Clinical and Economic Impact of a Combination Haemophilus influenzae and Hepatitis B Vaccine:  Estimating Cost-effectiveness Using Decision Analysis FREE

IMMUNIZATION AGAINST preventable childhood illnesses is a cornerstone of public health policy. Although public policy and immunization awareness programs have consistently improved vaccination rates in the United States, the number of children who receive the complete immunization series as recommended by the Advisory Committee on Immunization Practices (ACIP) of the US Public Health Service remains suboptimal. Lack of access to health care providers has been discussed as a possible barrier to immunization, but other factors also affect immunization compliance.

Vaccine compliance is related to the numbers of required clinic visits and of necessary injections at each visit.¹ Thus, efforts to combine 2 or more vaccines into a single injection may promote compliance.^2- 3 The incremental impact of multivalent injections is probably due to increased uptake of those vaccines included in the combination that had a less than desirable acceptance rate before the combination was available.

In the United States, compliance with the required 3-injection series of hepatitis B virus (HBV) vaccine was slightly higher than 75% in 1996.⁴ Despite an ACIP recommendation that all newborns be vaccinated⁵ and numerous articles demonstrating the favorable cost-effectiveness of the vaccine in infants and children,^6- 10 this acceptance rate remains lower than those of other ACIP-recommended vaccines, all of which are 90% or greater. In 1996, nearly 95% of infants completed the Haemophilus influenzae type b (Hib) vaccine series in monovalent form or as part of a multivalent vaccine (diphtheria and tetanus toxoids, pertussis, and Hib). A new combination HBV-Hib vaccine (Comvax, Merck and Company, Inc, West Point, Pa) was recently approved for use.¹¹

If compliance with the Hib vaccine remains high, HBV-related morbidity and mortality may be reduced further should caregivers decide to substitute the HBV-Hib combination for the existing Hib vaccine, without the need for an extra provider visit or another injection. Although it is unclear how many children currently receiving Hib vaccine but not monovalent HBV vaccine would actually receive the HBV-Hib combination, the clinical and economic consequences of incorporating this new vaccine into routine practice should be assessed.

In the present cost-conscious health care environment, clinical effectiveness and economic impact of medical innovations are being scrutinized. For example, an economic analysis found that introducing the safer inactivated polio vaccine into the routine vaccination schedule was not cost-beneficial from a societal perspective.¹² Accordingly, we used decision analysis to compare the clinical impact and cost-effectiveness of the following vaccination strategies: Hib vaccine given alone or as part of a multivalent injection that does not include HBV vaccine, with monovalent HBV vaccine given separately (current practice); and a multivalent HBV-Hib vaccine (combination strategy).

We developed a decision analytic model to estimate the clinical and economic consequences of HBV and Hib disease for a hypothetical cohort of newborns for each of the immunization strategies from the health care payer perspective. The model projected the lifetime burden of illness caused by HBV or Hib infection, or both, and the total costs associated with vaccination and HBV and Hib disease. Natural history models for HBV and Hib disease were constructed as decision trees with embedded Markov processes^13- 14 by using variables based on a critical review of published reports. The target population was a representative cohort of 1 million neonates born in the United States.

We evaluated both infant immunization strategies. In current practice, 75% of children complete the monovalent HBV series, and 95% complete the separate Hib series. These completion rates are similar to those recently reported by the Centers for Disease Control and Prevention (CDC).⁴ Infants received a 2.5-µg dose of monovalent HBV vaccine if results of maternal screening for HBV infection were negative; all other infants received a 5.0-µg dose. In the combination strategy, 95% of children complete the combined HBV-Hib series. For infants of mothers in whom HBV infection is not detected, these vaccines are given at 2, 4, and 12 months of age.

Both immunization strategies incorporated maternal screening for active HBV infection (testing for hepatitis B surface antigen [HBsAg]). Compliance with maternal screening in the United States averages 85%.¹⁵ Among mothers with a positive test result, we assumed that passive HBV immunization (hepatitis B immune globulin) would be administered to their infants. On the basis of currently available prescribing information, the HBV-Hib vaccine should not be used at birth for infants born to mothers with test results positive for active HBV infection. In both strategies, therefore, only monovalent HBV and Hib vaccines were administered to these infants. The HBV vaccine was administered at 0, 1, 2, and 6 months of age, and the Hib vaccine at 2, 4, and 12 months of age.

