# Projection of the Future Dimensions and Costs of the Genital Herpes Simplex Type 2 Epidemic in the United States

Background Infection with herpes simplex virus type 2 (HSV-2) currently affects approximately 22% of adult Americans and increased markedly in prevalence between the late 1970s and early 1990s. Although some estimates of the costs of prevalent disease due to HSV-2 are available, selection of interventions to prevent HSV-2 infection, as well as evaluation of their potential cost-effectiveness, should take into account projected future costs that will result if the epidemic is left unchecked.

Goal The goal was to estimate the future health and economic consequences attributable to the HSV-2 epidemic in the absence of interventions to slow the epidemic.

Study Design A mathematical model was constructed to project future increases in HSV-2 seroprevalence in the United States. The probability of heterosexual transmission of HSV-2 was estimated from cross-sectional estimates of infection prevalence reported by the National Health and Nutrition Examination Survey (NHANES). Per-infection expected costs were calculated on the basis of data obtained from the published medical literature.

Results Without intervention, the prevalence of HSV-2 infection among individuals aged 15 to 39 years was projected to increase to 39% among men and 49% among women by 2025. Annual incidence was projected to increase steadily between 2000 and 2025, from 9 to 26 infections per 1000 men and from 12 to 32 infections per 1000 women in this age group. The cost of incident infections in the year 2000 were estimated to be $1.8 billion; the cost of incident infections was predicted to rise to $2.5 billion by 2015 and $2.7 billion by 2025. The projected cumulative cost of incident HSV-2 infections occurring over the next 25 years was estimated to be $61 billion; at a 3% discount rate, this sum has a present value of $43 billion.

Conclusion The costs of incident HSV-2 infection in the United States are substantial and can be expected to increase as both the incidence and prevalence of this disease increase in the first half of the century. The level of resource allocation for HSV-2 prevention strategies should reflect the economic benefits that would result from control of this epidemic.

A mathematical model of herpes simplex virus type 2 infection in the United States suggested that herpes incidence and prevalence could continue to increase, with substantial economic consequences.

From the *City of Hamilton Social and Public Health Services Department and the †Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada; ‡Division of Infectious Diseases, University of Alabama School of Medicine, Birmingham, Alabama; and Departments of §Epidemiology and ∥Health Policy and Management, Harvard School of Public Health, Boston, Massachusetts

The authors thank Dr. Geraldine M. McQuillan, of the National Center for Health Statistics, for her provision of NHANES-II, NHANES-III, and NHANES-1999 seroprevalence data, and Dr. Karen M. Kuntz, for her review of the manuscript and many insightful comments. The authors acknowledge the support, guidance, and mentorship provided by the late Dr. Jonathan Freeman in the initiation of this project.

Dr. Fisman was supported by a National Research Service Award #5-T32-HS00020-15 from the Agency for Healthcare Research and Quality.

Reprint requests: Dr. David N. Fisman, Hamilton Department of Social and Public Health Services, 1 Hughson Street North, 4th Floor, Hamilton, Ontario, Canada L8R 3L5. E-mail: dfisman@city.hamilton.on.ca.

Received for publication October 2, 2001,

revised January 4, 2002, and accepted January 7, 2002.

HERPES SIMPLEX VIRUS TYPE 2 (HSV-2) is the most common cause of ulcerative genital disease in the United States. ^{1,2} The National Health and Nutrition Examination Survey III (NHANES-III) found that by the early 1990s, almost a quarter of American adults had serologic evidence of infection with HSV-2, a 30% increase in comparison with the late 1970s. ^{3,4} In some segments of the population, HSV-2 antibody seroprevalence is as high as 50%. ^{4}

Although HSV-2 infection appears to be extremely common, only 9% to 50% of HSV-2-infected individuals are aware of their infection. ^{4–7} This range is due in part to misdiagnosis of individuals with atypical manifestations of HSV-2 infection. ^{7} Nonetheless, symptomatic infection with HSV-2 affects a large proportion of the US population, with clinical manifestations include primary genital herpes syndrome, ^{8} recurrent genital ulceration, ^{9,10} neonatal herpes simplex virus infection, ^{11,12} and anxiety and depression. ^{13,14} Many symptomatic infections with HSV-2 are acquired from individuals with asymptomatic HSV-2 infection. ^{15,16}

A variety of public health interventions currently being formulated have the potential to help to slow the spread of HSV-2. These include serologic screening to identify asymptomatically infected individuals, ^{17,18} programs to increase condom usage, ^{19,20} suppressive antiviral therapy to prevent viral shedding, ^{21,22} and vaccines against HSV-2. ^{23,24} However, the prioritization of investment in such programs will be influenced not only by the cost of the interventions but also by the dimensions and costs of the HSV-2 epidemic that would occur in the absence of any intervention.

