New Vaccines Against Otitis Media: Projected Benefits and Cost-effectiveness
OBJECTIVE. New vaccines that offer protection against otitis media caused by nontypeable Haemophilus influenzae and by Moraxella catarrhalis are under development. However, the potential health benefits and economic effects of such candidate vaccines have not been systematically assessed.
METHODS. We created a computerized model to compare the projected benefits and costs of (1) the currently available 7-valent pneumococcal conjugate vaccine, (2) a candidate pneumococcal–nontypeable H influenzae vaccine that has been tested in Europe, (3) a hypothetical pneumococcal-nontypeable H influenzae–Moraxella vaccine, and (4) no vaccination. The clinical probabilities of acute otitis media and of otitis media with effusion were generated from multivariate analyses of data from 2 large health maintenance organizations and from the Pittsburgh Child Development/Otitis Media Study cohort. Other probabilities, costs, and quality-of-life values were derived from published and unpublished sources. The base-case analysis assumed vaccine dose costs of $65 for the 7-valent pneumococcal conjugate vaccine, $100 for the pneumococcal-nontypeable H influenzae vaccine, and $125 for the pneumococcal-nontypeable H influenzae–Moraxella vaccine.
RESULTS. With no vaccination, we projected that 13.7 million episodes of acute otitis media would occur annually in US children aged 0 to 4 years, at an annual cost of $3.8 billion. The 7-valent pneumococcal conjugate vaccine was projected to prevent 878000 acute otitis media episodes, or 6.4% of those that would occur with no vaccination; the corresponding value for the pneumococcal–nontypeable H influenzae vaccine was 3.7 million (27%) and for the pneumococcal–nontypeable H influenzae–Moraxella vaccine was 4.2 million (31%).
Using the base-case vaccine costs, pneumococcal–nontypeable H influenzae vaccine use would result in net savings compared with nontypeable 7-valent pneumococcal conjugate use. Conversely, pneumococcal–nontypeable H influenzae–Moraxella vaccine use would not result in savings compared with pneumococcal–nontypeable H influenzae vaccine use, but would cost $48 000 more per quality-adjusted life-year saved. The results were sensitive to variations in assumptions on vaccine effectiveness and vaccine dose costs but not to variations in other assumptions.
CONCLUSIONS. New candidate vaccines against otitis media have the potential to prevent millions of disease episodes in the United States annually. If priced comparably with other recently introduced vaccines, these new otitis vaccines could achieve cost-effectiveness comparable with or more favorable than that of the 7-valent pneumococcal conjugate vaccine.
Otitis media is a leading cause of childhood morbidity. Its occurrence is the most common reason for prescribing antimicrobial agents1,2 and serves as the main basis for undertaking tympanostomy-tube insertion, the most frequently performed operation in children in the United States.3 Pneumococcal conjugate vaccine is somewhat effective against acute otitis media (AOM) but prevents only a small percentage of all episodes.4–6 Before this vaccine's introduction, Streptococcus pneumoniae was identified in ∼25% to 40% of all AOM cases, whereas nontypeable Haemophilus influenzae (NTH) was found in 23% to 30% and Moraxella catarrhalis was found in 10% to 15%.7,8 Because of the introduction of the 7-valent pneumococcal conjugate vaccine (PCV7) in 2000, the proportion of AOM episodes caused by S pneumoniae has decreased, whereas the proportion of AOM episodes caused by NTH has increased.9,10
Vaccines anticipated to be more effective against otitis media are under development.11 These include candidate vaccines with activity against NTH and M catarrhalis, which, together with S pneumoniae, constitute the leading bacterial causes of AOM. In a recent prelicensure trial in the Czech Republic, a combination vaccine against S pneumoniae and NTH reduced the occurrence of AOM episodes by 34%.12
For otitis media vaccines to be widely administered, their use needs to be accepted by parents, clinicians, and policy makers. One study suggests that parents would readily accept otitis media vaccines for their children, even if effectiveness were limited.13 National policy-making groups, including the Advisory Committee on Immunization Practices for the Centers for Disease Control and Prevention, routinely request information on the projected health benefits and economic impact of new vaccination programs being considered. Thus, economic evaluations of candidate vaccines against otitis media are needed to inform choices regarding their additional development and clinical use.5,14–16
Estimating the projected health benefits and economic effects of new otitis media vaccines is complex for a number of reasons. First, AOM is often recurrent to a varying degree, and recurrent AOM is a main indication for tympanostomy-tube insertion.14 Second, studies of PCV7 suggest that vaccination may have different degrees of effectiveness against recurrent AOM and against tympanostomy-tube insertion than against single episodes of AOM.5,15,16 And third, vaccine effectiveness may wane over time. Thus, developing accurate projections of the effects of new otitis media vaccines requires decision-analytic modeling as well as the availability of detailed empirical data on the natural history of otitis media.
