OBJECTIVES. Because of the herd-immunity phenomenon, the benefits of immunization against hepatitis A extend beyond those received by those who are vaccinated. This analysis estimates the impact of herd immunity on the cost-effectiveness of routine hepatitis A immunization among US children.
PATITENS AND METHODS. In an economic model, the costs and benefits of hepatitis A immunization were estimated for immunizing all US children at age 1 year over a 10-year period starting in 2005. The future burden of disease from hepatitis A was also estimated with this model, and the fraction that would be prevented by herd immunity was modeled by using a previously published analysis of the relationship between hepatitis A vaccination coverage and declines in hepatitis A incidence.
RESULTS. Without accounting for herd-immunity effects, the costs of routine immunization would average $32000 per quality-adjusted life-year gained for the first 10 cohorts immunized starting with the 2005 birth cohort. Herd-immunity effects would be expected to produce substantial additional benefits, lowering the cost of the immunization program to $1000 per quality-adjusted life-year gained for the first 10 cohorts. Herd-immunity benefits would be greatest for the first few cohorts, more than doubling the benefits of immunization, and would decline over time. In a univariate sensitivity analysis, estimates were most sensitive to vaccination costs but remained below $20000 per quality-adjusted life-year under all of the assumptions.
CONCLUSIONS. Herd-immunity effects more than double the savings from hepatitis A immunization during the first 10 years of the program. After accounting for these effects, immunization is close to cost-neutral on a cost-per-quality-adjusted-life-year basis.
Hepatitis A has been one of the most common communicable diseases in the United States.1 The illness generally lasts 2 weeks to a few months, is characterized by jaundice, dark urine, nausea, anorexia, and abdominal discomfort and is followed by a prolonged convalescent period. Most cases resolve completely, but a small proportion progress to fulminant liver failure, which can be life threatening and require liver transplantation. Thus, hepatitis A is characterized by high rates of morbidity but relatively modest rates of mortality. Among children, infection with hepatitis A virus (HAV) is often asymptomatic and frequently unrecognized.
Before the availability of vaccine, ∼25000 to 35000 cases of acute hepatitis A were reported to public health agencies each year in the United States.2 Because many cases of HAV infection are unrecognized and many cases of acute hepatitis A are not reported, the true number of HAV infections is estimated to be 11-fold higher.3 The highest incidence of infection is in young children.3
Since 1995, 2 highly effective vaccines have been licensed in the United States for the prevention of hepatitis A. In 1996, the Advisory Committee on Immunization Practices (ACIP) issued recommendations for the use of these vaccines in certain high-risk populations.4 In 1999, the committee expanded these recommendations to include routine childhood immunization in 11, predominantly Western, states with historic rates that were at least twice the national average.1 In 6 other states, where rates had been between 1 and 2 times the national average, the ACIP recommended that routine immunization be considered. For the remaining 33 states, the ACIP made no recommendation for statewide immunization.
Since 1999, hepatitis A incidence rates have declined to historic lows, despite incomplete implementation of the committee's recommendations and only modest levels of vaccine coverage.2,5 In a Poisson regression analysis of nationwide data, hepatitis A declined by much higher proportions than would be expected were the vaccine protecting only the individual vaccine recipients.6 This suggests that children and adults who did not receive the vaccine were being protected by herd-immunity effects.
Herd immunity is a phenomenon in which the incidence of disease drops among the nonvaccinated population as an indirect effect of vaccination.7,8 This effect occurs because vaccinees, protected by the direct effects of vaccination, are less likely to be sources of infection for others. Herd-immunity effects from routine childhood vaccination are believed to be particularly strong when incidence is highest among young children, as with HAV infection.8
By increasing the impact of the intervention, herd-immunity effects may make communicable disease prevention programs more cost-effective than they would be otherwise.7 For example, herd immunity is believed to increase the cost-effectiveness of vaccination against pneumococcal9 and meningococcal disease.10 In this analysis, we estimate the impact of herd immunity on the cost-effectiveness of hepatitis A immunization in the United States.
PATIENTS AND METHODS
Definition of Herd-Immunity Effects
Immunization against hepatitis A was assumed to result in a decrease in the incidence of acute hepatitis A both because of the direct effects of immunization (the direct protection of individual vaccinees against infection) and the indirect effects of immunization (protection of unvaccinated persons) also know as herd-immunity effects. Herd-immunity effects were further divided into within-cohort herd-immunity effects (those occurring within the cohort being analyzed) and out-of-cohort herd-immunity effects.