Mathematical models are useful tools for comparing the impact of health care interventions, either before or in lieu of clinical trials. In addition, mathematical modeling permits estimation of the susceptibility of different facets of disease epidemiology to intervention. Our objective was to project the future burden of HSV-2 infection in the United States, using a mathematical model that incorporated epidemiologic trends documented between 1976 and 1994. We also estimated the per-patient average costs of HSV-2 infection in men and women and combined these costs with projections of future disease prevalence to estimate the future costs of HSV-2 infection in the United States.

## Methods

### The Epidemic Model

We constructed a compartmental model to simulate the course of the HSV-2 epidemic in individuals aged 15 to 39 years in the United States. We restricted the model to this age group because the age-related increase in HSV-2 is attenuated in individuals aged 40 years and older, while seroprevalence of HSV-2 is low before age 15 years. ^{3,4} Health states in the model were broadly characterized as infected or susceptible (i.e., uninfected) and were further stratified by age to capture heterogeneity in risk of infection and annual live birth rates in different age groups. The model assumed random, heterosexual mixing of the population. ^{25} Infectiousness was assumed to be unrelated to the presence of symptomatic disease. ^{26}

Individuals enter and leave the model at a rate that is defined by aging of the population and by the rate of population growth, estimated to be approximately 0.7% per year. ^{27} This model is described in further detail in Appendix 1.

The rate at which HSV-2 is transmitted between men and women is a function of the infectiousness of sexual contact with an infected individual, the rate of partner change among sexually active individuals, and the prevalence of infection. ^{25} While estimates of infectiousness of HSV-2 exist, ^{5,28–30} they are derived from monogamous heterosexual couples and may not be applicable to the general heterosexual population. Therefore, we estimated age- and gender-specific rates of acquisition of HSV-2, using the changes observed in population prevalence of HSV-2 infection between NHANES-II and NHANES-III. These studies collected demographic data and serum from large, representative cohorts of Americans between 1976 and 1980 and between 1988 and 1994, respectively. ^{3,4}

We calculated the rates of infection that would have been necessary to change the prevalence in a given age cohort in NHANES-II (e.g., 20–29-year-olds) to the prevalence observed in the next oldest age cohort (e.g., 30–39-year-olds) at the time of NHANES-III (Appendix 2). The resulting estimates of initial age- and gender-specific rates of HSV-2 acquisition are presented in Table 1. ^{3–5,7,9–12,27–32,35–37,47,48,54,55,71,89–94} Seroprevalence data from these studies were assumed to apply to the midpoint of the study. Age-specific seroprevalence estimates for 1978 were used as initial model inputs in the age-stratified model (Table 1). The model was constructed with use of Microsoft Excel version 6.0 (Microsoft Corporation™, Redmond, WA).

### Natural History of HSV-2 Infection

In the model, we stratified the infected health state such that individuals with symptoms were distinguished from those who were asymptomatic. Individuals who were symptomatic could initially experience a primary HSV-2 syndrome ^{2,8} or could progress directly to a health state characterized by relapsing, remitting ulcerative genital disease. ^{9}

Estimates of the probability of symptomatic genital herpes in an individual infected with HSV-2 vary widely. ^{4,5,7,31} In NHANES-III, only 9% of individuals with serologic evidence of HSV-2 infection reported a prior diagnosis of genital herpes. ^{4} However, clinical evidence of genital herpes can be demonstrated in approximately 50% of asymptomatic individuals with serologic evidence of infection followed prospectively by experienced practitioners. ^{7} We estimated the probability of symptomatic disease by comparing the annual incidence of HSV-2 infection projected by our model with estimates of incident, symptomatic genital herpes from the Centers for Disease Control and Prevention (CDC, Atlanta). ^{31} Such a comparison suggests that approximately 17% of infected individuals receive medical care for recognized genital herpes.

The frequency with which individuals relapse varies widely. ^{10} We estimated a weighted average lifetime number of relapses by applying a linear decline in annual relapse rate to published gender-specific distributions of symptom frequency in the first year of infection ^{32} (Appendix 3).

