016/j.ajog.2005.05.042

not changed appreciably in the last 20 years.1 More than

50% of infants with disseminated disease die, even with
antiviral therapy; among neonates with encephalitis,
O50% of survivors are left with severe neurologic impairment.2 Because neonatal HSV is dicult to diagnose
and eorts to identify the infection earlier have not
succeeded, the incidence and clinical outcome of neonatal
HSV has changed little over the last 25 years.2 Thus,
preventing vertical transmission is the most viable way to
limit disease.

Little and Caughey

1275

Figure 1 Decision tree of acyclovir versus no acyclovir for women with a history of genital herpes infection. Women who
experience side eects incur the cost but not the benet of treatment.

Approximately 85% of early neonatal herpes infections

are acquired during delivery from virion-contaminated
maternal vaginal secretions.2 Current clinical practice is
to oer cesarean delivery to women with active genital
lesions at the time of labor.2 Cesarean delivery has been
shown to virtually eliminate the risk of vertical transmission but misses women who are shedding asymptomatically.3,4 Another strategy is to oer acyclovir prophylaxis
at 36 weeks of gestation to women who have experienced
symptomatic HSV during their pregnancy. Acyclovir has
been proved eective at reducing both symptomatic
recurrences and asymptomatic shedding.5,6 For women
with symptomatic HSV outbreaks in pregnancy, acyclovir prophylaxis has been shown to be more cost-eective
than expectant management with cesarean delivery for
lesions.7,8
However, not all women with a history of genital
HSV experience symptomatic recurrence in pregnancy.
Current clinical practice is to oer acyclovir prophylaxis
only to those women who have experienced a recurrent
outbreak during pregnancy. Our study addresses whether
it would be clinically benecial and cost-eective to oer
prophylactic acyclovir to women with a history of HSV
but no recurrence during pregnancy.

Material and methods

A decision analytic model (Figure 1) was developed with
TreeAgePro 2004 software (TreeAge Software Inc, Williamstown, Mass). There are approximately 4 million
births per year in the United States, and 5% of these
women have been diagnosed with genital HSV. Of these,
80% of the women do not experience an HSV recurrence

in pregnancy and comprise a cohort of approximately

160,000 women.5 Our model compared acyclovir prophylaxis to no prophylaxis for these women with diagnosed genital herpes but no recurrence during pregnancy.

Probabilities
The input probabilities for our decision tree are displayed in Table I. Although most studies have shown
that short-term acyclovir therapy has no side eects, we
conservatively estimate that 2% of patients will experience minor side eects and thus discontinue treatment,
thereby incurring the cost of therapy but not receiving
the benets.9 We assumed that, if lesions were present at
the time of delivery, a woman would be delivered by
cesarean. If lesions were not present, we used the current
US cesarean delivery rate. We used the US maternal
mortality rates for both vaginal and cesarean delivery.10,11 For cesarean deliveries because of lesions, we
used the mortality rate for elective cesarean deliveries.12
To calculate the baseline probability of lesions or
asymptomatic shedding, we used a study that includes
our target population and assumed 85% of genital HSV
is HSV-2.3,6,13 We used composite odds ratios determined by meta-analysis to reduce the probabilities of
either having lesions or asymptomatic shedding by 75%
and 91%, respectively, with acyclovir prophylaxis.5 For
neonatal transmission rates, we assumed no transmission with a cesarean delivery, because this is consistent
with the largest study of neonatal herpes epidemiology
to date.3 We varied transmission rate by serotype14 and
reduced transmission by 89% with acyclovir prophylaxis.5 If HSV was transmitted to the neonate, we
accounted for neurologic morbidity.