Definition of the 2 Models
All of the analyses were done by birth cohort, defined as all (∼4 million) children born in the United States in a particular year. Two analyses are presented, the direct-effects model, which includes only direct effects, and the full model, which also includes all herd-immunity effects. Because herd-immunity benefits are greatest for the first immunized cohort and decline over time as more and more cohorts are immunized, we arbitrarily chose to estimate the impact of immunization for the first 10 birth cohorts immunized (ie, starting with the 2005 birth cohort and ending with the 2014 birth cohort). Because of this decline in herd-immunity effects over time, the direct-effects model would give an estimate of the economics of hepatitis A immunization far into the future if one were to assume no changes in such factors as vaccine cost, medical costs, and potential hepatitis A rates without immunization.
The economic model used in this analysis, fully described elsewhere,11 estimated the cost-effectiveness, from a societal perspective, of routine childhood vaccination against hepatitis A at age 1 year. In brief, the model followed a cohort of children from birth through age 95 years, during which members of the cohort were subject to mortality rates equal to the current US age-specific mortality rates. Throughout the modeling period, members of the cohort were also subject to acute hepatitis A and its associated complications. The incidence of hepatitis A that would have occurred without immunization was assumed to have declined by 1.4% per year during the entire modeling period based on a regression analysis of US hepatitis A rates after adjusting the 1995–2001 rates to compensate for the effects of immunization.6,11 The model tracked all hepatitis A–and hepatitis A vaccine-associated costs, the number of life-years (LYs) lived by the cohort, and the amount of time during which the cohort lived in various states of ill health because of hepatitis A. To assess the impact of childhood immunization, the model was run twice for each cohort, once with and once without routine childhood immunization. The results of the 2 runs were compared to assess the incremental impact of immunization.
The cost per LY saved and the cost per quality-adjusted life-year (QALY) saved were used to assess cost-effectiveness of routine immunization. The costs in the cost/LY ratio included net medical and public health costs, as well as costs from 3 categories of productivity loss: productivity loss after death from hepatitis A, productivity loss while ill with hepatitis A, and productivity loss by parents caring for children with acute hepatitis A. The costs in the cost/QALY ratio included all of these costs except productivity loss while ill with hepatitis A, which was taken into account as a component of the QALY weights and productivity losses after death.12–14 All of the costs and benefits, including dollar costs, LY, and QALY, were discounted by 3% per annum to their value in 2005.
All of the estimates were generated separately according to the 3 US regions specified in the 1999 ACIP recommendations. Region 1 included 11 states that had historically had the highest incidence of hepatitis A and for which the 1999 guidelines recommended routine immunization. Region 2 included 6 states with intermediate incidence for which the guidelines called for consideration of routine immunization. Region 3 included the District of Columbia and 33 states that had historically had the lowest incidence and for which the guidelines made no statewide recommendation. These 3 regions were also combined for a nationwide analysis.
The full model adds to the direct-effects model11 by including the additional benefits gained from herd-immunity effects by immunizing 10 birth cohorts of children, starting with the 2005 birth cohort and ending with the 2014 birth cohort. As a first step, the burden of disease attributable to hepatitis A in 2005 was estimated using the same model11 in which the age-specific incidence of hepatitis A was held constant at the estimated rate that would have occurred in the United States in 2005 had immunization not been implemented. The age-specific rates of hepatitis A–related morbidity and mortality from this model were then applied to the projected 2005 US resident population to estimate the costs incurred and the LY and QALY lost from hepatitis A had no immunization taken place in or before 2005. Costs, morbidity, and mortality attributable to hepatitis A in the years after 2005 were assumed to decline at 1.4% per year, the same rate as the decline in incidence.
In each year of the model, the benefits from herd immunity were divided equally among all of the childhood cohorts that had been immunized. For example, in 2006, the first year of nationwide routine childhood immunization in the model, all of the herd-immunity benefits were assigned to the cohort that was 1 year old in 2006. In 2007, 2 birth cohorts would have been immunized (1- and 2-year-olds), so the herd-immunity benefits were divided in 2 and applied equally to each cohort. In each subsequent year, the herd-immunity benefits were divided into progressively smaller proportions because of the increasing number of immunized cohorts. For each cohort, the total benefits from herd-immunity effects were calculated by summing these benefits over the lifetime of the cohort after discounting these benefits to their value in 2005.