### Pregnancy and Maternal–Fetal Transmission

An important contributor to the health and economic consequences of HSV-2 relates to infection and pregnancy and may result either from maternal–fetal transmission of HSV-2, with resultant infection in the neonate, ^{33–37} or from the increased risk of cesarean section observed for women with a history of symptomatic HSV-2 infection. ^{38} We estimated the probability of pregnancy in women by using average age-specific live birth rates for the US population from the period between 1980 and 1997. ^{27}

Modeling maternal–fetal transmission of HSV-2 infection is complicated by several factors, including possible changes in sexual behavior and acquisition of HSV-2 infection associated with pregnancy, ^{39–41} the difficulty in estimating the incidence of spontaneous abortion due to early trimester infections, ^{42–44} and by protective effects associated with prior HSV-1 infection. ^{11,12,45,46} The best available data suggest that neonatal infection with either HSV-1 or HSV-2 usually occurs following maternal acquisition of infection in the third trimester of pregnancy. ^{11,12,47,48} We assumed in the base case analysis that maternal–fetal transmission of HSV-2 occurs exclusively in this context. We conservatively estimated the risk of third-trimester acquisition of HSV-2 infection by calculating the decrease in risk of third-trimester maternal infection that would be necessary to explain the observed incidence of neonatal HSV infection. These calculations are described in detail in Appendix 4.

Current published guidelines ^{48–50} advocate cesarean delivery in women with known genital herpes if herpetic lesions are present at the time of labor. However, practicing obstetricians may perform cesarean delivery for women with a history of genital herpes, even in the absence of visible lesions, ^{51,52} and there appears to be an absolute increase of approximately 20% in the probability of cesarean section among American women with known genital herpes. ^{38} We estimated the future increase in cesarean sections among women with symptomatic HSV-2 infection as the product of the expected number of future term pregnancies and the absolute increase in risk of cesarean section associated with a history of genital herpes. These calculations are described in detail in Appendix 5.

### Costs

Selected cost inputs for the model are presented in Table 2. ^{2,8,35–38,49,53–57,71,95} Individuals infected with symptomatic HSV-2 accrued medical costs due to hospitalization (if they experienced a severe primary HSV-2 syndrome), physician visits and drugs prescribed for symptomatic episodes and long-term suppressive therapy, and the excess occurrence of cesarean sections among women with a history of genital herpes. ^{38,53} It was assumed that patients would be treated in accordance with the *1998 CDC Guidelines for Treatment of Sexually Transmitted Diseases*. ^{48}

Direct medical costs resulting from neonatal herpes simplex virus infection were estimated as a weighted average of costs associated with neonatal herpes without sequelae, neonatal herpes causing death, and neonatal herpes resulting in long-term disability. ^{35–37,54,55} Medical costs for the first year were estimated as a weighted average of the costs of caring for infants with skin, eye, and mucous membrane disease (SEM), herpetic encephalitis, and disseminated neonatal HSV infection. ^{38} Infants who survived with moderate or severe neurologic sequelae were projected to incur additional lifetime medical and long-term care costs, based on published estimates of the future costs of caring for children with neurologic impairment. ^{37,56} Our model also included patient time costs associated with travel, waiting room time, and time spent receiving medical care.

The total future cost of HSV-2 infection for an individual was discounted to present value with use of a 3% real discount rate. ^{58} Thus, we estimated the total cost of infection for that individual as though the entire future stream of medical and nonmedical costs resulting from HSV-2 infection had to be paid for at the time of the initial infection. The annual costs attributable to incident HSV-2 infection in the United States were estimated as the product of the projected number of incident infections and the average present value of the costs attributable to a single HSV-2 infection. Both undiscounted estimates and estimates discounted to present value were calculated. All costs were converted to year-2000 US dollars with use of the US Consumer Price Index for medical goods and services. ^{57}

### Sensitivity Analysis

We performed sensitivity analysis by varying parameter values over plausible ranges, in order to assess the robustness of our projections in the face of reasonable variation in data inputs. We also performed best-case and worst-case analyses by running the model using upper and lower bounds of parameters to which our results proved sensitive.

In our best-case estimates, we assumed that the true prevalence of HSV-2 infection in the United States. was represented by upper 95% confidence limits reported from NHANES-II and by lower 95% confidence limits reported from NHANES-III. This would result in an attenuated estimate of the force of infection acting on the population during that time period, as the true increase in HSV-2 seroprevalence between 1978 and 1991 would have been less. Conversely, our worst-case estimates assumed that the true prevalence of HSV-2 infection in the United States. was represented by lower 95% confidence limits from NHANES-II and upper 95% confidence limits from NHANES-III, which would result in estimation of a larger force of infection than in the base case simulation.

Upper-bound estimates of HSV-2-related complications of pregnancy were obtained by increasing the probability of symptomatic disease in women (which would increase the risk of cesarean section) and by introducing a 1/2000 risk of maternal–fetal transmission of HSV-2 for pregnant women with chronic HSV-2 infection. ^{48} The estimates of neonatal HSV-2 infection used in the base case represented lower-bound estimates and thus were also used in the best-case analysis.