We developed a decision-analytic model to address the lack of information concerning the potential benefits and cost-effectiveness of current and candidate vaccines against otitis media. Our aims were to (1) evaluate their projected benefits and cost-effectiveness, and (2) describe the requirements for effectiveness and costs that would make such vaccines economically viable.
We developed a model to estimate the health benefits, costs, and cost-effectiveness of 3 different vaccines against pathogens causing AOM in infants and young children. The model compared (1) PCV7, (2) a candidate pneumococcal-NTH (pneumo-NTH) vaccine that has been tested in Europe, (3) a hypothetical pneumo-NTH–Moraxella vaccine, and (4) no vaccination. The analysis was conducted for a US birth cohort of 4.2 million children. Each child was followed via the model from birth to 48 months of age for acute events including otitis media (both AOM and otitis media with effusion [OME]), meningitis, bacteremia, and pneumonia. In addition, the analysis included the lifetime health consequences of death or disability because of meningitis that occurred during the period covered by the model.
We created a microsimulation model of the natural history of otitis media that included 3 stable health states (well, AOM, and OME) and 1 temporary health state, tympanostomy-tube insertion (Fig 1). The model incorporated monthly cycles in which each child could transition to AOM, OME, or well, on the basis of the probabilities specified below. Each month the child's probability of being in each health state was determined by several demographic factors as well as the child's clinical history of AOM and OME. In addition, each child who had experienced 3 or more episodes of AOM during the preceding 6 months of life had a specified probability of having a tympanostomy-tube operation during a given month. To analyze other relevant vaccine-preventable conditions, including meningitis, bacteremia, and pneumonia, we linked the otitis media model to data from an existing model of PCV7 vaccine.17
Table 1 shows the probabilities of otitis-related events and the estimates of vaccine effectiveness used in the model.
Natural History of Otitis Media
The monthly probabilities of making transitions among the stable health states (well, AOM, and OME) were generated on the basis of logistic regression analyses by using 3 primary data sources. The base-case analysis used probabilities generated from the computerized medical charts from January 1996 to December 1999 of Northern California Kaiser Permanente, an integrated healthcare delivery system with an annual birth cohort of ∼30 000. In alternative analyses, we used probabilities from (1) the data sets of the Pittsburgh Child Development/Otitis Media Study, which enrolled 6350 children between 1991 and 1995 and followed 2747 of them with monthly observations through the 36th month of life,18,19 and (2) the computerized medical charts from January 1996 to July 2000 of Harvard Vanguard Medical Associates (HVMA), a large provider group in the greater Boston area whose population had ∼3500 births annually during this period. From each population, we created data sets that indicated the monthly number of visits or diagnoses relevant to otitis media.
We developed 1 set of logistic regression models to predict a child's having AOM in the current month, given the child's otitis media history and other variables, and another set to predict the child's having OME in the current month. For each outcome, we conducted extensive preliminary analyses to identify associated demographic and clinical variables. We formulated and tested many independent variables on the basis of the child's clinical history of otitis media (eg, whether the child had AOM in the preceding month); the number of months with AOM during the preceding 3 or 6 months; whether the child had ever had AOM between birth and the preceding month; and the child's age at the time of the first AOM episode. Lagged disease state variables were created to account for correlation between serial observations of the same child. We created similar independent variables using OME, and using otitis media of either type (AOM or OME). A child predicted to have both AOM and OME in a given month was classified as having AOM. Logistic regression models were developed by using the data set from Kaiser Permanente. Final model specifications were tested and refined for application to the data from Pittsburgh and from HVMA.
Likelihood ratio tests were used to compare alternative models with different predictors. The final logistic regression models were selected on the basis of a combination of statistical measures, such as the area under the receiver operating characteristic curve and the statistical significance of the variables in the model, as well as clinical judgment about the importance of each variable. The final models (Table 2) included age in months, gender, race, whether the child had any history of AOM (yes or no), age at the time of the first AOM episode, the number of months with otitis media in the previous 6 months (or, if the child was <9 months of age, the proportion of months with otitis media in the previous 6 months), the child's health state 1 month earlier (AOM, OME, or well), the child's health state 2 months earlier, and calendar month. The models also included interaction terms for the different possible combinations of health states in the previous 2 months. Age was included in the logistic model as a higher-order polynomial based on goodness of fit.
The hypothetical population for the analysis was assumed to have demographic characteristics similar to those of the general US population in terms of gender, race, and birth-month distributions, as reported in the US census.20–22 The probability of a child having an episode of AOM or OME was calculated at the beginning of each month by using the β coefficients from the logistic regression models.
