## Abstract

*Context.* The US Food and Drug Administration approved a meningococcal conjugate A/C/Y/W-135 vaccine (MCV-4) for use in persons aged 11 to 55 years in January, 2005; licensure for use in younger age groups is expected in 2 to 4 years.

*Objective.* To evaluate and compare the projected health and economic impact of MCV-4 vaccination of US adolescents, toddlers, and infants.

*Design.* Cost-effectiveness analysis from a societal perspective based on data from Active Bacterial Core Surveillance (ABCs) and other published and unpublished sources. Sensitivity analyses in which key input measures were varied over plausible ranges were performed.

*Setting and Patients.* A hypothetical 2003 US population cohort of children 11 years of age and a 2003 US birth cohort.

*Interventions.* Hypothetical routine vaccination of adolescents (1 dose at 11 years of age), toddlers (1 dose at 1 year of age), and infants (3 doses at 2, 4, and 6 months of age). Each vaccination scenario was compared with a “no-vaccination” scenario.

*Main Outcome Measures.* Meningococcal cases and deaths prevented, cost per case prevented, cost per life-year saved, and cost per quality-adjusted life-year saved.

*Results.* Routine MCV-4 vaccination of US adolescents (11 years of age) would prevent 270 meningococcal cases and 36 deaths in the vaccinated cohort over 22 years, a decrease of 46% in the expected burden of disease. Before program costs are counted, adolescent vaccination would reduce direct disease costs by $18 million and decrease productivity losses by $50 million. At a cost per vaccination (average public-private price per dose plus administration fees) of $82.50, adolescent vaccination would cost society $633000 per meningococcal case prevented and $121000 per life-year saved. Key variables influencing results were disease incidence, case-fatality ratio, and cost per vaccination. The cost-effectiveness of toddler vaccination is essentially equivalent to adolescent vaccination, whereas infant vaccination would be much less cost-effective.

*Conclusions.* Routine MCV-4 vaccination of US children would reduce the burden of disease in vaccinated cohorts but at a relatively high net societal cost. The projected cost-effectiveness of adolescent vaccination approaches that of recently adopted childhood vaccines under conditions of above-average meningococcal disease incidence or at a lower cost per vaccination.

N*eisseria meningitidis* is a leading cause of bacterial meningitis and septicemia in children and adolescents in the United States, with high case fatality and morbidity despite good medical care. Disease rates are highest among children <2 years of age, and slightly elevated rates have been observed among adolescents and young adults.^{1}

Vaccines against meningococcal disease, based on the polysaccharide capsule, have been in use in the United States since the 1970s. The formulation currently available covers serogroups A, C, Y, and W-135, which have caused approximately two thirds of US cases of meningococcal disease in recent years.^{1} Although the currently licensed polysaccharide vaccine is safe and effective, it is not recommended for routine use because of its relative ineffectiveness in children <2 years of age and its short duration of protection. Economic analysis of routine meningococcal polysaccharide vaccination of entering college freshmen who will live in dormitories (a group known to have a moderately elevated risk of meningococcal disease relative to the general population) projected substantial social costs for this program relative to estimated benefits.^{2}

Widespread use of *Haemophilus influenzae* type b (Hib) and *Streptococcus pneumoniae* conjugate vaccines, in which a carrier protein is conjugated to the polysaccharide to produce a T-cell–dependent response, has resulted in dramatic reductions in the burden of disease caused by these 2 pathogens.^{3,4} Serogroup C conjugate meningococcal vaccine has been developed and administered routinely in the United Kingdom since 1999,^{5} resulting in a dramatic decline in disease.^{6} A quadrivalent meningococcal conjugate polysaccharide A/C/Y/W-135 vaccine (MCV-4) was shown recently to be safe and immunogenic in healthy adults.^{7} In January, 2005, the US Food and Drug Administration approved MCV-4 for use in adolescents and adults aged 11 to 55 years. Submissions of applications for licensure for this and similar vaccines for use in younger age groups are expected to follow.

Analysis of US meningococcal disease surveillance data in the context of comparing potential strategies for routine use of a conjugate meningococcal vaccine has shown that a toddler strategy consisting of 1 dose administered at 1 year of age would bring about the largest reductions in cases and deaths.^{8} However, the potential costs and cost-effectiveness of such strategies in the United States have not been evaluated. National policy makers will soon formulate recommendations for the use of meningococcal conjugate vaccines in the United States, first among adolescents and adults but, in the near future, among toddlers and infants as well. To inform the most immediate MCV-4 policy decision, we evaluated the projected health benefits, costs, and cost-effectiveness of routine conjugate meningococcal vaccination of US adolescents from a societal perspective. For the purposes of comparison and to inform likely policy decisions in the near future, we also projected the cost-effectiveness of toddler and infant MCV-4 vaccination strategies.

## METHODS

### Design

We used a cohort-simulation approach in which meningococcal disease outcomes and costs were measured in a hypothetical population over a defined time frame. To evaluate the cost-effectiveness of routine adolescent vaccination, we modeled routine administration of a single dose of MCV-4 to a hypothetical “adolescent” cohort of US children 11 years of age. To evaluate the cost-effectiveness of routine infant and toddler vaccination, we modeled 2 MCV-4 vaccination strategies in a hypothetical 2003 birth cohort: an infant strategy consisting of a 3-dose regimen given at 2, 4, and 6 months of age and a toddler strategy consisting of a single dose given at 1 year of age. Immunization schedules for each strategy were based on serogroup C meningococcal vaccine immunogenicity and efficacy studies.^{9,10} Although recent data suggest that the addition of a booster dose to a 2-, 3-, and 4-month infant schedule may be required,^{11} our analysis included only the 3-dose primary vaccination series for infants (Fig 1).