## Results

### Projected Incidence and Prevalence of HSV-2 Infection

Under base case assumptions, our model predicted that the prevalence of HSV-2 infection in 1991 would be 18% in men and 25% in women, compared to the reported prevalence of 16% and 23% in NHANES-III. We calibrated our model by applying an empirically derived correction factor of 0.9 to rates of partner change for men and women in all age groups, which resulted in a projected prevalence of 16% among men and 22% among women. All subsequent projections were made with use of the calibrated model.

The estimated incidence of HSV-2 infection among those aged 15 to 39 years in the year 2000 was 25 infections per 1000 women and 19 per 1000 men. Incidence was projected to increase to 32 cases per 1000 women and 26 cases per 1000 men by 2025 and to remain fairly constant between 2025 and 2050 (Figure 1).

The prevalence of HSV-2 infection in this age group during the year 2000 was estimated to be 30%, while the prevalence among men was estimated to be 23%. Prevalence was projected to increase sharply between 2000 and 2025, rising to 52% among women and 41% among men, and to increase more gradually between 2025 and 2050 (Figure 2).

With an assumed constant rate of population growth of 0.7% per year, the number of infected individuals in the age group of 15 to 39 years was projected to increase from 28 million in 2000 to 43 million by 2015 and to 51 million by the year 2025.

### Per-Infection Costs of Symptomatic Genital Herpes

Assuming that 17% of infected individuals developed symptomatic genital herpes, we estimated that infected men would experience an average of 19 lifetime symptom-days, while infected women would experience an average of 16 lifetime symptom-days, following infection with HSV-2. The present value (at the time of infection) of the cost of genital herpes–related symptomatology was estimated to be $620 for men and $510 for women.

### Incidence and Costs of Neonatal HSV-2 Infection

We calibrated the probability of third-trimester infection to produce a baseline incidence of neonatal HSV-2 infection of 9.3 cases per 100,000 live births. ^{59} If age-specific live birth rates were to remain constant, we would project an incidence of neonatal HSV-2 infection of 29 cases per 100,000 live births by 2025, with the incidence remaining nearly constant between 2025 and 2050 (Figure 3). The average cost of future HSV-2-related complications of pregnancy per infected female (including maternal–fetal transmission of HSV-2 infection and excess cesarean sections among women with known HSV-2 infection) varied according to age at infection and was $128 for women aged 15 to 19 years, $160 for women aged 20 to 29 years, and $65 for women aged 30 to 39 years.

### Projected Costs of HSV-2 Infection in the United States

The direct medical costs associated with incident HSV-2 infections in the year 2000 were estimated to be $1.6 billion. When patients’ time, travel, and lost wages were included, the cost of incident infections increased to $1.8 billion. The cost of incident infections was projected to rise to $2.5 billion by 2015 and to $2.7 billion by 2025. The cumulative cost of incident HSV-2 infections occurring between 2000 and 2025 was estimated to be $61 billion. With a 3% discount rate, this sum has a present value of $43 billion (Table 3).

### Sensitivity Analyses

With use of upper-bound values for transmission rates, it was projected that nearly 70% of individuals aged 15 to 39 years would be HSV-2-infected by the year 2025, at which time prevalence would be expected to plateau. With use of lower-bound estimates, it was projected that 30% of individuals aged 15 to 39 years would be HSV-2-infected by the year 2015 and 34% would be infected by 2025 (Figure 4).

Our projections of cost proved most sensitive to the probability of symptomatic disease among infected individuals and to the future prevalence of HSV-2 infection in the population. Varying the probability of symptomatic infection from 9% to 50% resulted in lower- and upper-bound estimates of average symptom-associated costs of $300 and $1680 for infected men and $230 and $1260 for infected women.

When we introduced a 1/2000 chance of maternal–fetal transmission of infection in women with chronic HSV-2 infection, the annual incidence of neonatal HSV-2 infection increased from 18 infections per 100,000 live births in 1981 to 55 infections per 100,000 live births by 2030 (Figure 3).

When we combined worst-case estimates for HSV-2 prevalence with upper-bound cost estimates, it was estimated that the current cost of incident infections is over $9 billion per year (Table 3). The total projected cost of incident HSV-2 infections between 2000 and 2025 would be $269 billion, with a discounted present value of $177 billion. The use of best-case prevalence and lower-bound cost estimates resulted in an estimated cost of incident HSV-2 infections of $700 million to $1 billion per year between 2000 and 2025, with cumulative costs of $23 billion; this sum has a discounted present value of $16 billion.