1276
Table I

Little and Caughey

Probability estimates

Table II

Cost estimates

Variable

Probability Reference

Variable

Cost

Reference

Genital HSV infections because

of HSV type 1
Probability of side effects
because of acyclovir therapy
Probability of a Cesarean Delivery:
If lesions present at delivery
If no lesions present
Maternal death
Vaginal delivery
Cesarean delivery, not for
lesions
Cesarean delivery, for lesions
Status at delivery
Lesions
HSV-1
HSV-2
Shedding asymptomatically
HSV-1
HSV-2
Probability of neonatal
transmission if shedding
Cesarean for lesions
All other delivery
Maternal HSV-1
Neonatal HSV-1
Neonatal HSV-2
Maternal HSV-2
Neonatal HSV-1
Neonatal HSV-2
Outcomes of neonatal
transmission
Moderate neurologic disability
HSV-1
HSV-2
Severe neurologic disability
HSV-1
HSV-2
Neonatal death
HSV-1
HSV-2
Effectiveness of acyclovir
Reducing lesions
Reducing asymptomatic shedding
Reducing transmission

0.150

15

$46

16
18

0.020

1.000
0.244

Assumption
10

0.000092
0.000350

11
11

Acyclovir prophylaxis
Delivery
Vaginal
Cesarean
Cesarean for lesions
Initial hospital treatment
for neonatal HSV
Lifetime treatment of child
Moderate neurologic disability
Severe neurologic disability

0.000239

12
6,13
Table III

0.0037
0.0110

4,14
0.000

0.028
0.113
0.000
0.040
14

16
$32,483
17
$349,753
$1,049,260

Life expectancy/utility estimates

Variable

0.0018
0.0055

$4,939
$9,490
$7,608

Life expectancies*
Maternal
Normal neonate
Moderate neurologic disability
Severe neurologic disability
Utilityy
Maternal
Cesarean delivery
Having an impaired child
Losing a child
Neonatal
Moderate disability
Severe disability

Value

Reference

55.4
77.2
62.0
28.7

11
11
22
22

0.99
0.81
0.92

20
21
21

0.9
0.3

23
23

* Life expectancies are in years.

y
Utilities are in utils (range, 0 to 1).

0.01
0.14
0.02
0.17
0.28
0.20
5
0.75
0.91
0.89

cost of having a child with moderate neurologic disability, we assumed that these children would be one
third as expensive, because this was the ratio used in
previous studies.7 We used an estimate of $4939 for a
vaginal delivery, $7608 for a cesarean delivery because
of lesions, and $9490 for all other cesarean deliveries.
These estimates reect cost rather than charge and
account for the fact that cesarean deliveries are less
expensive when performed before a trial a labor, as is
the case for a woman with lesions.18

Quality-adjusted life years (QALYs)

Costs
The input cost estimates for our decision tree are
presented in Table II. All costs were projected to 2005
dollars by ination with the medical component of the
consumer price index.15 For the baseline cost of acyclovir prophylaxis, we used $46, which reects cost rather
than charge.16 For the lifetime cost of having a child
with severe neurologic disability, we used the estimated
cost of a child with cerebral palsy.17 To interpolate the

We generated QALYs that account for both the neonate

and mother. The utility and life expectancy data are
presented in Table III. When calculating the QALYs, we
assumed a 3% discount rate.19 To calculate total
QALYs, we added both maternal and neonatal QALYs.
We used the US average maternal age and life expectancy18 and assumed that women would slightly prefer
to have vaginal deliveries.20 If the child died, maternal
utility decreased to 0.92; this is the utility estimated for

Little and Caughey

women who experience miscarriage.21 If the child had
either moderate or severe neurologic impairment, maternal utility decreased to 0.81; this is the estimated
utility for having a child with Down syndrome.21 We
used the average US life expectancy for a normal
neonate.10 For a neonate with neurologic disability, we
used the life expectancy for a child with either severe or
moderate cerebral palsy.22 For the utility of being a
child with either severe or moderate neurologic disability, we used utility estimates from Saigal et al.23 For the
denition of a child with moderate neurologic impairment, we used a child who can perform the activities of
daily living on his/her own, is free of pain, and performs
schoolwork more slowly than his peers. For the denition of a child with severe neurologic impairment, we
used a child who needs assistance with eating, bathing,
or using the toilet, is very slow at schoolwork, in
moderate to no physical pain, and is blind, deaf, or
unable to talk.

1277
Table IV

Decision analysis results

Variable
Cost-effectiveness data
Cost ($2005)
Effectiveness (QALYs)
Needed to treat (n)
To prevent 1 neonatal death
To prevent 1 affected child
To prevent 1 cesarean delivery
Outcomes for 160,000 pregnant
women with known history
of HSV*
Total cost (millions of $2005)
Total QALYS (in thousands)
Cesarean deliveries
Neonatal deaths
Severely impaired children

Acyclovir

No acyclovir

6,102
56.7117

6,122
56.7074

22,286
8,985
177

d
d
d

976
9,074
39,341
0.15
0.13

979
9,073
40,244
7.33
5.96

* Number of pregnant women with a diagnosed history of genital

HSV but without symptomatic recurrence during pregnancy each year
in the United States.