We used a 22-year time frame to examine the impact of vaccination through 2 phases of life during which rates of disease have historically peaked: early childhood and late adolescence/early adulthood.^{1} For each case occurring within the 22-year time frame, lifetime costs and effects of meningococcal disease were calculated based on age-adjusted life expectancy at disease acquisition.

We used a population size of 1000000 for each cohort for performing model simulations and calculating cost-effectiveness ratios. We multiplied sum values for health outcomes and costs by the appropriate ratio derived from US Census estimates of the size of the 2003 US birth cohort (*n* = 4026538) and number of children 11 years of age (*n* = 4238672) so that reported totals would reflect expected values if modeled strategies were adopted as policy (Appendix 1).

We projected costs and outcomes under a “no-vaccination” strategy to create a baseline for comparison to vaccination strategies. We calculated the cost of vaccination programs, including the cost of care for associated adverse events to vaccination. Based on differences between vaccination and no-vaccination options and the cost of vaccination programs we estimated the potential benefits of conjugate meningococcal vaccination, which include reduced morbidity and mortality from meningococcal disease and its sequelae, treatment cost savings, and productivity savings. All costs were adjusted to 2003 US dollars. Based on recommendations of the US Panel on Cost Effectiveness in Health and Medicine, all future costs and benefits were discounted at a 3% annual rate.^{12} Models were created and analyses were performed by using Excel 2002 (Microsoft Inc, Redmond, WA) and @RISK 4.5.2 (Palisade Corporation, Newfield, NY).

### Disease Incidence

Disease-incidence estimates were based on surveillance for invasive disease caused by *N meningitidis* conducted as a part of Active Bacterial Core Surveillance (ABCs).^{1,13} This ongoing, active, laboratory- and population-based surveillance is a part of the Emerging Infections Program Network coordinated by the Centers for Disease Control and Prevention in collaboration with participating state and local health departments and universities. We used ABCs data collected from January 1, 1993, through December 31, 2002. The continuously participating surveillance sites during this time were: Connecticut, Maryland, Minnesota, and Oregon and multicounty areas in Georgia, California, New York, and Tennessee. The aggregate population under surveillance ranged from 12 million in 1993 to 34 million in 2002.

We defined a case of meningococcal disease as the isolation of *N meningitidis* from a normally sterile site, such as blood or cerebrospinal fluid, from a resident of the surveillance area between January 1, 1993, and December 31, 2002. Ten-year average age- and serogroup–specific incidence rates among the surveillance population were calculated by using population denominators from the US Census (Fig 2). These rates were race adjusted and projected to the US population. The following age groups were used: 0 to 4 months, 5 to 11 months, 1 year (12 to 23 months), 2 to 4 years, 5 to 10 years, 11 to 17 years, 18 to 22 years, and 23 to 32 years. The following serogroup categories were used: B, C, Y, W-135, and “other.” There were no reported cases of serogroup A meningococcal disease. When the serogroup was unknown, serogroup was assigned based on the distribution of known serogroups.

### Disease Incidence Under Vaccination Strategies

To each age- and serogroup-specific rate, we applied the formula
where, *DI*_{vacc} is disease incidence under vaccination strategy, *DI*_{novacc} is disease incidence without vaccination, *V*_{cov} is vaccination coverage, and *V*_{eff} is vaccine efficacy.

For each strategy (vaccination and no vaccination) and within each age group, the serogroup-specific rates were summed, generating an age-group–specific all-serogroups incidence. This rate then was multiplied by the population denominator to generate the number of expected cases of meningococcal disease. The populations of birth and adolescent cohorts (*n* = 1000000) were decreased each year by subtracting the expected number of deaths calculated by multiplying the appropriate age-specific all-cause US mortality rate (year 2000) by the population of the cohort.^{14} We assumed a constant annual incidence rate with no variation in serogroup distribution.

### Disease Outcomes

Our model allowed for 3 primary outcomes of meningococcal disease: survival without sequelae, survival with long-term sequelae, and death (Fig 1). The most common long-term sequelae among survivors of meningococcal disease are skin scarring, amputations, hearing loss, and neurologic disability, but survivors also may be left with renal failure and septic arthritis requiring joint replacement.^{15–17} Although a small proportion of meningococcal disease survivors have been reported to have >1 long-term sequela,^{16} for the purposes of this analysis we divided long-term sequelae into 5 mutually exclusive categories: skin scarring, single amputation, multiple amputations, hearing loss, and significant neurologic disability (Fig 1). Probabilities of each of these 5 sequelae were based on published sources (Table 1). Although recurrent meningococcal disease has been reported under exceedingly rare circumstances in individuals with specific immune deficiencies, we assumed that meningococcal disease could occur only once. To determine the number of deaths from meningococcal disease each year, we multiplied the projected number of cases by age-group–specific case-fatality ratios (CFRs) derived from ABCs data (Fig 2). We calculated years of life lost (for deaths) and years of survival with sequelae by subtracting the age at which meningococcal disease was acquired from the appropriate age-group–specific US Census estimate of average life expectancy in 2000.^{18}