## Discussion

We constructed a mathematical model to project the future dimensions of the HSV-2 epidemic in the United States. Any prophesy of the future epidemiology of an infectious disease needs to be interpreted with caution. ^{60} However, a model that simulates rates of transmission of HSV-2 necessary to produce the change in seroprevalence seen in the United States between the late 1970s and early 1990s would project a continued increase in the incidence and prevalence of HSV-2 infection in the early 21st century. Our base case simulation projected a modest increase in incidence between 2000 and 2025; however, because of the chronic nature of HSV-2 infection, this was projected to produce an almost twofold increase in the prevalence of HSV-2 infection during the same time period. We estimate that the cumulative economic consequences of such an increase would be substantial.

In our base case simulation, the prevalence of HSV-2 infection reached a plateau during the second half of the 21st century. Our projections of equilibrium prevalence in the population aged 15 to 39 years as a whole are similar to current estimates of HSV-2 prevalence among blacks, which did not change substantially between 1978 and 1991. ^{3,4} Our model suggests that one possible explanation for this difference would be that the HSV-2 epidemic is more mature and nearer to equilibrium in the African American population.

Our initial estimates of HSV-2 transmission rates, using data from NHANES-II and NHANES-III, reproduced the change in prevalence that was reported between the late 1970s and early 1990s. Our estimates of the number of incident cases of HSV-2 infection during that time are similar to those generated by Armstrong and colleagues, using the same data but a different statistical model. ^{60} For example, Armstrong and colleagues estimated approximately 1.34 million new infections with HSV-2 among individuals less than 39 years of age in 1985, as compared with 1.48 million new infections projected by our model.

Our projection of HSV-2 prevalence in the late 1990s is higher than that reported during the first year of the NHANES-IV study, in 1999. ^{62} This may be related to the fact that sexual behavior remained constant in our model, whereas sexual behaviors that would affect HSV-2 transmission (particularly numbers of sex partners and rates of condom use) may have changed in the 1990s, perhaps as a result of the AIDS epidemic. ^{63–65} Increased use of antiviral drugs or cyclic variation in sexual risk behavior, as has been seen with the US AIDS epidemic ^{66–68} and previously with the syphilis epidemic, ^{69,70} might explain the discrepancy between our model outputs and these data. Alternatively, this discrepancy may reflect the preliminary nature of the NHANES-IV data or the regional nature of annual sampling in the NHANES studies. Future updates on NHANES-IV will permit ongoing calibration of our model.

Maternal-fetal transmission of HSV-2 infection is thought to occur with high frequency when maternal infection occurs in the third trimester of pregnancy and with much lower frequency with chronic maternal infection. ^{11,12,33,48,59} We initially restricted maternal–fetal transmission to women with third-trimester infection; even so, our base case analysis projected rates of neonatal HSV-2 infection higher than those reported in the population. ^{46,59} When we added the possibility of transmission by chronically infected women, the difference between our projections and empirical estimates grew even larger. Other authors have also noted that mathematical estimates of neonatal HSV-2 are larger than empirically reported rates. ^{34,46,61} An important attribute of models such as this one is their ability to highlight discrepancies such as this one. This suggests that the epidemiology of neonatal herpes virus infection is incompletely understood and highlights the importance of this area for future research.

We purposefully chose to focus on the costs of incident disease, rather than combine costs of incident and prevalent HSV-2 infection, and we valued the total expected future stream of costs generated by an incident infection as a cost incurred at the time of infection. Given that goals for HSV-2 control focus on prevention of incident infections, emphasizing the cost of incident disease highlights modifiable costs of the HSV-2 epidemic and makes our model potentially useful for assessing the cost-effectiveness of herpes control interventions. Because the costs associated with genital herpes are “front-loaded,” (i.e., severe medical sequelae, including neonatal infections, are most likely to occur with initial infection) ^{11,12,71} and the frequency of relapse decreases with time, ^{32} the impact of this assumption is blunted.

Our annual direct medical cost estimates are higher than some existing estimates, which have ranged from $100 million (early 1980s) to $207 million (between 1992 and 1994). ^{38,72,73} Szucs and colleagues recently estimated total costs of genital herpes 1996 (including patients’ time and lost wages) at $1.2 billion. ^{74} However, although we may have overestimated the amount of medical care required by HSV-2-infected individuals, we have certainly underestimated some costs associated with HSV-2 infection, such as costs associated with enhanced HIV transmission among individuals with genital herpes ^{75–77} and economic costs associated with the physical pain, depression, disruption of interpersonal relationships, ^{13,78,79} and even legal action ^{80,81} that may accompany HSV-2 infection.

In addition, other published estimates, particularly those derived from large databases, may be underestimates. For example, Tao and colleagues ^{38} estimated that only 13% to 28% of acyclovir claims not associated with a specific diagnostic code were for the treatment of genital herpes. As drugs accounted for over half the costs attributed to genital herpes, underestimation of drug costs would substantially decrease the estimated annual cost of genital herpes. When we removed drug costs associated with chronic suppressive antiviral therapy from our model, in conjunction with our lower-bound estimate for the probability of symptomatic disease, our direct medical cost estimate for 1993 decreased to $350 million, closer to the estimate of $207 million generated by Tao and colleagues.