Children who experienced 3 episodes of AOM in 6 months or 4 episodes in 1 year were considered eligible for tympanostomy-tube insertion.23 In the otitis media model, 18% of such subjects were assumed to actually receive tubes. This estimate was derived by calibrating the model to fit the expected incidence of tube insertion in the total population, based on national statistics.3 Tympanostomy tubes were assumed to be 75% effective in preventing AOM episodes,24and to remain in place for an average of 1 year.25
The effects of 3 different vaccines were evaluated: (1) PCV7; (2) a candidate pneumo-NTH vaccine that has been tested in Europe12; and (3) a hypothetical pneumo-NTH–Moraxella vaccine. We assumed that each vaccine would be administered at 2, 4, 6, and 18 months of age, with maximum effectiveness reached after the second dose. For each vaccinated child, the monthly probability of having AOM was reduced by a defined percentage, which we termed the monthly vaccine-effectiveness factor. This factor was empirically derived by testing a range of percentages and choosing the 1 that produced model results that best matched vaccine-effectiveness estimates from 2 published studies.5,12
Studies of PCV7 have suggested that, as would be predicted mathematically, the vaccine reduces the risk of frequent episodes of AOM to a greater degree than the risk of single episodes.5,15 In the Fireman et al5 study, vaccine use resulted in an 8% reduction in the risk of single otitis visits, a 10% reduction in the risk of 2 otitis visits in 6 months, and a 26% reduction in the risk of 10 otitis visits in 6 months. Our decision model was structured to predict each child's health state (AOM, OME, or well) each month. This structure required that vaccine effectiveness and other factors that affected the probability of otitis media be specified on a monthly basis. Our model's estimates of monthly PCV7 effectiveness were calibrated to reproduce the empirical results reported by Fireman et al5 for both single AOM episodes and recurrent AOM. Estimates of the monthly effectiveness of the pneumo-NTH vaccine against single AOM episodes were calibrated to reproduce the empirical results reported by Prymula et al.12 The relative effectiveness of the pneumo-NTH and pneumo-NTH–Moraxella vaccines against recurrent AOM compared with single episodes of AOM was assumed to be similar to the relative effectiveness of PCV7 against recurrent AOM compared with single episodes of AOM.
Vaccine effectiveness was assumed to wane to 50% of its original level in the third year of life and to 25% in the fourth year of life, based on the consensus of an expert panel assembled to advise on key assumptions in the model.
Costs and Quality-Adjusted Life-Years
Table 3 shows the assumptions we used regarding costs and quality-adjusted life-years (QALYs) in the decision-analysis model.
Costs were estimated by analyzing existing health care use data from the study populations and from other published and unpublished sources. All costs were estimated in 2006 US dollars; costs from previous years were adjusted for inflation by using the medical component of the Consumer Price Index.26
Otitis Media Costs
The cost of each episode of otitis media included direct medical costs and parental time costs. In the base case, the direct medical costs attributable to AOM and OME were estimated by applying estimated unit costs to the utilization identified in the computerized medical charts of HVMA and included the costs of outpatient visits and antimicrobial agents. Episodes of AOM resulted in an average of 1.2 outpatient visits, and episodes of OME resulted in an average of 1.08 outpatient visits. The antimicrobial costs attributable to AOM were derived by taking a weighted average of the average wholesale prices of the antimicrobial agents actually prescribed in the HVMA population.27 Because the actual costs of producing antimicrobial agents are unknown, the average wholesale price was used as a proxy for the actual reimbursement rate. Antimicrobial costs were stratified by age and by whether the AOM episode was recurrent. Recurrent episodes, each defined as 1 of 3 or more episodes in 6 months, were assigned higher costs for antimicrobial agents.
The parental time cost for each office visit was assumed to be 2 hours.28 The value of parental time was based on average wage data from the Bureau of Labor Statistics.29 Travel time and time off from work to care for a child with AOM were not included in the numerator when calculating the cost-effectiveness ratios, because the value of this lost time had already been included in the calculations of QALYs lost because of AOM, as described below. The cost of tympanostomy-tube insertion included the cost of the procedure on the basis of the Medicare Fee Schedule, parental time costs, and costs arising from potential complications.30,31 The only cost for treatment of a complication that we considered was the cost of tympanoplasty,3 generally undertaken because of residual tympanic membrane perforation.