Analysis
First, the costs and QALYs for acyclovir prophylaxis
and no acyclovir treatment were calculated. The number
needed to treat was calculated for several outcomes:
cesarean deliveries, neonatal deaths and severely neurologically impaired children. Next, univariate sensitivity
on every variable in our model was performed. Over
particularly sensitive inputs, threshold analysis was
performed to determine values for which the input
would remain cost-eective. Finally, a Monte Carlo
simulation was used to test the robustness to simultaneous multivariable changes in the theoretic cohort of
160,000 women. For the Monte Carlo simulation,
triangular probability distributions were used. Triangular distributions were preferable to normal distributions
because they did not extend beyond either the zero or
one probability threshold.

Results
We found that acyclovir prophylaxis was both less
expensive (an average cost of $6102 vs $6122 per
woman) and more eective (an average composite
QALY for mother and child of 56.712 vs 56.707) for
women with a history of diagnosed genital HSV, but
without a recurrent infection during pregnancy (Table
IV). Thus, acyclovir prophylaxis dominates (less expensive, better outcomes) not giving acyclovir to this
population. When clinical outcomes were examined,
177 women must be treated to prevent 1 cesarean
delivery; 8985 women must be treated to prevent
1 aected child, and 22,286 women must be treated to
prevent 1 neonatal death. Table IV shows the outcomes
with or without acyclovir prophylaxis for 160,000
women. By providing acyclovir prophylaxis, we save

approximately $3.2 million, gain approximately 1000

QALYs, and prevent 6 severely neurologically impaired
children, 7 neonatal deaths, and approximately 1000
cesarean deliveries.
Univariate sensitivity analysis showed that our model
was quite robust. Changing all probabilities and utilities
down to one-third and up to 3 times their baseline
estimate had no discernible eect on outcome. Similarly,
there was no eect in varying either the cost of a
cesarean delivery (all the way down to a cost equal to
that of vaginal delivery) or the cost of having an aected
child. Changing the cost of the acyclovir, however, did
have an impact. Acyclovir was the dominant strategy up
to a cost estimate of $67 (150% of baseline). At a cost of
$280, the acyclovir prophylaxis strategy cost approximately $50,000 per QALY, and at a cost of $495, the
strategy cost $100,000 per QALY.
Monte Carlo simulation further conrmed the robustness of our model. Figure 2 displays the outcomes
of our 160,000 trials. As the graph demonstrates, most
trials placed acyclovir in the upper left quadrant (both
cheaper and more eective). In our 160,000 trials,
acyclovir was the dominant strategy in all but 26 trials
(0.02%). For these 26 trials, acyclovir was cost-eective
(at a cost of !$20,000 per QALY).

Comment
Our results demonstrate that acyclovir prophylaxis is
not only cost-eective but also cost saving for pregnant
women with a diagnosed history of genital herpes who
do not experience recurrence during pregnancy. This
conclusion is consistent with, although more expansive

1278

Figure 2 Monte Carlo simulation. Of 160,000 trials that

simultaneously vary probability estimates, prophylactic acyclovir therapy is both cheaper and more eective than no
acyclovir O99% of the time. From the origin of the dotted
line, a move to the left represents an increase in QALYs for
acyclovir, and a move up represents a cost savings. The curved
line represents the 95% CI.

than, previous cost-eective analyses, which have shown

acyclovir to be cost-eective for pregnant women who
have experienced symptomatic outbreaks in pregnancy.7
Our ndings are robust to both univariate sensitivity
analysis and Monte Carlo simulation.
Of note, not only did we nd the use of acyclovir
prophylaxis in this population to be cost-eective and
cost-saving, but also we found that the overall clinical
outcomes were improved. Although, the actual cost
savings of approximately $20 and the 0.01 improvement
in QALYs may seem small on an individual level, these
eects should be considered at the population level.
When considering the 160,000 women each year who
could benet from therapy, this becomes $3.2 million in
savings of direct medical costs and an increase of 1600
QALYs. As policy makers, we should consider the costeectiveness of therapeutic interventions. However, as
clinicians, we should seek to optimize the risks and
benets of our patients. Although the outcomes would
be improved in only 13 neonates, the benet to them and
their families is dramatic. Further, to the almost 1000
women who would not require a cesarean delivery, the
benets would surely outweigh the costs. Our ndings
are not dissimilar to the current group B streptococcus
screening protocol in terms of the ratio of numbers