Our model has additional important limitations. We did not stratify our population according to ethnicity due to the difficulty in making assumptions about interethnic sexual mixing. ^{82,83} However, the higher prevalence of HSV-2 infection among blacks relative to the population as a whole ^{3,4} suggests that the burden of morbidity and economic cost associated with incident HSV-2 infection falls disproportionately on this population.

We used a mathematical model to project the future costs and consequences of the HSV-2 epidemic in the United States. Given the future economic burden that is likely to be imposed on society by the HSV-2 epidemic, substantial investment in HSV-2 prevention strategies makes economic sense. The recent development of inexpensive serologic assays for the detection of asymptomatic HSV-2 infection ^{84–86} and promising new vaccines ^{23,24,87,88} might provide superior means of control of the HSV-2 epidemic in the United States. Programs incorporating these new technologies are anticipated to be expensive, but in light of the large costs that the HSV-2 epidemic is projected to impose, such programs may prove to be cost-effective or cost-saving.

### Appendix 1: Compartmental Model for Genital HSV-2 Infection

We constructed a compartmental model to simulate the course of the HSV-2 epidemic in individuals aged 15 to 39 years in the United States. Health states in the model were broadly characterized as infected (*I*) or susceptible to infection (*S*). Assuming random, heterosexual mixing of the population, susceptible individuals would become infected in proportion to the prevalence of infection among members of the opposite sex at some time EQUATION

the annual probability of sexual transmission of HSV-2 within a sexual partnership (β), and the age- and gender-specific weighted average number of partners per person in a year (*c*). ^{25}

Our model incorporated three age strata: stratum 1 included those aged 15 to 19 years, stratum 2 included those aged 20 to 29 years, and stratum 3 included those aged 30 to 39 years. We developed a series of difference equations to calculate the annual number of susceptible and infected individuals in each age stratum.

#### Susceptible Individuals

The first, youngest age stratum comprises individuals who have just become sexually active. Because no individual is infected with HSV-2 before initiating sexual activity, the number of susceptible individuals in this stratum at some time *t + 1* will be equal to the number of individuals remaining in this stratum at time *t* (i.e., the number not lost due to age or acquisition of infection), plus those individuals who enter the stratum by reaching an age at which sexual activity is initiated. The number of individuals entering the youngest stratum by initiating sexual activity will be proportional to the size of the youngest age stratum (both infected and susceptible individuals). Individuals move out of the youngest susceptible age stratum either by growing older or by acquiring HSV-2 infection.

The number of individuals in the youngest age stratum at some time *t + 1* will be the number of susceptible individuals:EQUATION

where the constants μ*1* and μ*2* represent the rate at which individuals move in and out of the youngest age stratum, respectively, the constant β*c* *1* is the product of the infectivity of HSV-2 and the rate of partner change in the youngest age stratum, and the term EQUATION

represents the prevalence of HSV-2 infection among members of the opposite sex at some time *t.* The rate at which individuals move out of the youngest age stratum, μ*2*, is defined as *1/D*, where *D* is the duration of that age stratum. For a given rate of population growth, *g*, the term μ*1* will be equivalent to *(1 + g)/D.*

For all subsequent age strata, the rate at which susceptible individuals enter the stratum is a function of the size and duration of the next youngest age stratum, *S* _{i − 1}. For the *i* *th* age stratum, the number of susceptible individuals in the stratum at some time *t + 1* will be:EQUATION

Summing across all *n* age strata, the number of susceptible individuals in the population at some time *t + 1* will be:EQUATION

#### Infected Individuals

The number of infected individuals in the first, youngest age stratum at time *t + 1* will be equal to the number of individuals infected at time *t*, plus those who acquired infection during the interval between *t* and *t + 1*, minus those individuals who leave the youngest age stratum by growing older:EQUATION

In subsequent age strata, the number of infected individuals in the *i* *th* stratum at some time *t + 1* will be equal to the number of infected individuals in that stratum at time *t*, plus those individuals infected in younger age strata who enter the stratum by growing older and those individuals with newly acquired infections, minus those individuals who leave the stratum by growing older:EQUATION

Summing across all *n* age strata, the number of infected individuals in the population at time *t + 1* will be EQUATION

### Appendix 2: Estimation of Rates of Partner Change

The rate at which HSV-2 is transmitted between men and women is a function of the infectiousness of sexual contact with an infected individual (β), the rate of partner change (*c*) among sexually active individuals, and the average prevalence of infection among members of the opposite sex:EQUATION