Vaccination costs included the costs of vaccine doses, vaccine administration, and vaccine reactions. In the base case, published public- and private-sector costs were used to calculate a weighted value of $64.84 for the dose cost of PCV7. Costs were estimated at $100 per dose for the combined pneumo-NTH vaccine and at $125 per dose for the combined pneumo-NTH–Moraxella vaccine, based on expert opinion. The range for sensitivity analysis was $5 to $200 per individual vaccine dose. Each vaccine was assumed to require 4 doses. We assumed that the vaccines would be administered during routine well-child visits and that vaccine administration would not involve additional parental time costs. Two percent of all vaccine doses were assumed to result in moderate vaccine reactions.32 Assumed costs associated with such reactions consisted of the cost of a physician visit and 1.5 days of parental time.32
The primary measure of health benefit was the number of QALYs saved. The QALY estimates used in this analysis were based on time-trade-off values (the amount of time persons were willing to give up to avoid a given illness) collected from community members in a study of preferences regarding pneumococcal disease (meningitis, bacteremia, inpatient and outpatient pneumonia, and AOM) and vaccination.33 The values of QALY losses for these conditions took into account both the child's and the parents' pain and suffering and the family's inconvenience from these diseases.33 As shown in Table 3, an AOM episode was estimated to result in the loss of 0.011 QALYs (4 days). An OME episode that was detected clinically was estimated to result in the same QALY loss, whereas an OME episode not detected clinically was not counted in the model. Tympanostomy-tube insertion was estimated to result in the loss of 0.11 QALYs (40 days). The latter estimate was based on the estimated value of 0.011 QALYs for AOM, together with the report from another study that participants were willing to trade-off 10 otitis episodes for 1 tympanostomy-tube insertion.34
The base-case analysis was conducted from the societal perspective, which includes all costs, with both health benefits and costs discounted by 3% per year. The analysis used first-order Monte Carlo simulations of 150 000 trials and was conducted by using TreeAge Pro (TreeAge Software Inc, Williamstown, MA). The number of trials was chosen on the basis of acceptable convergence of model results.
The primary outcomes were AOM and OME episodes and tympanostomy-tube insertions averted, QALYs saved, costs, cost-effectiveness ratios (ie, the cost per QALY saved), and incremental cost-effectiveness ratios (ie, the additional cost per additional QALY saved when a vaccination strategy was compared with the next most effective strategy). The numerator in the cost-effectiveness ratio was the net cost of the vaccination program (ie, the cost of vaccine doses and administration minus the cost of savings from disease averted). The denominator was QALYs saved. As noted earlier, the study from which we derived QALYs asked parents to incorporate their work time lost when answering the time-trade-off questions.33 Thus, when we calculated the cost per QALY saved, we excluded from the numerator the cost of parent work-loss associated with otitis media, because it had already been accounted for via the QALYs in the denominator.
One-way sensitivity analyses were completed for all key model parameters by using ranges described in Table 1. We also conducted analyses in which vaccine effectiveness was assumed to wane more slowly or more rapidly than in the base-case analysis. Two-way sensitivity analyses and threshold analyses were performed for several key variables including vaccine effectiveness, vaccine dose cost, and QALY assumptions.
Table 4 shows the projected health benefits of existing and hypothetical otitis media vaccines for 4.2 million healthy US infants from birth to 48 months of age. During that age period, with no vaccination, the birth cohort of US children was projected to experience 13.7 million episodes of AOM and 350 000 tympanostomy-tube insertions. Of the 13.7 million AOM episodes, PCV7 was estimated to prevent 878 000 episodes (6.4%), the pneumo-NTH vaccine to prevent 3.7 million episodes (27%), and the pneumo-NTH–Moraxella vaccine to prevent 4.2 million episodes (30%). With no vaccination, the cohort would lose 210 000 QALYs because of otitis-related outcomes. PCV7 would save 27 000 (13%) of these QALYs, the pneumo-NTH vaccine would save 71 000 (34%), and the pneumo-NTH–Moraxella vaccine would save 78 000 (37%).
Each of the vaccines was projected to prevent higher proportions of tympanostomy-tube insertions and of lost QALYs than of AOM episodes. Thus, PCV7 was projected to prevent 6.4% of the AOM episodes, 12% of the tympanostomy-tube insertions, and 13% of the QALYs projected to be lost in the absence of vaccination. The corresponding values for the pneumo-NTH vaccine were 27%, 44%, and 34%, respectively, and for the pneumo-NTH–Moraxella vaccine were 31%, 49%, and 37%, respectively.
Costs and Cost-effectiveness
As shown in Table 5, the medical cost savings attributable to averting episodes of otitis media (both AOM and OME) by means of vaccination were estimated at $147 million for PCV7, $574 million for the pneumo-NTH vaccine, and $642 million for the pneumo-NTH–Moraxella vaccine. The 3 vaccines were estimated to save an additional $139 million, $582 million, and $651 million, respectively, in parental time costs. The total savings, because of otitis episodes averted, were $286 million, $1.2 billion, and $1.3 billion, respectively. Table 5 shows that for each of the 3 vaccines, the projected savings from averting otitis media episodes was greater than the $220 million in projected savings from averting all other pneumococcal diseases combined.
We assumed base-case costs per individual vaccine dose of $65 for PCV7, $100 for the pneumo-NTH vaccine, and $125 for the pneumo-NTH–Moraxella vaccine. Compared with no vaccination, vaccination with any of the vaccines would result in additional financial cost. The net costs of the vaccination programs, after accounting for the savings realized from averting otitis media and other pneumococcal diseases, were estimated at $954 million for PCV7, $944 million for the pneumo-NTH vaccine, and $1.3 billion for the pneumo-NTH–Moraxella vaccine.