The natural history model was divided into the following 2 stages: a perinatal stage for the transmission of HBV from mother to neonate (vertical transmission) and a horizontal stage for transmission after birth. Separate natural history models were created for patients in whom acute HBV or Hib disease developed. On the basis of a review of the literature,^16- 74 we derived variables describing the outcomes of acute and chronic phases of HBV (Table 1) and Hib disease (Table 2).

The purpose of the perinatal stage was to predict transmission of HBV from mother to neonate at birth. Because maternal transmission of Hib does not occur, only HBV infection was included in this stage.

The model divided mothers into 1 of the following 3 risk strata: those with negative results of HBsAg screening (no risk), those with positive results of HBsAg screening and negative results of hepatitis B e antigen (HBeAg) screening (high risk), and those with positive results of HBsAg and HBeAg screening (highest risk). Risks for transmission with no intervention (ie, no screening or treatment with HBV immune globulin) were derived from published reports of mother-child studies and from the control arms of HBV vaccine trials.^16- 20

For each infant, there were 5 possible HBV-related outcomes at the end of the perinatal period. The infants could be susceptible to HBV, vaccinated (1 dose) against HBV, or immune to HBV as a result of recovery from infection; they could have chronic HBV infection as a result of infection; or they could die of fulminant HBV infection (Figure 1). All infants who survived the perinatal period remained susceptible to Hib, because no Hib vaccination had been given.

The distribution of the outcomes of the perinatal stage was based on the distribution of mothers in the risk strata, the use of maternal screening, the rate of vertical transmission of HBV from mothers to newborns, and the effectiveness of passive and active vaccination. The variables describing the risks for perinatal infection, type of acute infection, and outcome of HBV infection for this period are listed in Table 1.

The remainder of the model (ie, the horizontal stage) was designed as a Markov process. Depending on the perinatal outcome, infants progressed through the simulation in specific health states with or without HBV or Hib disease (Table 3 lists the potential health states). Movement among states was determined by a set of transition probabilities based on the epidemiological data and natural history of both diseases (Figure 2 and Figure 3).

An individual infant could progress through many health states in a lifetime. Some of these states were standard transition states (eg, chronic carrier infection); some were temporary and thus were assigned only resource use (eg, acute asymptomatic hepatitis); and others were considered absorbing states (eg, hepatoma), in that on entrance to one of them, patients remained there until death. For each immunization strategy, the sum of the time alive in each of the standard transition and absorbing states (except death) equaled life expectancy.

Time periods (cycles) for the model were 1, 5, or 10 years in duration; the longer periods were used for the last 30 years of the model. Decisions about the length of a cycle for particular ages in the patients' lives were based on variability in the natural history of disease (eg, probabilities of onset or progression of disease) and on the availability of data.

Patients progressed through various health states in the model, as determined by their risk for acute HBV-Hib disease and disease attack rates. Patients remained in the simulation until they died of an HBV- or Hib-related illness or died of other causes.

The epidemiological data of HBV differ from one population to another, making an accurate simulation of the natural history of this disease complex. Infection with HBV can be divided into acute and chronic stages. Acute HBV disease can be transmitted from an infected mother to a newborn or horizontally through blood exposure or intimate physical contact. Perinatal HBV transmission is rare in the United States, unlike HBV-endemic areas. Acute HBV infection occurs most commonly in the young or middle adulthood years by way of sexual contact, blood exposure, or injected drug use (horizontal transmission).

Horizontal HBV infection was modeled in varying time intervals by using age-related rates of acute HBV infection. The risk for HBV infection was estimated from surveillance reports published by the CDC and modified to account for underreporting (most acute HBV infection in adults is asymptomatic).^21,75

Independent of the mode of transmission, acute HBV infection was usually asymptomatic but could be symptomatic with jaundice and/or other symptoms or with fulminant hepatic failure. The outcomes of acute HBV infection were modeled as complete recovery and natural immunity from further HBV infection, chronic HBV infection and possible development of chronic sequelae, or death associated with fulminant hepatitis (Figure 2).