### Quality of Life

Our model accounted for decrements in quality of life among survivors of meningococcal disease with long-term sequelae. We did not attempt to measure the presumptive decrease in quality of life that occurs during an episode of acute disease, because meningococcal disease follows a rapid clinical course.^{19,20} Because there are few data on the quality of life among survivors of meningococcal disease,^{21} we obtained health-utility rates for conditions closely resembling each of the 5 long-term sequelae in the model (Table 1). For skin scarring we assumed that there would be no adjustment for quality of life in the base case (health-utility rate = 1.0). We based this assumption on multiple studies of pediatric burn victims with skin scarring that failed to show any long-term impairments in physical or psychosocial functioning.^{22,23}

Our health-utility indices (HUIs) for meningococcal disease survivors with single or multiple amputations were based on observations of the quality of life of trauma patients after undergoing amputation of an extremity.^{24} For hearing loss, we used the general health assessment of postlingually deaf adults after cochlear implant placement.^{25} For neurologic disability, we used the mean global health utility as scored by patients with severe Alzheimer's disease using the Health Utilities Index Mark-3 questionnaire.^{26} We calculated quality-adjusted life-years (QALYs) lost by multiplying the years of life remaining with each long-term sequela by the sequela-specific health-utility rate.

### Vaccine: Efficacy and Coverage

The vaccine used in this analysis is a quadrivalent conjugate meningococcal vaccine (covering serogroups A, C, Y, and W-135). We based our estimates of vaccine efficacy on that of the serogroup C conjugate meningococcal vaccine used routinely in the United Kingdom^{6,10,27} (Table 1). Persons to whom the vaccine is administered as toddlers (1 year of age) and adolescents (11 years of age) were assumed to have full protection immediately after receiving their first and only dose of vaccine. We assumed no vaccine-induced protection from disease in infant-vaccinated members of the birth cohort until 1 month after the first 2 doses of vaccine were administered (ie, at 5 months of age).^{28}

Because other protein antigen vaccines have been shown to confer protection that lasts for up to 18 years,^{29–31} and based on their ability to induce immunologic memory,^{28} conjugate meningococcal vaccines are expected to offer an extended duration of protection from disease. Based on this expectation, we assumed vaccine-induced protection for the time frame of the analysis (22 years). To account for the presumed waning of vaccine-induced immunity, we assumed that the protective efficacy dropped by 25%, 10 years after vaccination, and then remained constant at this level for the remainder of the analytic time frame (22 years). We did not incorporate a herd-immunity effect into any of our assumptions, maintaining our focus on only the direct protective effects of vaccination.

We sought to model the effects of a conjugate meningococcal vaccination program that was fully implemented; thus, we used vaccination-coverage data from established US infant, toddler, and adolescent routine vaccination programs (Table 1). For infants, we used national 3-dose Hib vaccination coverage in US children 19 to 35 months of age.^{32} For toddlers, we used national 1-dose measles-mumps-rubella (MMR) vaccination coverage in US children by 24 months of age.^{32} Because national estimates of coverage for vaccines administered to adolescents are unavailable, we based our adolescent vaccination-coverage estimate on observations of 3-dose hepatitis B vaccination coverage in San Diego County, California, seventh-graders 1 year after a school-entry law went into effect mandating hepatitis B vaccination for entry into middle school.^{33}

### Costs Associated With Disease

To determine the total cost of meningococcal disease under each strategy, we added all costs from the following categories: costs related to the acute event of meningococcal disease, costs related to meningococcal sequelae, and the economic value of life/cost of productivity losses (Appendix 2).

### Cost Related to the Acute Event

Costs related to the acute episode of meningococcal disease (as opposed to those related to long-term sequelae of meningococcal disease) were those arising from (1) medical care for acute meningococcal disease, (2) parents' time attending to a child with meningococcal disease, and (3) public health response to an isolated case of meningococcal disease. We used parents' lost time from work as a proxy for the value of time spent attending to a child with meningococcal disease and based our estimate on previous observations of the value of lost wages of parents of children with bacteremia and pneumonia.^{34}

We calculated the cost of the public health response to an isolated case of meningococcal disease by formulating estimates for the average number of hours of work devoted by public health departments to a single reported case of meningococcal disease,^{35} the average number of case contacts requiring chemoprophylaxis (G. Birkhead, MD, MPH, New York State Department of Health, written communication, 2004), the average hourly wage of an epidemiologist working in the public sector,^{36} and the average wholesale price of meningococcal chemoprophylaxis (2-day course of oral rifampin 600 mg administered twice per day)^{35–37} (Appendix 3).

### Costs Related to Meningococcal Sequelae

Cases of sequelae were treated as a subset of nonfatal cases. Persons in the model who suffered sequelae incurred sequela-specific medical costs in addition to costs related to acute meningococcal disease (Appendix 2). Lifetime sequelae-specific medical and other direct costs are detailed in Table 2.