Thus, the rate of acquisition of HSV-2 infection within the *i* *th* age stratum may be represented as EQUATION

where *λi* is the “force of infection” in the *i* *th* age stratum, or the rate at which infection is acquired by susceptible individuals. ^{96} This *λi* can also be represented as EQUATION

where *p* *t1* is the prevalence of infection in the *i* *th* age group at some time *t* *1*, while *p* *t0* is the prevalence in the *i* *th* age group at some earlier time *t* *0*, and Δ*t* is equal to the time elapsed between *t* *1* and *t* *0*. For a given age cohort, we estimated *p* *t1* as prevalence in that age cohort at the time of NHANES-III, while *p* *t0* was estimated as the prevalence in the next youngest age cohort at the time of NHANES-II. We estimated Δ*t* as 13 years, the time elapsed between the midpoint of NHANES-II (1978) and NHANES-III (1991).

Given equations (2.1) and (2.2), and assuming some average seroprevalence in the opposite sex between *t* *0* and *t* *1*, the product of the infectiousness of an individual with HSV-2 and the rate of partner change in a given age cohort, (β*c*) *i*, can be estimated as EQUATION

We attempted to derive rates of partner change using other sexual mixing assumptions, including complete age-assortative mixing (i.e., individuals choose partners from within only their own age group), ^{25} and using several less extreme forms of age-assortative mixing (e.g., men mix with women from their own and younger age groups at one rate and with older women at another rate, while women behave in the opposite manner). The use of age-assortative mixing resulted in extremely high estimates of HSV-2 seroprevalence in the early 1990s. The use of a matrix algebraic approach ^{96} demonstrated that other mixing patterns were mathematically implausible.

### Appendix 3: Projection of Future Number of Relapses and Relapse-Associated Costs

We calculated weighted average relapse rates in the year after infection from published distributions of symptom frequency. Suppose *n* different frequencies of symptomatic relapse are observed in a population in the first year after development of symptomatic genital herpes. The weighted average rate of relapse in the first year may be represented as EQUATION

where *L* *q* is the rate of relapse in the *q* *th* stratum, while *P* *q* is the proportion of the population in the *q* *th* stratum. The frequency of relapse in individuals with symptomatic genital HSV-2 infection appears to decrease in a linear fashion over time. ^{10} If the annual change in the number of relapses for a given individual is Δ*L*, that individual will cease to experience relapses at time *t* *end* such that EQUATION EQUATION

The total number of future relapses can be represented as a sum of the sum of relapses over time, or EQUATION

The average per-relapse cost at some time *t*, represented as *C* *relapse* (*t*) will change over time, as individuals who have high initial rates of relapse are likely to be given chronic suppressive antiviral therapy, while those with lower relapse rates will be treated with intermittent antiviral therapy. In a population with *n* different relapse rates, the average present value of total relapse-associated costs among individuals with symptomatic disease can be calculated as EQUATION

where EQUATION

represents the discount factor, with a 3% annual discount rate and discounting at the beginning of each year.

### Appendix 4: Incidence and Costs of Maternal–Fetal Transmission Among Women With Third-Trimester HSV-2 Infection

Transmission of HSV-2 infection from a pregnant woman to a neonate occurs most commonly in the context of new maternal infection in the third trimester of pregnancy, with delivery occurring before the development of humoral immunity to HSV-2 in the mother. ^{12} Transmission of HSV-2 infection to the neonates of women with chronic HSV-2 infection appears to be much less efficient. ^{11} In our base case analysis, we assumed that maternal–fetal transmission of HSV-2 occurred exclusively in the context of third-trimester infection.

In a woman with incident HSV-2 infection, the probability of transmission of HSV-2 to a neonate will be a function of the probability that the woman is in the third trimester at the time of initial infection. The expected fraction of a year during which a woman will be in the third trimester of pregnancy will be approximately EQUATION

where EQUATION

is the live birth rate in a woman in the *i* *th* age stratum. The “force of infection” among initially susceptible women in the third trimester, λ*third*, will be related to the prevalence of infection in the opposite sex, as in equation (2.1) above. However, it might be expected that the rate of partner change and the type of sexual activities engaged in during the third trimester of pregnancy would be different from those in the nonpregnant population, such that EQUATION

where λ*third* is the product of the rate of transmission within a partnership and the rate of partner change among women in the third trimester of pregnancy. However, λ*third* will differ from β*c* in the population as a whole by some constant ** k**, such that EQUATION

Thus, the “force of infection” in the third trimester would be EQUATION

and the probability of third-trimester infection among initially susceptible women in the *i* *th* age stratum would be EQUATION

The probability of maternal–fetal transmission among initially susceptible women in a given age stratum is EQUATION

where EQUATION

is the probability of perinatal transmission of HSV-2 infection from a woman to a neonate, given third-trimester infection.