All patients who contracted acute HBV infection without fulminant hepatic failure survived the initial episode. Patients who had fulminant hepatic failure had different outcomes, with death being the predominant consequence. In contrast to acute HBV infection occurring in childhood, acute cases of HBV infection occurring in adolescence and adulthood were less likely to result in chronic HBV infection. The age-dependent risk for HBV infection, type of infection, and outcomes are listed in Table 4.

Depending on the age of the patient at which the acute infection developed, HBV persisted in some infected patients and resulted in chronic HBV infection. The clinical presentation of chronic HBV disease ranged from asymptomatic to fatal complications of end-stage cirrhosis or liver cancer. Transition probabilities among chronic HBV health states were derived primarily from longitudinal studies of native Alaskans with chronic HBV infection.^35,38 At any time in the model, patients with chronic HBV could be in 1 of the following 5 clinical health states: immune after hepatitis, chronic carrier, chronic active hepatitis, cirrhosis, and hepatoma.

Patients who recovered from acute or chronic HBV infection were naturally immune to acute hepatitis, and therefore chronic hepatitis could never develop. Patients remained in this state unless acute Hib disease developed (assuming they were not immune).

Survivors of acute HBV infection who remained infected but did not show clinical signs or symptoms of ongoing hepatitis became chronic carriers. Most were asymptomatic, although liver enzyme levels could fluctuate. The costs of surveillance of these laboratory test results to monitor for hepatic sequelae were included. On the basis of natural history data, patients could continue in this state or progress to active hepatitis, cirrhosis, or hepatoma. We assumed that a small proportion of patients (1.8% per year) would spontaneously recover from infection and become immune from further infection.⁴²

Some survivors of acute HBV infection remained actively infected with HBV. This state included persons with chronic active and chronic persistent hepatitis, which we distinguished by their histological characteristics. Patients could continue in this state or regress back to a chronic carrier state. These persons are also at a high risk for progressing to cirrhosis or hepatoma. A small proportion of patients received interferon therapy. We assumed that a small fraction of patients in this state would also spontaneously recover from infection and become immune from further infection.

Chronic carriers and patients with chronic active HBV infection may progress to cirrhosis. We assumed that patients who entered this state would eventually die of a liver-related complication, ie, decompensated cirrhosis or hepatocellular carcinoma, with survival ranging from 4 to 8 years.^76- 79 A small proportion of patients undergo liver transplantation but eventually die of an HBV-related complication.^80- 81

Chronic carriers and patients with chronic active hepatitis or cirrhosis may progress to hepatoma (hepatocellular carcinoma). These patients were assumed to have a poor overall survival, with an average life expectancy of 1 year.^82- 83

We previously developed a decision analytic model of Hib disease that was used to evaluate vaccination policy in Australia.⁸⁴ This model was updated for use in this analysis. A natural history decision tree for Hib disease is shown in Figure 3. Important Hib-related variables used in the principal analysis are listed in Table 2. We assumed that preventable Hib disease occurred between birth and 5 years of age. Acute cases of Hib disease were calculated at monthly intervals until 3 years of age and then at annual intervals until 5 years of age.

For patients susceptible to Hib, the incidence of Hib disease was based on surveillance studies in the United States during the time of early introduction of Hib vaccination programs.⁴³ Current Hib incidence rates in the United States do not reflect those of previous cohorts of nonimmunized children. The number of cases of Hib disease has sharply decreased as a result of excellent vaccination coverage. Thus, use of incidence of Hib infection before the availability of an effective vaccine would probably overestimate the number of cases in a nonimmunized cohort because these rates do not account for the effect of herd immunity. The probability of Hib infection was independent of the probability of HBV infection.