For single and multiple amputations, we included both an acute treatment cost (eg, cost of the amputation procedure)^{2} and lifetime medical costs related to rehabilitation, such as fitting and maintenance of a prosthetic limb.^{38} We assumed that each survivor with hearing loss would receive a cochlear implant and incorporated lifetime costs related to its placement and maintenance.^{39} For neurologic disability, we based our estimate on observed costs of caring for individuals with developmental disabilities, and we incorporated the cost of long-term residential care and special education.^{40,41}

### Economic Value of Life/Cost of Productivity Losses

To estimate the economic value of life, we used the human capital approach, which assumes that the value to society of an individual's life is equivalent to the value of productivity lost as measured by the present value of the individual's future income stream. We used published age-specific estimates of productivity losses due to death based on average wages for males and females in the United States (labor market earnings) and the value of unpaid labor (household production) to calculate the economic value of life lost.^{42} For meningococcal disease survivors with neurologic disability, we assumed the cost of productivity losses to be equivalent to the projected labor market earnings only. We assumed that persons with multiple amputations would lose 30% of lifetime labor market earnings and persons with hearing loss would lose 33% of lifetime labor market earnings.^{43}

### Cost Per Vaccination

The manufacturer has yet to announce its US list price for the quadrivalent conjugate meningococcal vaccine. We assumed that the cost per dose of vaccine would be equivalent to the inflation-adjusted price of the pneumococcal conjugate vaccine the first year that it was licensed for use in the United States (2000).^{44} We calculated an overall average cost per vaccination that included assumptions for (1) differing public- and private-sector prices per dose of vaccine, (2) proportion of all vaccine doses bought at the public- versus private-sector price, (3) differing public- and private-sector administration fees, and (4) proportion of all children receiving vaccinations in public health versus private-sector clinics (Table 2; Appendix 4).^{45}

### Vaccine-Associated Adverse Events

We used the moderate and severe adverse-event rates noted during the United Kingdom meningococcal serogroup C conjugate vaccination campaign.^{46} We assumed that the cost of a moderate adverse event was equivalent to a physician visit for otitis media excluding fees for antibiotics and laboratory services.^{47} We assumed that the cost of a serious adverse event was equivalent to US hospitalization and outpatient charges for anaphylaxis.^{45}

### Cost of Vaccination Program

To determine the cost of vaccination programs, we multiplied the number of persons receiving vaccine by the cost per vaccination and number of doses required and then added the total cost of moderate and serious adverse events.

### Economic Analysis

To estimate the economic returns of a societal investment in routine conjugate meningococcal vaccination, we calculated the net present value (NPV) of each vaccination strategy (Appendix 5). NPV is equal to the cost of the vaccination program minus the benefits of averting disease through vaccination. We calculated the benefits of averting disease by subtracting the total cost of disease under a vaccination strategy (including all costs related to acute meningococcal disease, meningococcal sequelae, and the cost of productivity losses due to meningococcal deaths and long-term sequelae) from the cost of disease in the baseline no-vaccination strategy.

Average cost-effectiveness ratios were calculated as dollars invested in the vaccination program minus dollars saved due to disease episodes averted, divided by health outcomes. (Appendix 5) Four cost-effectiveness ratios were calculated based on different health outcomes in the denominator: cost per case prevented, cost per death averted, cost per life-year saved, and cost per QALY saved. We excluded all costs of productivity loss from the numerator of the cost per QALY saved, because this cost is implicit in the assessment of HUIs in the denominator.^{12} Consistent with recent published economic analyses of childhood vaccination programs,^{34,45} productivity losses due to death were not included in calculating dollars per life-year saved.

We calculated the benefit/cost (BC) ratio for each scenario to provide a summary measure of each vaccination program relative to the “no-program” option. The BC ratio is equal to the benefits ($) of averting disease through vaccination divided by vaccination-program costs.

### Sensitivity and Scenario Analyses

We performed a 1-way deterministic sensitivity analysis focused on the following variables: (1) disease incidence, (2) CFR, (3) rates of long-term sequelae, (4) HUIs, (5) vaccine efficacy, (6) vaccination coverage, (7) acute meningococcal disease costs, (8) lifetime costs of meningococcal disease sequelae, and (9) cost per vaccination. For each variable, we set high and low estimates according to the specified ranges for each component of the variable (Tables 1 and 2; Appendix 6).

We also analyzed the sensitivity of our assumption for waning immunity by calculating outcomes under the assumptions of (1) full vaccine-induced protection for the time frame of the analysis (22 years) and (2) a drop in protective efficacy to 0% 10 years after vaccination.

We modeled best-case (most favorable to vaccination) and worst-case (least favorable to vaccination) scenarios by using the appropriate high or low estimate for variables 1 to 8 above. For example, our best-case scenario used estimates at the high end of our defined range for the following variables: disease incidence, CFR, rates of long-term sequelae, HUIs, vaccine efficacy, acute meningococcal disease costs, and lifetime costs of meningococcal sequelae.

For base-, best-, and worst-case scenarios, we conducted threshold analyses to determine the break-even value for cost per vaccination, ie, the value needed such that the sum of the benefits equals the sum of the costs.

We performed a Monte Carlo analysis in which several parameters were varied simultaneously over specified probability distributions. All incidence and CFR estimates had triangular distributions; types of distributions used for other estimates varied in the Monte Carlo analysis are specified in Tables 1 and 2. The numbers of iterations were determined based on the optimal number needed to reach stability in simulated values with a 1.5% convergence from iteration to iteration. We selected the median as the primary result and 5th and 95th percentiles to approximate confidence intervals around the primary result. We conducted a scenario analysis for cost per vaccination in which a new Monte Carlo simulation was performed at each of several fixed values for cost per vaccination.