The base-case analysis was incremental in nature. We thus compared the benefits and costs of the pneumo-NTH vaccine with those of PCV7, and the benefits and costs of the pneumo-NTH–Moraxella vaccine with those of the pneumo-NTH vaccine. Compared with PCV7, the pneumo-NTH vaccine would avert an additional 2.8 million AOM episodes, an additional 113 000 tympanostomy-tube insertions, and the loss of an additional 44 000 QALYs. Thus, at a base-case cost of $100 per dose, the pneumo-NTH vaccine would be preferable to PCV7 because of both greater health benefits and lower net costs.
Compared with the pneumo-NTH vaccine, the pneumo-NTH–Moraxella vaccine would avert an additional 400 000 AOM episodes, an additional 17 000 tympanostomy-tube insertions, and the loss of an additional 7000 QALYs. However, use of the Pnuemo-NTH–Moraxella vaccine would entail an additional $220 million in vaccination program costs and, after subtracting savings from additional disease prevented, would result in an additional net cost of $325 million. Thus, compared with the pneumo-NTH vaccine, the pneumo-NTH–Moraxella vaccine would cost an additional $48 000 per additional QALY saved as shown in Table 6.
In an alternative analysis, we compared each vaccine with no vaccination. Compared with no vaccination, the cost per QALY saved was projected to be $13 000 for the pneumo-NTH vaccine, $16 000 for the pneumo-NTH–Moraxella vaccine, and $35 000 for PCV7.
One-Way Sensitivity Analyses
In 1-way sensitivity analyses comparing the pneumo-NTH vaccine and PCV7, the results were most sensitive to the effectiveness and to the dose cost of the pneumo-NTH vaccine. With vaccine effectiveness held constant at base-case estimates, the pneumo-NTH vaccine would be cost-saving relative to PCV7 if the cost per dose of the pneumo-NTH vaccine were ≤$101. On the other hand, if its cost were $167 per dose, the pneumo-NTH vaccine would cost, relative to PCV7, an additional $25 000 per additional QALY saved, and if its cost were $200 per dose, an additional $38 000 per additional QALY saved.
In other sensitivity analyses, we varied the effectiveness of the pneumo-NTH vaccine with the vaccine cost held constant at $100 per dose. If monthly effectiveness were reduced to 15% in the first 2 years of life and were reduced commensurately in subsequent years of life, the pneumo-NTH vaccine would cost an additional $20 000 per additional QALY saved relative to PCV7. If its monthly effectiveness were reduced to 10% in the first 2 years of life with commensurate reductions in subsequent years of life, it would cost an additional $63 000 per additional QALY saved relative to PCV7.
In 1-way sensitivity analyses, results were found to be somewhat sensitive to the incidence of AOM. When the incidence of AOM was set at the highest end of the plausible range, vaccination with pneumo-NTH became cost saving compared with no vaccination, and compared with the pneumo-NTH vaccine, the pneumo-NTH–Moraxella vaccine had an incremental cost-effectiveness ratio of $8000 per additional QALY saved. The results were not sensitive to variations in other key assumptions, including those concerning the number of QALYs lost per otitis media episode, the cost of an otitis media visit, the cost of tympanostomy-tube insertion, and the probability of vaccine adverse events. When these assumptions were varied over plausible ranges, the pneumo-NTH vaccine remained cost-saving relative to PCV7 in most instances. Where pneumo-NTH was not cost-saving relative to PCV7, its incremental cost-effectiveness (ie, the additional cost per additional QALY saved) ranged from $300 to $7000 per QALY saved.
Two-Way Sensitivity Analyses
Figure 2 shows the results of a 2-way sensitivity analysis in which vaccine effectiveness and dose cost were varied simultaneously. As shown, the pneumo-NTH vaccine was cost-saving compared with PCV7 over a wide range of assumptions (area to the left of the cutoff line between cost and cost-saving [the “$0” line]). Thus, if the monthly effectiveness of the pneumo-NTH vaccine for children 4 to 24 months of age were 20% (instead of 28% as in the base case), it would be cost-saving relative to PCV7 at a cost of $85 per dose. However, with monthly effectiveness remaining at 20% and at a cost of $131 per dose, the cost per QALY saved would be $25 000 greater than that of PCV7. Similarly, if the monthly effectiveness of the pneumo-NTH vaccine were 10%, it would be cost-saving relative to PCV7 at a cost of $70 per dose, and at a cost of $82 per dose the cost per QALY saved would be $25 000 greater than that of PCV7.