Estimated Proportion of Infections Prevented by Herd Immunity
Approximately 10% of cases of hepatitis A in the United States are acquired abroad.15 In our analysis, we assumed that these cases would be prevented only by the direct effects of immunization. The other 90% are acquired domestically and, thus, can be prevented by both the direct effects of immunization and by the indirect (ie, herd-immunity) effects. We assumed, based on a previous analysis,6 that immunizing children would result in herd-immunity effects in both children and adults. Although this earlier model showed that vaccination of adults also produced herd-immunity effects among adults, we chose to ignore these in this model because of the greater degree of uncertainty of these effects and the relatively small impact they would have had on the model. The effect of this decision was to slightly reduce the benefits of immunization and, thus, to make the model slightly more conservative.
To estimate the impact of herd immunity on domestically acquired infection among children, we used model-based estimates6 that, from 1995 to 2001, a 1% increase in immunization coverage among children 0 to 19 years old resulted in an ∼3.9% decline in incidence or, more precisely, that I = I0 × (1 − v)C, where I0 is the incidence in the absence of immunization, I is the observed incidence when vaccination coverage is v, and C is a constant (3.9) determined by regression.6 For this analysis, we assumed that if the vaccination coverage was 1%, then 1% of the decline was attributable to the direct effects of immunization (ie, only the 1% of children who received the vaccine would have been protected by the direct effects), and the other 2.9% was attributable to herd-immunity effects, both within cohort and out of cohort. Because the model used here differed from the original model in that we assume that 10% of infections were not preventable by herd immunity, we adjusted C proportionately (ie, to 4.3).
We used a similar approach to estimate the impact of herd immunity from childhood immunization on adults. In the original model, D for this effect (D is analogous to C for the effect among children, above) was estimated to be 1.0. In the model presented here, we assumed that 25% of infections among adults were because of transmission from children based on national notifiable disease data and sentinel surveillance (Centers for Disease Control and Prevention, unpublished data, 1995–2004). To account for this, we adjusted D proportionately (ie, to 4.0). These relationships between vaccination coverage and the proportion of cases prevented in the population are shown graphically in Fig 1.
In a univariate analysis, we examined the sensitivity of the results to all of the herd-immunity–related variables, as well as to those variables to which the direct-effects analysis was most sensitive, including vaccination costs, the discount rate, the incidence and rate of decline in incidence of hepatitis A, the duration of immunity from hepatitis A vaccine, immunization coverage, and the quality of life lost during acute hepatitis A.11
Impact of Out-of-Cohort Herd Immunity on the First 10 Cohorts
In the direct-effects model, immunization would cost society an average of $113 million per year and would save $65.9 million, including $32.2 million in savings from decreased medical costs, $29.2 in savings from increased productivity, and $4.5 million in savings to the public health system (Table 1). The net cost to society (ie, the cost of vaccination minus the total savings) would be $47.1 million per year. The reduced rates of hepatitis A resulting from immunization would save 1792 QALYs. This gain in QALYs would be attributable mostly to reduced morbidity from hepatitis (1586 QALYs) and, to a much lesser degree, to reduced mortality from hepatitis A (206 LYs). In the direct-effects model, immunization would cost $32000 per QALY saved and $228000 per LY saved.
In the full model, inclusion of the herd-immunity benefits adds no costs to the vaccination program but increases the annual savings by $66.9 million. Most of these herd-immunity benefits ($62.7 million; Table 1) are because of infections prevented among the older, nonimmunized cohorts rather than among the relatively few unimmunized children within the birth cohort ($4.2 million). Inclusion of herd immunity in the model more than doubles the QALYs saved and more than triples the LYs saved (Table 1). Thus, in the full model, immunization costs $1000 per QALY saved and is cost-saving for each LY saved.
In the full model, immunization is cost-saving for at least the first 10 years in Regions 1 and 2 (Table 2). In Region 3, routine hepatitis A immunization would cost approximately $53000 per QALY gained and $222000 per LY gained.