## RESULTS

Table 3 shows projected disease outcomes for adolescent, toddler, and infant vaccination strategies compared with a no-vaccination strategy of the same cohort. In the base case or most-likely scenario, routine MCV-4 vaccination of US 11-year-olds would prevent 270 cases and 36 deaths in this cohort over 22 years. Compared to the adolescent strategy, toddler or infant vaccination of a birth cohort would prevent more meningococcal cases and a similar number of deaths resulting from meningococcal disease over the same amount of time.

In an unvaccinated US 11-year-old cohort, the societal cost of meningococcal disease is projected to be $146 million over 22 years, including $38 million (26%) in medical and other direct costs and $108 million in productivity losses. Adolescent vaccination would reduce the societal cost by 46%, including $18 million in direct disease costs and $50 million in the cost of productivity loss (Table 4). In comparison, the societal cost of meningococcal disease in a US birth cohort over the same time frame is projected to be $191 million, including $79 million (41%) in direct disease costs. Routine toddler vaccination would reduce the societal cost of disease by 35%; infant vaccination would reduce the same cost by 40%.

Using our most-likely estimates for the public- and private-sector prices for cost per dose of MCV-4 and administration fees, we project overall cost per MCV-4 vaccination to be $82.50. At this cost per vaccination, total adolescent vaccination-program costs would be $227 million. Subtracting the expected savings in meningococcal disease costs, the net cost of a routine adolescent MCV-4 vaccination strategy is $159 million (societal perspective) or $210 million (direct costs only). Routine toddler and infant vaccination-program societal net costs would be $239 million and $854 million, respectively.

### Cost-Effectiveness and BC Ratios

Table 5 shows the estimated cost per meningococcal case prevented and other cost-effectiveness ratios for adolescent, toddler, and infant strategies. Adolescent and toddler strategies both would cost approximately $630000 per case prevented, whereas infant vaccination would cost >3 times this amount per case prevented. Adolescent vaccination would cost slightly less per death averted and per life-year saved than would toddler vaccination. Figure 3 shows how the cost per life-year saved varies depending on the cost per vaccination. At costs per vaccination below $70, the adolescent strategy's cost per life-year saved is less than $100000; at costs per vaccination below $50, the cost per life-year saved is less than $50000.

### Sensitivity Analysis

Of the 9 parameters varied, the cost-effectiveness outcomes of all 3 vaccination strategies were most sensitive to variations in disease incidence, followed by CFR and cost per vaccination. For the adolescent cohort, as incidence estimates were increased from values approximating the lowest rates in the last 10 years to 10-year highs, the cost per life-year saved decreased from $190000 to $95000. Figure 4 shows the range of values for cost per life-year saved at high, low, and average disease-incidence rates and at varying levels of cost per vaccination.

Variation of estimates for acute meningococcal disease costs (medical care, public health response, etc) and lifetime costs of long-term sequelae (limb prostheses for amputees, residential care for the neurologically disabled) had negligible effects on the cost-effectiveness of any of the 3 strategies. Increases or decreases in vaccination coverage were linearly related to overall vaccination-program costs but did not affect cost-effectiveness ratios. Cost per QALY saved was most sensitive to variations in the rate of long-term sequelae; alterations to HUIs had little effect on this ratio.

Changes to the estimate for waning vaccine-induced immunity had a much greater effect on the cost-effectiveness of toddler and infant strategies than on that of an adolescent strategy. If vaccine-induced protection disappears completely beginning 10 years after immunization in all vaccinated persons, the adolescent strategy cost per life-year saved increases by 25% to $153000, whereas the same ratio for the toddler strategy increases by 105% to $338000.

### Threshold Analyses and Best/Worst-Case Scenarios

The “break-even” cost per vaccination (cost at which conjugate meningococcal vaccination would be cost saving to society [NPV ≤ 0]) is $23 for an adolescent strategy and $18 for a toddler strategy in the base-case scenario. In the best-case scenario (most favorable to vaccination), adolescent vaccination would cost $29000 per life-year saved, and have a break-even cost per vaccination of $76. In the worst-case scenario (least favorable to vaccination), the adolescent strategy would cost $494000 per life-year saved, with a break-even cost per vaccination of less than $10. Break-even costs per vaccination for the infant strategy were less than $10, even under the best-case scenario.

## DISCUSSION

Routine quadrivalent (A, C, Y, W-135) conjugate meningococcal vaccination of US children would reduce substantially the burden of disease in vaccinated cohorts, but compared with many currently recommended preventive measures, projected costs are high. MCV-4 cost-effectiveness ratios (eg, cost per life-year saved) are consistent with a historical trend of decreasing overall cost-effectiveness of new childhood vaccines adopted for routine use (Table 6). Well-established childhood immunizations such as MMR, diphtheria-pertussis-tetanus, and polio vaccinations have been shown to be cost saving even when only the direct costs of disease are counted.^{45,48} Two routine vaccinations first recommended in the 1990s, infant varicella and hepatitis B (infant and adolescent), were projected to have a net cost when only direct costs were counted but to be cost saving when indirect costs were considered.^{49,50} Routine infant pneumococcal conjugate vaccination, adopted in 2000, was projected to have a net cost both from the health care payer (direct costs only) and societal perspectives.^{34} Adolescent MCV-4 vaccination, at its expected cost per vaccination, also has a net cost from both perspectives even when model assumptions match those of our best-case scenario. Under the most-likely conditions, MCV-4 adolescent and toddler cost-effectiveness ratios exceed those of recently adopted routine childhood vaccination programs, but this difference narrows with small but plausible alterations in key model assumptions such as disease incidence, CFR, and cost per vaccination. Routine infant MCV-4 vaccination would prevent the most cases but would be far less cost-effective than an adolescent or toddler strategy.