Waning of Vaccine Effectiveness
The base-case analysis assumed that vaccine effectiveness waned to 50% of its original level in the third year of life, and to 25% of its original level in the fourth year of life. Two additional waning patterns were examined: (1) no waning; and (2) more rapid waning, defined as a decrease to 25% of the original effectiveness in the third year of life and to 6.25% of the original effectiveness in the fourth year of life. The results were not sensitive to the rate at which vaccine effectiveness waned; irrespective of the rate examined, the pneumo-NTH vaccine remained cost-saving compared with PCV7. These results seemed most likely attributable to the fact that most episodes of otitis media occur within the first 2 years of life.
Analyses Using Alternative Populations
To test the robustness of our results, we conducted alternative analyses using assumptions derived from 2 alternative populations: the Pittsburgh Child Development/Otitis Media Study, and the HVMA population.
In the Pittsburgh analysis, the predicted number of AOM episodes for the US birth cohort under no vaccination was 18 million during the first 36 months of life, which was higher than in the base-case analysis (which predicted 13.7 million AOM episodes during the first 48 months of life). In the Pittsburgh analysis, the pneumo-NTH–Moraxella vaccine was projected to have greater benefits and lower net costs than both PCV7 and pneumo-NTH, making it the preferred strategy. The cost of pneumo-NTH–Moraxella compared with no vaccination was $16 000 per QALY saved.
In the HVMA analysis, the predicted number of AOM episodes for the US birth cohort under no vaccination was 16 million during the first 48 months of life. The pattern of cost-effectiveness results in the HVMA analysis was similar to that in the base-case analysis. Compared with PCV7, pneumo-NTH prevented more otitis media episodes and had lower net costs, making it the preferred strategy. The incremental cost-effectiveness ratios were $13 000 per QALY saved for pneumo-NTH relative to no vaccination, and $49 000 per QALY saved for pneumo-NTH–Moraxella relative to pneumo-NTH.
These alternative analyses suggest that the results are sensitive to the incidence and natural history of otitis media in the population. In a population with a much higher incidence of AOM than in the base case, the pneumo-NTH–Moraxella vaccine would be the preferred strategy.
This study found that candidate combination vaccines against otitis media have the potential to prevent millions of episodes of AOM in the United States each year. The candidate pneumo-NTH vaccine would be cost-effective or even cost-saving relative to the PCV7 vaccine, provided its price were within a range comparable to that for other vaccines introduced in the United States since 2000, which is approximately $30 to $125 per dose.35 At prices of up to $200 per dose, both the pneumo-NTH and pneumo-NTH–Moraxella vaccines would most likely cost $25 000 or less per QALY saved. This cost-effectiveness ratio is within the range of values for other recently introduced vaccination programs, most of which have costs of approximately $10 000 to $50 000 per QALY saved.36–41
Comparisons With Other Studies
We believe the current study to be unique in its focus on the population benefits and the economic viability of new candidate vaccines against otitis media. Other studies have described the cost-effectiveness of PCV7 but have not evaluated new combination vaccines.17,33,42,43
Our analysis suggests that AOM in US children between birth and 48 months of age leads to annual costs of more than $3.8 billion. Other published estimates of the annual costs of otitis media in the United States are more than a decade old and have ranged from $3.5 billion in a 1989 report to more than $5 billion in a 1996 report.44,45 Our estimate was conservative in that it did not include costs for AOM occurring after the 48th month of life. We chose this cutoff because the incidence of AOM decreases after this age and because it is not clear how much the effectiveness of infant vaccination might wane after this age. Relatly, some studies using computerized diagnostic data suggest that the observed incidences of AOM, treatment failure, and relapse have been decreasing over the past several years.46–49 Should this actually be the case and should such a trend continue, the costs of AOM might decrease, and vaccination against otitis media might accordingly be less cost-effective than suggested in the present analysis.
We estimate that the savings from AOM prevented by pneumococcal conjugate vaccines, and the projected savings from AOM prevented by candidate otitis media vaccines, far outweigh the savings from other pneumococcal diseases prevented. Our estimate of the cost per QALY saved by pneumococcal conjugate vaccination, approximately $35 000, was higher than in past studies, in which the estimated costs ranged from $4000 to $6500.33,50 Most of this difference can be accounted for by differences in several key assumptions. The present study's assumptions were based on more recent data that suggest that vaccine effectiveness may wane over time.5 The present study also gave vaccination credit for preventing disease for the first 4 years of life, whereas past studies assumed that it would be efficacious for the first 5 years of life. In addition, the estimate of vaccine effectiveness against recurrent AOM (≥3 episodes in 6 months) was calibrated to empirical data from the study by Fireman et al,5 whose results were not available when the previous studies33,50 were conducted. Finally, the current study used a more conservative estimate of the QALY losses associated with recurrent AOM than did the previous studies, to avoid double-counting the QALY losses we assigned to tympanostomy-tube insertion.