Little and Caughey

that are needed to treat to prevent neonatal death or
severe injury.
Our study does have limitations. We used a model to
represent the clinical picture, which can be limited in its
scope and may miss a number of intangibles that can be
realized in a clinical study. Our model incorporates
probabilities from a variety of sources, some of which
may lack the power to demonstrate the full range of
eects that we are attempting to address. For example,
our estimate for the number of women with lesions or
asymptomatic shedding is from a study with just 201
women.6 However, this is the only published study that
is representative of our target population. In settings
such as this, the strength of decision analysis is to be
able to consider a wide condence interval around
estimates from small clinical studies. We used this point
estimate and varied it through a wide range in the
sensitivity analysis. Throughout this range, our results
remained robust, and the use of prophylactic acyclovir
both was cheaper and led to better outcomes. To
estimate the eects to a mother of having a severely
neurologically impaired child, we used preferences towards that of having a child with Down syndrome. Of
note, these estimates are likely to underestimate the
disutility of a child that has been severely neurologically
devastated by a neonatal HSV infection. Further, our
QALY estimates of how this outcome will aect society
have likely also been underestimated because they only
include the mother and the child and ignore the impact
on other members of the family.
Most neonatal herpes infections are not addressed in
this study; most of the neonates who are born with HSV
are delivered by women without a known history of
genital HSV.4 Only 17% to 40% of reproductive-aged
women with antibodies to HSV-2 have a history of
symptomatic infection. Also, approximately 2% of
women seroconvert during pregnancy.24 Women who
are seronegative for HSV have the highest probability of
transmitting HSV to their neonate (0.054% vs 0.026% for
HSV-1 seropositive women and 0.022% for HSV-2) as
primary infection is transmitted with far greater eciency
than recurrent infection.3 Screening for HSV status during pregnancy has been proposed as a way to identify
women or their partners who are either unaware of a past
infection or at risk of seroconverting during their third
trimester. Studies have started to address whether such
screening is cost-eective.14
Although we would have to treat almost 9000 women
to prevent one neonatal infection among women with a
known history of HSV, the risk for experiencing an
unnecessary cesarean delivery in the women with no
prophylaxis is estimated at 1 in R200. The societal costs
of acyclovir prophylaxis are quite low, and acyclovir has
no known harms in pregnancy.25 Overall, our ndings
suggest that obstetricians should prescribe acyclovir
with a low index of suspicion. We hope that these

Little and Caughey

robust ndings, which demonstrate the cost-eectiveness of acyclovir prophylaxis for a broad group of
women, spur continued research into further populations that may benet from therapy.