We used a static cohort design for this analysis to facilitate comparisons to cost-effectiveness analyses of other childhood vaccines done before their routine use, which used similar methods.^{34,49,50} This design, while conservative, does not have the capacity to account for a major potential benefit of routine conjugate meningococcal vaccination: reduction of disease in unvaccinated persons through routine vaccination, or herd immunity. In the United Kingdom, a 67% reduction in disease in unvaccinated children was observed in the third year of their serogroup C conjugate meningococcal vaccination program.^{51} Whether the United States would reap similar benefits is unclear, especially because the timing of US licensure for different age groups precludes replication of the United Kingdom's massive catch-up campaign targeting all children 0 to 17 years of age.^{5} The United Kingdom's approach probably maximized herd immunity by dramatically decreasing asymptomatic carriage of *N meningitidis*, a phenomenon that has been observed only in the first few years after conjugate meningococcal vaccination, and may not persist.^{52}

Many of our assumptions are drawn from the United Kingdom's experience with routine serogroup C conjugate vaccination; thus, our analysis is limited by the ways in which meningococcal epidemiology and available conjugate vaccines differ in the United States and United Kingdom. We assumed that the proportion of vaccine-preventable serogroups would remain stable, as they have in the United Kingdom,^{52} despite both the relatively unpredictable nature of meningococcal serogroup distribution^{53} and the theoretical risk of increased carriage of vaccine-resistant meningococci induced by routine vaccination. We assumed that the rate of adverse events to vaccination in the United States would be as low as that observed in the United Kingdom. We assumed a duration of immunity of 22 years, with a modest decline in efficacy after 10 years. If a booster dose is necessary to sustain high levels of protective efficacy, as suggested by recent observations among children vaccinated according to the 2-, 3-, and 4-month schedule in the United Kingdom,^{11} the cost-effectiveness of a US vaccination program would decline substantially. Waning immunity had a greater impact on toddler and infant strategy cost-effectiveness, because the assumed declines in protection coincided with the modest rise in disease incidence that occurs during adolescence.

We did not consider the potential economic benefits of a reduction in the number of outbreaks of meningococcal disease. Routine vaccination would probably decrease the overall number of outbreaks; however, the direct and measurable costs of outbreak response are small relative to those that are significant yet intangible. These costs include public panic, intense media coverage, ad hoc meningococcal disease education campaigns, school closures, and general upheaval in local health departments and health care facilities.^{54} Likewise, we did not account for the cost of psychologic trauma in survivors, which is commonly observed^{16,21} but difficult to quantify. The impact of meningococcal disease cannot be wholly accounted for by any single analysis, and cost-effectiveness is only 1 of the measures that should be used to inform a policy decision on the routine use of conjugate meningococcal vaccines in the United States.

Our sensitivity analyses identified disease incidence as a highly influential variable regardless of strategy. Given known geographic patterns in US meningococcal disease epidemiology, in which some US states and counties have had an elevated incidence of meningococcal disease sustained over several years^{55} (Centers for Disease Control and Prevention, unpublished data, 2004), geographically targeted recommendations might maximize the cost-effectiveness of conjugate meningococcal vaccination policy, if feasible from a programmatic perspective.

The most imminent decision faced by policy makers is whether to initiate routine vaccination of US adolescents. Vaccination of all 11-year-olds in the United States, as described in this model, would lower the incidence of disease within that cohort and may be deemed cost-effective by policy makers when weighed against competing priorities. However, because meningococcal cases in a single age cohort make up a small part of the overall US burden of meningococcal disease, it is unlikely that such a policy would result in dramatic reductions in US meningococcal disease in the near term. “Catch-up” vaccination of older adolescents (12–18 years of age) not previously vaccinated, in addition to 11-year-olds, may soon become a policy consideration. Given its greater potential for generating herd immunity through widespread reduction in asymptomatic carriage of *N meningitidis*, the cost-effectiveness of such an approach warrants additional study.

## APPENDIX 3. Cost of Public Health Response Equation

where; *PH*_{wage} is the per-hour wage for a public health worker, *PHW*_{hours} is hours of work per case of meningococcal disease, *C*_{chemoprophy} is the cost of chemoprophylaxis, and *N*_{contacts} is the average number of contacts requiring chemoprophylaxis per case of meningococcal disease

## APPENDIX 4. Calculation of the Cost Per Vaccination

where *C*_{vacc} is cost per vaccination, *V*_{pub} is the public-sector price of vaccine ($), *P*_{pub} is the proportion of vaccine purchased at the public-sector price (%), *AF*_{pub} is the administrative cost for vaccination during a visit to a public clinic ($), *A*_{pub} is the proportion of vaccinees receiving vaccine in public health clinics (%), *V*_{prv} is the private-sector price of vaccine ($), *P*_{prv} is the proportion of vaccine purchased at the private-sector price (%), *AF*_{prv} is the administrative cost for vaccination during a visit to a private health care provider or other private-sector clinic ($), and *A*_{prv} is the proportion of vaccinees receiving vaccine from a private health care provider or other private-sector clinic (%).