In this analysis we used assumptions about the effectiveness of the pneumo-NTH vaccine based on a recent randomized, controlled trial.12 That trial's definition of AOM was based on the presence of a middle ear effusion alone, so that some of the counted episodes might have been OME rather than AOM. In addition, no active surveillance for otitis media was conducted; instead, children were seen by doctors when symptomatic, according to standard local clinical practice, so that the frequency of occurrence of AOM may have been under-ascertained. The lack of precision in diagnosis and the limited follow-up in the trial may raise questions about the true effectiveness of the vaccine being studied. The results of the current analysis should be interpreted with this caveat in mind.
This study was focused on 2 candidate vaccines against otitis media that, on the basis of recent research, seem most likely to be developed.11,12 We did not conduct extensive analyses of other hypothetical vaccines, namely, vaccines against NTH alone, Moraxella alone, respiratory syncytial virus, and influenza virus. In view of the bacterial and viral etiologies of AOM, each of these vaccines would likely prevent smaller numbers of AOM episodes than the vaccines we evaluated in our final analysis. Thus, they would be far less viable candidates for effectively and cost-effectively preventing AOM.
An alternative to the routine vaccination strategy evaluated here would be to selectively vaccinate only those children who have already experienced recurrent AOM. The target population would likely be relatively small: during the period from 1988 to 1994, only 8.2% of infants in the first year of life were reported to have experienced recurrent AOM, defined as ≥3 episodes,51 and after 2 years of age the incidence of AOM begins to decline.52 Be that as it may, studies suggest that, in previously unvaccinated young children with a history of recurrent AOM, PCV7 is not effective in reducing additional recurrences of AOM.53,54 In addition, our preliminary analyses suggested that this would be a less effective and less cost-effective approach than routine vaccination of all infants.
This analysis offers an assessment of the benefits and cost-effectiveness of candidate otitis media vaccines at the present time. However, future shifts in overall pathogen distribution could lead to different findings. We did not attempt to evaluate the cost-effectiveness of a vaccine that included pneumococccal serotype 19A, which is an emerging otopathogen that is resistant to multiple antimicrobial agents.55 Studies of the population-based dynamics and economics of antimicrobial resistance, and of its possible prevention via vaccination, are complex, and are an important area for future research but were beyond the scope of the present study. Our analysis did not give any of the vaccines credit for preventing disease in unvaccinated persons via herd immunity. This might have led to conservative estimates of the vaccines' cost-effectiveness but probably would not have affected the comparative results for the 3 vaccines. In addition, our analysis is based on assumptions that may not be applicable universally to all populations. However, our alternative analyses using populations from the Pittsburgh and Harvard cohorts suggest that our findings are relatively robust.
This analysis suggests that new candidate vaccines against otitis media have the potential to prevent millions of disease episodes in the United States annually. If priced comparably with other childhood vaccines recently introduced in this country, these new vaccines could achieve, in relation to the pneumococcal conjugate vaccine, comparable cost-effectiveness or even cost savings.
We are grateful to our expert panelists, Steve Black, MD, M.Elizabeth Halloran, MD, MPH, DSc, Jerome Klein, MD, Marc Lipsitch, MD, DPhil, and Henry Shinefield, MD. We thank Howard Rockette, PhD, for advice at the inception of the study.
- Accepted September 26, 2008.
- Address correspondence to Tracy A. Lieu, MD, MPH, Harvard Medical School and Harvard Pilgrim Health Care, Center for Child Health Care Studies, Department of Ambulatory Care and Prevention, 133 Brookline Ave, 6th Floor, Boston, MA 02215. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
New vaccines with enhanced effectiveness against otitis media are under development, but very scant information exists about how much benefit they would provide or whether they would be economically viable if recommended in large populations.
What This Study Adds
This study shows that new candidate vaccines against otitis media could prevent millions of disease episodes in the United States annually and would have cost-effectiveness comparable with the current pneumococcal conjugate vaccine if priced similar to other recently introduced vaccines.