From equation (4.2) we can see that in a population with *n* age strata, the annual number of infants with HSV-2 infection is estimated to be EQUATION

The costs associated with transmission of HSV-2 to a neonate as a result of third-trimester infection will be EQUATION

where *C**N* is the present value of the total costs associated with neonatal HSV-2 infection.

The constant *k* was estimated by calibrating the model such that the model output matched empirical data on the incidence of neonatal herpes infection. ^{59} Between 1978 and 1981, the incidence of neonatal herpes infection in King County, Washington, was reported to be 11.9 cases per 100,000 live births. Approximately 78% of cases of neonatal herpes infection were due to HSV-2 infection; thus, the incidence of neonatal HSV-2 infection was estimated to be 9.3 per 100,000 live births.

On the basis of the estimates of incident HSV-2 infection provided by our model and using age-specific live birth rates, we estimated the number of third-trimester HSV-2 infections occurring in 1981 among women aged 15 to 39 years to be 8670 *k*. If the probability of maternal–fetal transmission following third-trimester infection is assumed to be approximately 10%, the number of infants with congenital HSV-2 infection as a result of third-trimester infection would be 867 *k*. We estimated a total of 3,601,270 live births in the United States in 1981, so that the incidence of neonatal HSV-2 infection due to third-trimester infection in the mother would be 22.5 *k* cases per 100,000 live births. Assuming that King County, Washington, was representative of the United States as a whole:EQUATION

so that *k* has a value of 0.41. We assumed that changes in sexual behavior associated with the third trimester of pregnancy remained constant over time and used *k* to estimate the projected future number of cases of neonatal HSV-2 infection resulting from third-trimester maternal infection.

### Appendix 5: The Incidence and Costs of Obstetrical Complications of Chronic HSV-2 Infection

In our base-case analysis, we assumed that all maternal-fetal transmission of HSV-2 would result from third-trimester maternal infection. However, there is likely a small risk of maternal–fetal transmission of HSV-2 even for women with chronic HSV-2 infection. In addition, women with a history of chronic HSV-2 infection appear to be more likely to undergo cesarean section, in part because of existing recommendations that cesarean section be performed if visible genital lesions are present in the mother at the time of delivery, although practicing obstetricians may be more likely to perform cesarean sections on women with a history of symptomatic HSV-2 even in the absence of visible lesions at the time of labor. ^{51,52}

The cumulative future risks of cesarean section and of maternal–fetal transmission of HSV-2 as a result of chronic maternal infection will be a function of the age at which initial HSV-2 infection is acquired, as individuals who are infected at a younger age will have, on average, more future term pregnancies and thus more future chances to incur obstetrical complications of chronic HSV-2 infection.

If a woman in the *i* *th* age stratum becomes infected, on average, in the middle of that age stratum, she will have *z* future years of potential child-bearing, such that:EQUATION

where ** D** is the duration of a given age stratum and

*n*is the total number of age strata during which childbearing is possible. We made the simplifying assumption that a woman would not be at risk for obstetrical complications of chronic HSV-2 infection until the year after acute infection. For a woman infected at some time

*t = 0*, the average number of future pregnancies that might be complicated by chronic HSV-2 infection would be EQUATION

For an individual with symptomatic HSV-2 infection and *z* future years of childbearing, the average total number of excess cesarean sections resulting from symptomatic infection with HSV-2 was estimated to be EQUATION

where Δ*P* *(cesarean)* is the excess probability of cesarean section associated with a history of symptomatic HSV-2 infection (estimated to be approximately 20%). ^{38} The present value (at the time of incident infection in the mother) of the future stream of costs due to excess cesarean sections and neonatal HSV-2 infection secondary to chronic maternal infection would be approximately EQUATION

where *C**cesarean* represents the excess costs associated with cesarean (as opposed to vaginal) delivery, and EQUATION

represents the discount factor, with a 3% annual discount rate and discounting at the beginning of each year.

Similarly, if EQUATION

is the probability of maternal–fetal transmission of HSV-2 by a chronically infected mother, the average future incidence of neonatal infection resulting from chronic maternal infection with HSV-2 would be approximately EQUATION

while the average present value of future costs due to maternal–fetal transmission of HSV-2 would be EQUATION

where *C**neonatal* represents the present value of the costs of neonatal infection at the time of neonatal infection.

We evaluated the health and economic impacts of maternal-fetal transmission of HSV-2 resulting from chronic maternal infection by assigning EQUATION

values of 1/2000 and 1/10,000.