Acute Hib infection could clinically manifest itself in the model as meningitis, epiglottitis, or other, less common presentations of Hib disease such as pneumonia or depending on the severity of acute disease and the patient's age.^52- 57

With the exception of persons with Hib-associated meningitis, all survivors of acute Hib disease were assumed to have no chronic sequelae from their illness and to have a normal life expectancy. Survivors of meningitis were at risk for permanent neurologic sequelae as a result of Hib infection. Rates of neurologic sequelae after acute meningitis were drawn from a modem series of studies.^58- 63 Severe chronic disabilities were defined as intractable seizures, incapacitating mental retardation, any institutionalization, or neurologic deficits as severe as or worse than hemiplegia. Less severe sequelae referred to all other deficits (eg, deafness) not included under the severe heading. Life expectancy was modeled separately for patients with severe neurologic sequelae on the basis of observational studies of institutionalized and noninstitutionalized mentally challenged children.^85- 86 Patients with mild deficits were assumed to have a mortality rate that is 50% higher than the rate in the general population.

At any time in the model, each disease-free patient was protected against HBV alone, protected against Hib alone, protected against HBV and Hib, or unprotected against both. We assumed that protection against HBV developed from successful vaccination or natural immunity acquired after HBV infection. Because the literature has identified rare cases of loss of protective immunity for HBV,⁷⁴ we assumed that a small proportion of patients who were fully protected against HBV became susceptible (0.2% per year, or a cumulative risk of 15% at 65 years of age). Protection against Hib was assumed to develop only after successful vaccination, and once protected, patients never again became susceptible. We assumed that the probabilities of being protected from HBV and Hib were independent.

Effectiveness of HBV vaccination was estimated from randomized and nonrandomized studies of recombinant HBV vaccines.^87- 92 In the perinatal stage, a complete series of HBV immunization and a single dose of HBV immune globulin at birth resulted in 92% to 98% protection from vertical HBV transmission. In the horizontal stage, we assumed that, on the basis of long-term pediatric follow-up and clinical trials in adolescents and adults,^93- 95 protection from HBV infection was nearly equivalent to protection achieved from vaccination at birth. On the basis of clinical data for monovalent and combination HBV vaccines, we assumed that the HBV component of the combination HBV-Hib vaccine had efficacy equivalent to that of the monovalent HBV vaccine.⁹⁶ No partial immunity was attributed to infants who had less than the complete series of HBV vaccine in the principal analysis. The impact of partial compliance was explored using sensitivity analysis.

The effect of Hib vaccination was calculated as the reduction in incidence of acute Hib disease and its sequelae in each cohort. Vaccine effectiveness was derived from long-term efficacy studies of Hib immunization within specific populations.^97- 98 We estimated that 3 doses of Hib vaccine would result in 95% protection against acute disease for at least 5 years. Given clinical trial data supporting the approval of the combination vaccine, we assumed that the Hib component of the combination HBV-Hib vaccine had efficacy similar to that of the monovalent Hib vaccine in the principal analysis.⁹⁶

Values for direct medical costs of health care used in the model for preventive services, acute HBV and Hib disease, and long-term sequelae for the most part were derived from published reports (Table 5). Costs of lost productivity (eg, indirect costs) and nonmedical direct costs (eg, travel to medical visits and lodging of family members) were not included in the principal analysis.

Cost of prevention program included direct costs related to maternal testing for HBsAg; however, because they were equal in both immunization strategies, they had little effect on incremental costs. However, the test results affected the total cost of care because they determined whether active and passive HBV immunization occurred during the hospitalization at the time of birth.

Costs of vaccines were determined from government and private sources and included the cost of supplies and administration fees (Table 5). For the principal analysis, March 1997 CDC prices were used. To minimize a potential advantage of the combination vaccine, we assumed no reduction in the number of office visits. A $2 administration fee was included for the second injection of the monovalent vaccines.

Direct medical costs of HBV and Hib disease were associated with time spent in each transition and absorbing state (except death). Costs were also assigned to temporary states that did not have a time component. The sum of these costs represented the total costs of disease for each cohort.

In the principal analysis, costs and years of life were discounted at 5% per year. All costs were reported in 1996 US dollars.

We compared the costs (including the costs of the preventive program and of acute and chronic illness related to HBV or Hib) and associated clinical outcomes (measured as years of life) of both immunization scenarios. If the total costs of both strategies differed, an incremental cost was determined. Similarly, if a difference in years of life between strategies resulted, an incremental number of years of life saved was calculated. Cost-effectiveness ratios were calculated as the ratio of the incremental costs and incremental years of life saved.