- ↵Freid VM, Makuc DM, Rooks RN. Ambulatory health care visits by children: principal diagnosis and place of visit. Vital Health Stat 13.1998;(137):1– 23
- ↵Owings MF, Kozak LJ. Ambulatory and inpatient procedures in the United States, 1996. Vital Health Stat 13.1998;(139):1– 119
- ↵Straetemans M, Sanders EA, Veenhoven RH, Schilder AG, Damoiseaux RA, Zielhuis GA. Pneumococcal vaccines for preventing otitis media. Cochrane Database Syst Rev.2004;(1):CD001480
- ↵Bluestone CD, Klein JO. Otitis Media in Infants and Children. Philadelphia, PA: W.B. Saunders Company; 2001
- ↵Prymula R, Peeters P, Chrobok V, et al. Pneumococcal capsular polysaccharides conjugated to protein D for prevention of acute otitis media caused by both Streptococcus pneumoniae and non-typable Haemophilus influenzae: a randomised double-blind efficacy study. Lancet.2006;367 (9512):740– 748
- ↵Wischnack LL, Jacobson RM, Poland GA, Jacobsen SJ, Harrison JM, Murtaugh PA. The surprisingly high acceptability of low-efficacy vaccines for otitis media: a survey of parents using hypothetical scenarios. Pediatrics.1995;95 (3):350– 354
- ↵Poehling KA, Szilagyi PG, Grijalva CG, et al. Reduction of frequent otitis media and pressure-equalizing tube insertions in children after introduction of pneumococcal conjugate vaccine [published correction appears in Pediatrics. 2007;119(6):1270]. Pediatrics.2007;119 (4):707– 715
- ↵Paradise JL, Rockette HE, Colborn DK, et al. Otitis media in 2253 Pittsburgh-area infants: prevalence and risk factors during the first two years of life. Pediatrics.1997;99 (3):318– 333
- ↵Paradise JL, Dollaghan CA, Campbell TF, et al. Language, speech sound production, and cognition in three-year-old children in relation to otitis media in their first three years of life. Pediatrics.2000;105 (5):1119– 1130
- ↵US Census Bureau; Census 2000 summary File 1, generated by Megan O'Brien; using American FactFinder; http://factfinder.census.gov. Accessed July 11, 2005
- US Census Bureau; Current Population Survey, annual social and economic supplement 2000; generated by Megan O'Brien; using American FactFinder; http://factfinder.census.gov. Accessed July 7, 2005
- ↵Centers for Disease Control and Prevention. Live births by day of birth. Available at: www.cdc.gov/nchs/data/statab/t001x16.pdf. Accessed June 23, 2005
- ↵Ah-Tye C, Paradise JL, Colborn DK. Otorrhea in young children after tympanostomy-tube placement for persistent middle-ear effusion: prevalence, incidence, and duration. Pediatrics.2001;107 (6):1251– 1258
- ↵Bureau of Labor Statistics. Consumer price index for medical care services. Available at: http://data.bls.gov/cgi-bin/surveymost. Accessed April 9, 2008
- ↵Drug Topics Red Book: Pharmacy's Fundamental Reference. Montvale, NJ: Thomson PDR; 2005
- ↵National Compensation Survey: Occupational Wages in the United States, June 2006. Summary 07–03. Washington, DC: US Department of Labor Statistics; 2007
- ↵Prosser LA, Ray GT, O'Brien M, Kleinman K, Santoli J, Lieu TA. Preferences and willingness to pay for health states prevented by pneumococcal conjugate vaccine. Pediatrics.2004;113 (2):283– 290
- ↵Centers for Disease Control and Prevention. CDC vaccine price list. Available at: www.cdc.gov/vaccines/programs/vfc/cdc-vac-price-list.htm. Accessed March 20, 2008
- ↵Lee GM, Lebaron C, Murphy TV, Lett S, Schauer S, Lieu TA. Pertussis in adolescents and adults: should we vaccinate? Pediatrics.2005;115 (6):1675– 1684
- Goldie SJ, Kohli M, Grima D, et al. Projected clinical benefits and cost-effectiveness of a human papillomavirus 16/18 vaccine. J Natl Cancer Inst.2004;96 (8):604– 615
- ↵Rein DB, Hicks KA, Wirth KE, 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
- ↵Gates GA. Cost-effectiveness considerations in otitis media treatment. Otolaryngol Head Neck Surg.1996;114 (4):525– 530
- ↵Finkelstein JA, Stille C, Nordin J, et al. Reduction in antibiotic use among US children, 1996–2000. Pediatrics.2003;112 (3 pt 1):620– 627
- Sox CM, Finkelstein JA, Ruihua Y, Kleinman K, Lieu TA. Trends in otitis media treatment failure and relapse. Pediatrics.2008;121 (4):674– 679
- ↵Zhou F, Shefer A, Kong Y, Nuorti JP. Trends in acute otitis media-related health care utilization by privately insured young children in the United States, 1997–2004. Pediatrics.2008;121 (2):253– 260
- ↵Lieu TA, McGuire TG, Hinman AR. Overcoming economic barriers to the optimal use of vaccines. Health Aff (Millwood).2005;24 (3):666– 679
- ↵Auinger P, Lanphear BP, Kalkwarf HJ, Mansour ME. Trends in otitis media among children in the United States. Pediatrics.2003;112 (3 pt 1):514– 520
- ↵Teele DW, Klein JO, Rosner B. Epidemiology of otitis media during the first seven years of life in children in greater Boston: a prospective, cohort study. J Infect Dis.1989;160 (1):83– 94
- ↵American Academy of Pediatrics. 2004/05 state reports on Medicaid, Medicare, and commercial payments by CPT code. Available at: www.aap.org/research/medmedcomm0405state.htm. Accessed September 13, 2007
- Copyright © 2009 by the American Academy of Pediatrics