Impact of Herd Immunity Over Time
The expected benefits from immunization at age 1 year, including savings from direct medical costs, public health costs, and productivity losses, decreased with each cohort immunized in the model (Fig 2). The direct benefits from immunization decreased each year by ∼1.4% because of the assumption that hepatitis A rates would continue to decrease even without immunization. The within-cohort benefits would be expected to continue as an almost constant proportion of the direct benefits. Out-of-cohort herd-immunity effects more than double the benefits of immunizing the first cohort of children but rapidly decline over the first 10 years of the program as benefits are divided between more and more cohorts. Of all out-of-cohort benefits during the first 50 years, 57% can be attributed to immunization of the first 10 cohorts.
Nationwide, hepatitis A immunization was most sensitive to vaccine acquisition and administration costs (Fig 3). Decreasing these costs to 50% of their base value or increasing them to 150% of their base value resulted in vaccination being either cost-saving or costing $17000 per QALY gained. Varying all of the other parameters resulted in smaller changes in the estimated cost per QALY. The results were insensitive to changes in the herd-immunity parameters (parameters C and D), varying from $6000 per QALY to being slightly cost-saving as these parameters were adjusted to between 50% and 150% of their base values. Assuming that children would be vaccinated at 2 years old rather than at age 1 year decreased the number of infections prevented and increased administration costs, because it was assumed that 1 of the doses would be given at a visit where no other vaccines were given. In this scenario, vaccination cost $9000 per QALY saved.
Two separate economic models are presented here, a direct-effects model, which considers only benefits accruing to the immunized cohort, and a full model, which considers both direct and herd-immunity benefits. The direct-effects model may be useful for comparison with other economic analyses, which generally disregard the population effects of public health interventions.7 However, one would expect a vaccine such as that for hepatitis A to have large herd-immunity benefits,6,7 given the underlying epidemiology of hepatitis A, as well as real-world experience where routine hepatitis A immunization has been implemented (see below). Therefore, the full model may provide a more realistic estimate of the returns that could be expected from investment in hepatitis A immunization.
Benefits of herd immunity diminish with each successively immunized cohort, and for this reason we arbitrarily chose to estimate the average economics of immunization for the first 10 cohorts of children immunized. Even before accounting for herd immunity, immunization becomes less cost-effective with each cohort because of the assumption that incidence of hepatitis A would continue to slowly decline even if there were no immunization. Therefore, because this analysis considers averages over a 10-year period, the economics of hepatitis A immunization in the cohort-only model are slightly less favorable than those estimated elsewhere11 for the 2005 cohort.
The full model shows immunization to be close to cost-neutral on average for the first 10 cohorts of children immunized. The herd-immunity benefits are greatest when immunization is first implemented because of the large impact of herd immunity in older cohorts of both children and adults. These benefits decrease with each subsequent cohort, both because there are fewer unimmunized persons in the population to benefit from herd immunity and because the benefits from herd immunity are divided among an increasing number of immunized cohorts of children, all of whom are assumed to contribute to this herd immunity. Nonetheless, the benefits of herd immunity persist for many years because of the reduced incidence among adults, for whom vaccination coverage in the model was assumed to remain low until cohorts immunized as children reach adulthood.
The large economic benefit from herd immunity is consistent with observations where hepatitis A immunization has been implemented. During a hepatitis A immunization demonstration project in Butte County, CA, for example, incidence declined by 79% among children and 44% among adults during a time when immunization coverage among children reached a maximum of 66%.16 These declines were much greater than seen elsewhere in California during the same time. Similar effects have been observed in Israel, where within 4 years of initiating of an immunization program aimed only at toddlers, the incidence of hepatitis A declined by >95% among children and >90% among adults,17 and in North Queensland, Australia, where during the first 4 years of routine hepatitis A immunization among indigenous infants the incidence of hepatitis A fell by 96% among indigenous populations and 90% among nonindigenous populations.18
A previously published economic analysis by Jacobs et al19 found routine childhood hepatitis A vaccination in the United States to be cost-effective, with a gain of 9611 QALYs nationwide at a societal cost of $1400 per QALY. In the analysis by Jacobs et al,19 herd-immunity benefits were also considered, although this was done by estimating the numbers of secondary cases among household contacts of infected children.20 As a result of this methodology and of other differences in their model, especially higher rates of hospitalization and mortality, the Jacobs et al19 analysis estimated a much larger benefit from immunization. Nonetheless, its results are similar to those from the first few cohorts of our model in that immunization was close to cost-neutral and in that almost two thirds of the benefits were attributable to out-of-cohort effects.