## APPENDIX 5. Economic Analysis: Formulas

We calculated NPV for vaccinated and unvaccinated cohorts according to the formula
(1) where *C* is cost of the vaccination program (including cost of adverse events), *B* is the benefits (savings in cost of disease with vaccination program), *t* is the time in years after immunization, *T* is life expectancy, and *r* is the discount rate (3%).

We calculated average cost-effectiveness ratios according to the formula
(2) where *ACE* is the average cost-effectiveness ratio, *NPV*_{vacc} is the NPV of the vaccination strategy, *t* is the time in years after immunization, *T* is the life expectancy, and *r* is the discount rate (3%).

The BC ratio is computed as
(3) where *C* is cost of the vaccination program (including cost of adverse events), *B* is the benefits (savings in cost of disease with vaccination program), *t* is the time in years after immunization, *T* is the life expectancy, and *r* is the discount rate (3%).

## APPENDIX 6. Rationale for Selection of Model Input Variables Listed in Tables 1 and 2 and in Fig 2

### Table 1

#### Sequelae

The upper ranges for single and multiple amputation rates are based on the total amputation rate in refs 16, 56, and 57, split into single and multiple amputation rates based on the ratio of single to multiple amputations among meningococcal disease survivors in ref 16.

#### Health-Utility Indices

The lower bound for skin scarring and the upper bound for single amputation both are based on the highest EuroQol score given by the 4 meningococcal survivors interviewed in ref 21. The lower bounds for single and multiple amputations are based on the HUI for “maximal major amputation” in ref 60. The base-case HUI for multiple amputations is the 3-month postdischarge Quality of Well-being score in ref 24. For hearing loss, the base case and upper bound are adult study subject post–cochlear implant utilities according to the SF-36 instrument and HUI-2, respectively, in ref 25. The lower bound is the HUI for deaf children post–cochlear implant as scored by their parents in ref 39. For neurologic disability, the base case and upper bound are equal to the HUI-3 for “severe” and “mild” Alzheimer's disease, respectively, as scored by patients in ref 26.

#### Vaccine Efficacy

For each age group, the upper and lower bounds are equivalent to the confidence intervals cited in ref. 10.

#### Vaccination Coverage

For infants 5 to 11 months old, the base-case estimate is equivalent to national 2+ Hib coverage by 5 months of age in 2002–2003; the lower bound is equivalent to national 2+ pneumococcal conjugate vaccine coverage by 5 months of age in 2002–2003; the higher bound is equivalent to the highest reported 2+ Hib coverage of any state in 2002–2003 (Massachusetts).

For infants ≥1 year of age, the lower bound is equivalent to 2002 national 1+ varicella coverage by 19 to 35 months; the higher bound is equivalent to 3+ dose diphtheria, tetanus, and pertussis vaccine (DT/DTP/DTaP) in 1999, which is the highest national coverage rate for any single vaccine from 1998 to 2002.

For toddlers, the lower bound is equivalent to national 3+ pneumococcal conjugate vaccine coverage in 2002–2003; the higher bound is equivalent to the highest reported 1+ MMR vaccination coverage of any state in 2002–2003 (Connecticut).

For adolescents, the lower bound is equivalent to the rate of 3+ hepatitis B vaccination coverage in fifth- and sixth-grade children in San Diego County, California, in 1998, the year before a seventh-grade school-entry law for hepatitis B vaccination went into effect^{33}; the higher bound is equivalent to the 2+ measles-containing vaccine coverage rate among seventh-graders in San Diego County, California, 2 years after a school-entry law requiring such vaccination was in place.^{33}

### Table 2

#### Medical Care for Single and Multiple Amputations

We used cost estimates from the orthopedic surgery literature for a temporary prosthetic limb, a durable or “permanent” prosthetic limb, annual prosthesis maintenance and supplies, and repairs and socket modifications to derive an estimate of the lifetime medical cost of amputation.^{38} We assumed that a prosthetic limb would be needed for a total of 50 years; that a temporary prosthesis would be needed only once; and that durable prostheses would need to be replaced every 4 years. Annual and recurring costs were converted to the present value assuming a 3% discount rate. For those persons with multiple amputations, we increased our assumptions for acute and lifetime medical costs by 20%.

#### Medical Care for Hearing Loss

For our estimate of the lifetime cost of hearing loss we relied on a recently published cost-utility analysis of cochlear implant use in profoundly deaf US children.^{39} We used the estimate of direct medical costs only: initial cost of the device, upgrades and maintenance, surgery to place the implant, outpatient preoperative and postoperative visits, and audiology follow-up. The lower bound for this estimate is derived from published national estimates of these costs^{62}: direct medical costs for hearing loss ($132000000) divided by the quotient of total costs for hearing loss ($2102000000) divided by the average cost per person ($417000).

#### Residential Care for Neurologic Disability

We assumed that persons with neurologic sequelae would require residential care for 50 years. The base case assumes care in an intermediate facility. The annual cost of residential care ($97307) was converted to lifetime costs by using a 3% discount rate.^{40} The higher bound is based on the estimate for care in a large state facility, and the lower bound is based on the estimate for the cost of home- and community-based care.