Because the efficacy of (95%) and rate of compliance with (95%) Hib vaccine were assumed to be similar in the principal analysis, the primary clinical outcome was years of life saved as a result of incremental reduction in the rate of HBV disease. Thus, the main result of the economic analysis was the estimation of a cost-effectiveness ratio (costs per HBV-related year of life saved) describing the extra cost necessary to achieve the added benefits of combination HBV-Hib vaccine when compared with its monovalent components. We used a ratio of $50,000 per year of life saved as a standard against which to compare strategies, in part because this value has been suggested as a threshold for acceptable interventions.⁹⁹

We performed a number of sensitivity analyses to determine the effects of several assumptions on the cost-effectiveness of treatment. We varied the compliance with the combination vaccine (±5%), the prevalence of maternal HBV carriage (±50%), the incidence of HBV infection (±50%), the percentage of cases of acute infection that become chronic (±50%), the lifetime loss of protection from HBV vaccination (±100%), the cost of the combination vaccine (±$5), and the direct medical costs associated with HBV disease (±50%). Because of recent recommendations of the Panel on Cost-Effectiveness in Health and Medicine,¹⁰⁰ we repeated our analysis using a discount rate of 3% and performed sensitivity analyses using discount rates of 1% and 7%. In addition, we evaluated the economic effect of providing a birth dose of HBV vaccine for infants of mothers who did not undergo screening along with combination vaccine at 2, 4, and 12 months of age, as this was not incorporated in the principal analysis.

Our study complied with published guidelines for minimizing bias in economic analysis funded by pharmaceutical companies.¹⁰¹

No difference in Hib-related outcomes was seen between strategies, because the efficacy and compliance rates of the Hib vaccine were assumed to be equivalent in the principal analysis (Table 4). In each strategy, with Hib vaccine compliance rates of 95%, the model estimated that 308 acute cases of Hib infection would occur per 1 million infants. Of infected infants, 12 (3.9%) would die of a Hib-related illness, and 47 (15.2%) would have chronic neurologic sequelae after surviving Hib-related meningitis.

By using an HBV vaccine compliance rate equal to that of the Hib vaccine in the combination strategy, the model estimated that the 80% reduction in noncompliance (from 25% to 5%) resulted in a 53% reduction in the number of acute cases of HBV infection (8541 per 1 million) when compared with current practice (5230 cases with the combination strategy and 18,044 cases with the current practice strategy). This decrease in the number of cases of acute HBV infection prevented more than 823 cases of chronic HBV disease. Compared with current practice, the combination strategy resulted in 205 fewer HBV-related deaths (30 deaths due to acute and 175 deaths due to chronic HBV infection averted). These averted HBV-related deaths translated into 417 discounted years of life saved per 1 million infants entered into the simulation.

We found that the cost of the vaccination program of the combination strategy was $11.5 million higher (12%) than the cost of current practice ($108.4 million compared with $96.9 million). This difference was almost entirely attributable to the increased coverage with the combination vaccine.

Consistent with the equivalent Hib clinical outcomes reported above, no difference in the resources devoted to the treatment of acute and chronic Hib disease was seen between the strategies ($11.3 million). The clinical benefits of the combination strategy translated into a 54% reduction ($4.0 million) in the costs of HBV-related disease compared with current practice ($7.5 million in current practice and $3.5 million in the combination strategy). The savings of the combination strategy were proportionally distributed between acute and chronic HBV-related disease.

When compared with monovalent components, incorporation of the HBV vaccine into a multivalent vaccine led to 417 additional years of life saved at an incremental cost of $7.3 million (extra cost of immunization program [$11.4 million] minus cost of disease saved [$4.1 million]). The incremental cost-effectiveness ratio was $17,700 per year of life saved. This ratio is considerably less than the $50,000 per year of life saved standard that we selected for evaluating both strategies.

The results of the 1-way sensitivity analyses indicated that for all variables, the HBV-Hib combination continued to satisfy the criterion of $50,000 per year of life saved ratio (Table 6). The cost-effectiveness ratios were relatively insensitive to assumptions about the prevalence of maternal carriage, the percentage of cases of acute infection that became chronic, the lifetime loss of HBV protection, and the direct medical costs of HBV disease.