The analysis presented here is subject to several limitations, including those of the underlying economic model on which it is based.11 There are few data about the magnitude of herd-immunity effects under real-life circumstances, and the parameters used in our model were based on a single analysis. Nonetheless, the model was relatively insensitive to these parameters. The assumption that a maximum of 25% of adult infections were preventable by herd immunity may have been overly conservative. The model also had to make assumptions about future trends in incidence. The model was more sensitive to the assumed decline in incidence, but cost-effectiveness ratios remained reasonable even when the assumed decline was doubled.
The sensitivity analysis of the full model showed hepatitis A immunization on a national scale to be reasonably cost-effective under a wide range of assumptions. The model was most sensitive to assumptions about vaccine and administration costs. Lowering the costs of vaccine acquisition or lowering the administration costs by, for example, incorporating the hepatitis A antigen into a combination vaccine, would both result in a more favorable cost-effectiveness ratio.
In October, 2005, the ACIP voted to extend routine infant hepatitis A immunization to all children born in the United States.21 The decision was based on several considerations, including the demonstrated effectiveness of immunization, questions about the long-term sustainability of the 1999 interim recommendations, and the favorable economics of hepatitis A immunization in a direct-effects model.8 This analysis provides further support for this policy in showing that herd-immunity benefits from hepatitis A immunization are potentially very large.
- Accepted August 16, 2006.
- Address correspondence to Gregory L. Armstrong, MD, Centers for Disease Control and Prevention, Mailstop E-03, 1600 Clifton Rd NE, Atlanta, GA 30333. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- ↵Centers for Disease Control and Prevention. Prevention of hepatitis A through active and passive immunization: recommendations of the Advisory Committee on Immunization Practices. MMWR Recomm Rep.1999;48(RR-12) :1– 37
- ↵Hopkins RS, Jajosky RA, Hall PA, et al. Summary of notifiable diseases: United States, 2003 [published correction appears in MMWR Morb Mortal Wkly Rep. 2006;55:779]. MMWR Morb Mortal Wkly Rep.2005;52(54) :1– 85
- ↵Armstrong GL, Bell BP. Hepatitis A virus infections in the United States: model-based estimates and implications for childhood immunization. Pediatrics.2002;109 :839– 845
- ↵Centers for Disease Control and Prevention. Prevention of hepatitis A through active or passive immunization: recommendations of the Advisory Committee on Immunization Practices (ACIP) [published correction appears in MMWR Morb Mortal Wkly Rep. 1997;46:588]. MMWR Recomm Rep.1996;45(RR-15) :1– 30
- ↵Amon JJ, Darling N, Fiore AE, Bell BP, Barker LE. Factors associated with hepatitis A vaccination among children 24 to 35 months of age: United States, 2003. Pediatrics.2006;117 :30– 33
- ↵Brisson M, Edmunds WJ. Economic evaluation of vaccination programs: the impact of herd-immunity. Med Decis Making.2003;23 :76– 82
- ↵Anderson RM, May RM. Infectious Diseases of Humans. Oxford, United Kingdom: Oxford University Press; 1991
- ↵Melegaro A, Edmunds WJ. Cost-effectiveness analysis of pneumococcal conjugate vaccination in England and Wales. Vaccine.2004;4203– 4214
- ↵Trotter CL, Edmunds WJ. Reassessing the cost-effectiveness of meningococcal serogroup C conjugate (MCC) vaccines using a transmission dynamic model. Med Decis Making.2006;26 :38– 47
- ↵Rein DB, Hicks KA, Wirth BA, et al. Cost-effectiveness of routine childhood vaccination for hepatitis A in the United States. Pediatrics.2007;119(1) . Available at: www.pediatrics.org/cgi/content/full/119/1/e12
- ↵Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-effectiveness in Health and Medicine. New York, NY: Oxford University Press; 1996
- Haddix AC, Corso PS, Gorsky RD. Costs. In: Teutsch SM, Corso PS, eds. Prevention Effectiveness: A Guide to Decision Analysis and Economic Evaluation. New York, NY: Oxford University Press; 2003
- ↵Centers for Disease Control and Prevention. Hepatitis Surveillance Report Number 60. Atlanta, GA: Centers for Disease Control and Prevention; 2005:1–51
- ↵Advisory Committee on Immunization Practices; Fiore AE, Wasley A, Bell BP. Prevention of hepatitis A through active or passive immunization: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep.2006;55(RR-7) :1– 23
- Copyright © 2007 by the American Academy of Pediatrics