#### Special Education for Neurologic Disability

Persons who contracted meningococcal disease before the age of 10 were assumed to require 16 years of special education; persons acquiring meningococcal disease between the ages of 10 and 18 years were assumed to require 10 years of special education. We did not assume any cost for special education if persons acquired their infection after they were 17 years old. Calculation of the base-case estimate uses the average annual per-pupil expenditure for mental retardation; calculation of the lower bound uses the same value for the “multiple disabilities” pupil; and calculation of the higher bound uses this value for the “speech/language impairment” pupil. The annual cost of special education ($16242) was converted to lifetime costs by using a 3% discount rate.^{41}

#### Cost Per Dose of Vaccine

The lower bound for the cost per dose of vaccine (public sector) is equal to the 2004 public-sector price per dose of pneumococcal conjugate vaccine,^{63} and the upper bound is equal to the price of meningococcal polysaccharide vaccine in the private sector,^{64} decreased by the same magnitude (16.4%) as the public-sector discount for pneumococcal conjugate vaccine.

The lower bound for cost per dose of vaccine (private sector) is equal to the 2004 private-sector price per dose of pneumococcal conjugate vaccine, and the upper bound is based on the price of a single dose of the most expensive vaccine available (Japanese encephalitis vaccine), as determined by an informal search by the authors.

### Figure 2: Upper and Lower Bounds for Sensitivity Analysis

We identified upper and lower bounds for disease incidence by multiplying age- and serogroup-specific incidence by 120% and 46%, respectively. The multiplier used to create the upper bound is based on disease incidence in 1995, the year in which the overall meningococcal disease incidence (ABCs) had the greatest upward deviation from the 1993–2002 average incidence. The multiplier used to create the lower bound is based on disease incidence in 2002, the year in which the overall meningococcal disease incidence (ABCs) had the greatest downward deviation from the 1993–2002 average incidence.

We identified the upper and lower bounds of CFR by multiplying each cell by 131% and 34%, respectively. The multiplier used to create the upper bound is based on the CFR in 2002, the year in which the CFR had the greatest upward deviation from the average CFR from 1993–2002 according to the National Notifiable Disease Surveillance System. The lower bound is based on the CFR in 1999, the year in which the CFR had the greatest downward deviation from the average CFR from 1993 to 2002 according to the National Notifiable Disease Surveillance System.

## Acknowledgments

The following are the Active Bacterial Core Surveillance Team members. California: G. Rothrock, P. Daily, L. Gelling, D. Vugia, and A. Reingold (Emerging Infections Program); Connecticut: N. Barrett, J. Hadler, P. Mshar, C. Morin, Q. Phan, and M. Cartter (Emerging Infections Program); Georgia: D. Stephens, W. Baughman, S. Whitfield, M. Bardsley, and M. Farley (Veterans Affairs Medical Center and Emory University School of Medicine, Atlanta); Maryland: D. Blythe, L. Sanza, A. Schmidt, and L. Harrison (Emerging Infections Program); Minnesota: M. Osterholm, R. Danila, J. Rainbow, C. Lexau, L. Triden, K. White, J. Besser, and R. Lynfield (Emerging Infections Program); Missouri: R. McPherson, M. Bright, and M. Huber; New York: D. Morse, P. Smith, N. Bennett, B. Hatheway, B. Damaske, S. Zansky, N. Spina, and G. Smith (State Department of Health); Oklahoma: L. Smithee, G. Istre, and P. Quinlisk; Oregon: K. Stefonek, B. Robinson, P. Cieslak, J. Donegan, M. Dragoon, and H. Homan (Oregon Department of Health Services and Multnomah County Health Department); Tennessee: A. Craig, B. Barnes, C. Gilmore, L. Lefkowitz, P. Adams, W. Schaffner, and P. Rados (Department of Public Health); and Centers for Disease Control and Prevention: G. Ajello, M. Berkowitz, B. Plikaytis, M. Reeves, K. Robinson, S. Bernhardt, E. Zell, K. Shutt, A. Schuchat, and S. Schmink.

We thank Mick Ballesteros, Oleg Bilukha, Gus Birkhead, Phaedra Corso, Reginald Finger, Marc Fischer, Daniel Fishbein, Carolyn Greene, Orin Levine, Jai Lingappa, Mark Messonnier, Martin Meltzer, Ted Shepard, Jim Turner, Kirk Winger, Elizabeth Zell, and the Meningococcal Working Group of the Advisory Committee on Immunization Practices.

## Footnotes

- Accepted February 7, 2005.
- Address correspondence to Colin W. Shepard, MD, Centers for Disease Control and Prevention, 1600 Clifton Rd, MS G37, Atlanta, GA 30333. cvs8{at}cdc.gov
No conflict of interest declared.

*Haemophilus influenzae*type b • MCV-4, quadrivalent meningococcal conjugate polysaccharide A/C/Y/W-135 vaccine • ABCs, Active Bacterial Core Surveillance • HUI, health-utility index • QALY, quality-adjusted life-year • MMR, measles-mumps-rubella • NPV, net present value • BC, benefit-cost • CFR, case-fatality ratio

## REFERENCES

- Copyright © 2005 by the American Academy of Pediatrics