The cost-effectiveness of the combination vaccine strategy was sensitive to assumptions about the incidence of HBV infection, cost and compliance with the combination vaccine, and the discount rate. The epidemiological data of HBV infection were validated in our simulation in 2 ways. First, the model predicted a lifetime risk for HBV infection similar to that of population-based studies before the advent of an effective vaccine. Second, the number of individuals with chronic HBV infection in our model was similar to that projected by the CDC model evaluating the role of a universal hepatitis B vaccination program.⁶

The impact of variation of the cost of the combination vaccine on the cost per year of life saved is shown in Figure 4. The cost per life-year saved increases with the combination vaccine price. For example, if the cost of the combination vaccine was $40 (rather than the $33.85 we assumed), the cost per year of life saved increases from less than $20,000 to $57,600. If we assumed the addition of a birth dose of HBV vaccine (with no associated benefit) for infants of mothers not undergoing screening in the combination strategy, this would increase total costs by $2.4 million, leading to a cost per life-year saved of $23,500.

The model was also sensitive to the level of compliance with the combination vaccine. Figure 5 displays a 2-way sensitivity analysis that explicitly demonstrates the trade-off between combination vaccine compliance and cost. At any given price for the combination vaccine, as the combination vaccine uptake falls from universal acceptance, the cost per life-year saved increases. For example, if at $34 a dose, only 90% of patients received the combination vaccine instead of the targeted 95%, the cost per life-year saved would increase from $18,700 to $35,800.

Because the clinical benefits of any immunization program take effect long after the financial investment is made, changes in the discount rate substantially affect the ratios. When we adopted a 3% rate, as recommended by the Panel on Cost-Effectiveness in Health and Medicine,¹⁰⁰ the incremental ratio for the combination strategy was reduced to $4100 per year of life saved. However, if we used a 7% rate (the high end of the recommended range) instead, the ratio would be as high as $46,100 per year of life saved.

The clinical and economic benefits attributable to combination vaccines are mainly related to improved vaccination compliance. In modern medicine, it is relatively uncommon that a more effective intervention leads to a net decrease in health care expenditures. In most instances, we must pay more to realize gains in clinical outcomes. To justify the routine use of a combination vaccine, the health outcome advantages that stem from incremental cases of preventable disease averted need to be balanced against the increased cost of the multivalent vaccines themselves.

Before broad public policy decisions about the value of a better vaccine are made, health benefits attained and subsequent added expenditures related to the innovation (prevention program and disease-related effects) must be carefully assessed. The resultant cost-effectiveness ratio should provide decision makers with important information on whether to recommend, provide, and reimburse the intervention under investigation.

Although relatively rare in the United States, acute and chronic HBV infection is responsible for more severe morbidity than many preventable childhood illnesses. Despite the strongest of recommendations, compliance with the required 3-dose series of HBV vaccine remains below that of other childhood immunizations. Reasons for this disappointing acceptance rate have not been well elucidated, but parents' desire to minimize the number of immunization visits and physicians' preferences to give a limited number of injections per visit most likely play a key role.^102- 103

Multivalent vaccine technology has produced a combination HBV-Hib vaccine now available in the United States. If caregivers who provide children with Hib vaccine in the monovalent or multivalent form accept the new combination product at established Hib compliance rates (95%), our analysis indicated that HBV-related benefits would be substantial. More than 9500 cases of HBV infection would be averted per 1 million infants, leading to approximately 417 extra years of life saved. If extrapolated to the US birth cohort of 4 million annually, nearly 40,000 cases of HBV infection would be prevented, translating to more than 1600 life-years saved.

The incremental health benefits of the combination vaccine, however, do not come without an incremental cost. The combined vaccine added $7 per infant in extra costs, all due to the incremental coverage with the combination vaccine. This cost translated into a ratio of cost per year of life saved of $17,700 for the combination strategy, compared with that of current practice.