References
1. Gutierrez KM, Falkovitz Halpern MS, Maldonado Y, Arvin AM.
The epidemiology of neonatal herpes simplex virus infections in
California 1985 to 1995. J Infect Dis 1999;180:199-202.
2. Whitley R. Neonatal herpes simplex virus infection. Curr Opin
Infect Dis 2004;7:243-6.
3. Randolph AG, Washington E, Prober CG. Cesarean delivery for
women presenting with genital herpes lesions. JAMA 1993;270:
77-82.
4. Brown ZA, Wald A, Morrow RA, Selke S, Zeh J, Corey L. Eect
of serologic status and cesarean delivery on transmission rates
of herpes simplex virus from mother to infant. JAMA
2003;289:203-9.
5. Sheeld JS, Hollier LM, Hill JB, Stuart GS, Wendel GD.
Acyclovir prophylaxis to prevent herpes simplex virus recurrence
at delivery: a systematic review. Obstet Gynecol 2003;102:
1396-403.
6. Braig L, Luton D, Sibony O, Edlinger C, Boissinot C, Blot P, et al.
Acyclovir prophylaxis in late pregnancy prevents recurrent genital
herpes and viral shedding. Eur J Obstet Gynecol Reprod Biol
2001;96:55-8.
7. Randolph AG, Hartshorn RM, Washington AE. Acyclovir prophylaxis in late pregnancy to prevent neonatal herpes: a costeectiveness analysis. Obstet Gynecol 1996;88:603-10.
8. Scott LL, Alexander J. Cost-eectiveness of acyclovir suppression
to prevent recurrent genital herpes in pregnancy. Am J Perinat
1998;15:57-62.
9. Masoaka T, Shibata H, Amaki I, Takeo H, Sakurai K, Ise T, et al.
[The clinical eects of a new antiviral 9-(2-hydroxyethoxymethyl)
guanine (acyclovir) against herpes virus infections]. Gan To
Kagaku Ryoho 1983;10:944-52.
10. National Center for Health Statistics. National Vital Statistics
United States for 2003. Hyattsville, Md: Public Health Services;
2003.
11. Baker D, Brown Z, Hollier LM, Wendel GD Jr, Hulme L, Griths
DA, et al. Cost-eectiveness of herpes simplex virus type 2
serological testing and antiviral therapy in pregnancy. Am J Obstet
Gynecol 2004;191:2074-84.

1279
12. Lilford RJ, van Coeverden de Groot HA, Moore PJ, Bingham P.
The relative risks of caesarean section (intrapartum and elective)
and vaginal delivery: a detailed analysis to exclude the eects of
medical disorders and other acute pre-existing physiological disturbances. BJOG 1990;97:883-92.
13. Solomon L, Cannon MJ, Reyes M, Graber JM, Wetherall NT,
Reeves WC. Epidemiology of recurrent genital herpes simplex
virus types 1 and 2. Sex Transm Infect 2003;79:456-9.
14. Rouse DJ, Stringer JSA. An appraisal of screening for maternal
type-specic herpes simplex virus antibodies to prevent neonatal
herpes. Am J Obstet Gynecol 2000;183:400-6.
15. United States Department of Labor Bureau of Labor Statistics.
Consumer Price Indices. Washington DC: United States Bureau of
Labor Statistics; 2005.
16. Barnabas RV, Carabin H, Garnett GP. The potential role of
suppressive therapy for sex partners in the prevention of neonatal
herpes: a health economic analysis. Sex Trans Infect 2002;78:
425-9.
17. Waitzman NJ, Romano PS, Scheer RM. Estimates of the
economic costs of birth defects. Inquiry 1994;31:188-205.
18. Rouse DJ, Owen J, Goldenberg RL, Cliver SP. The eectiveness
and costs of elective cesarean delivery for fetal macrosomia
diagnosed by ultrasound. JAMA 1996;276:1480-6.
19. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-eectiveness
in health and medicine. New York: Oxford University Press; 1996.
20. Angeja A, Caughey AB, Vargas J, Gomez R, Washington AE.
Chilean womens preferences regarding mode of delivery: Which
do they prefer and why? Obstet Gynecol 2004;103:117S.
21. Kuppermann M, Nease RF, Learman LA, Gates E, Blumberg B,
Washington AE. Procedure-related miscarriages and Down syndrome aected births: Implications for prenatal testing based on
womens preferences. Obstet Gynecol 2000;96:511-6.
22. Strauss D, Shavelle R, Day S. Life Expectancy Project. Available
from: http://www.lifeexpectancy.com/cp.shtml. Accessed April 25,
2005.
23. Saigal S, Stoskopf BL, Feeny D, Furlong W, Burrows E,
Rosenbaum PL, et al. Dierences in preferences for neonatal
outcome among health care professionals, parents and adolescents. JAMA 1999;281:1991-7.
24. Brown ZA, Selke S, Zeh J, Kopelman J, Maslow A, Ashley RL,
et al. The acquisition of herpes simplex virus during pregnancy.
N Engl J Med 1997;337:509-15.
25. Stone K, Rei-Eldridge R, White AD, Cordero JF, Brown Z,
Alexander ER, et al. Pregnancy outcome following systematic
prenatal acyclovir exposure: conclusions from the international
acyclovir pregnancy registry, 1984-1999. Birth Defects Res Clin
Mol Teratol 2004;70:201-7.