We also found that the cost-effectiveness of the combination strategy was relatively robust in 1-way sensitivity analysis. Variation of our assumptions about prevalence of maternal HBV carriage, incidence of HBV infection, risk for chronic HBV infection, loss of HBV protection, direct medical costs associated with HBV infection, and the discount rate all yielded cost-effectiveness ratios less than the standard of $50,000 per year of life saved. The cost of the combination vaccine would have to exceed $40 per dose to surpass this benchmark for cost-effectiveness if compliance with the combination vaccine reached 95%. If compliance were to approximate 90%, the combination vaccine remained cost-effective at prices less than $35 per dose.

In the principal analysis, we assumed that all infants who would have received Hib vaccine in any form would receive the combination HBV-Hib vaccine instead. This may seem unrealistic, given that many children receive the Hib vaccine as part of a combined diphtheria and tetanus toxoids, pertussis, and /Hib vaccine.¹⁰⁴ In this case, a switch to a combination HBV-Hib vaccine would require an additional injection, thereby negating one of the main advantages of a combination vaccine. However, Hib vaccine is not currently licensed for use in a primary series for infants in combination with the newly recommended acellular pertussis vaccine. When the acellular pertussis vaccine is used, the HBV-Hib vaccine would probably be substituted for the monovalent Hib vaccine, because in either circumstance, Hib immunization would require a separate injection.

There is also controversy as to the necessity of the birth dose of HBV vaccination. Many obstetric centers are trained to administer a birth dose as part of routine neonatal care. The combination strategy evaluated in this model does not incorporate such a dose, unless results of maternal screening for HBsAg were positive. Advocates for the birth dose believe it may provide protection for newborns whose mothers escaped screening and are potential carriers of HBV. If such a dose is added to the current combination strategy, it remains cost-effective (ie, a cost-effective ratio <$25,000), even under the assumption that no benefit is gained from this earlier dose. Efforts should be directed toward screening to determine who gains from passive and active immunization at birth.

The reductions in HBV disease worthy of incremental expenditures on a combination vaccine as estimated by the simulation are clearly linked to the enhanced uptake of combination vaccine when compared with monovalent substitutes. Given there are no empirical data to validate our assumption that combination vaccine compliance will reach 95%, sensitivity analysis that explicitly examines the trade-off between the price of the combination vaccine and compliance provides cost-effectiveness estimates under a number of potential situations.

Our analysis indicated that if the adoption of the combination vaccine were to reduce coverage for Hib, its value for the additional cost of the vaccine would be reduced. Alternatively, if the combination vaccine were to increase Hib coverage, its value would be further enhanced. Theoretically, the higher cost of the combination vaccine could lead to lower acceptance rates than a lower-priced product. However, coverage that is decreased because of increased cost may be offset by factors relating to the convenience and efficacy of the combination vaccine. The fact that the combination allows for 2 vaccines in the same injection, thereby reducing the pincushion effect, may be viewed as favorable by parents and clinicians. Second, the Hib component of the combination vaccine requires 1 less injection than many of the widely used Hib vaccines (3 compared with 4).

Finally, our results may undervalue the use of the combination, because we only accounted for direct medical costs to the payer. The model estimates included no benefits that can be attributed to the convenience of the combination vaccine (eg, fewer visits), no nonmedical costs associated with premature morbidity and mortality (eg, lost productivity), and no secondary effect of a smaller infectious population on the rate of transmission (ie, herd immunity).

We conclude that, from the perspective of the health care payer, an investment in a combination HBV-Hib vaccine to increase HBV compliance rates is wise on clinical and economic grounds. If we were to extend our analysis to include patient preferences, improvements in quality of life, and indirect cost savings associated with preventable disease, this cost-effective innovation may well be cost saving if examined from the societal perspective.

Accepted for publication July 13, 1998.

Supported by a grant from Merck and Company, Inc.

Presented in part at the 31st Annual National Immunization Conference, Detroit, Mich, 1997.

Reprints: A. Mark Fendrick, MD, Division of General Medicine, University of Michigan Medical Center, 3116 Taubman Center, Ann Arbor, MI 48109-0376 (e-mail: amfen@umich.edu).

Editor's Note: With a superficial glance, some of you might assume that the length of this article is not a cost-effective way to spend your time. However, I assure you that your decision to carefully analyze the entire article would be very wise.—Catherine D. DeAngelis, MD