Objective. To measure parents’ and other adults’ values for preventing disease associated with pneumococcal infection and to evaluate how including these values changes the economic appraisal of pneumococcal conjugate vaccine.
Methods. Data on preferences and willingness to pay to reduce risk of illness were collected for 6 illnesses that are preventable by pneumococcal conjugate vaccine (simple otitis media, complex otitis media, moderate pneumonia, severe pneumonia, bacteremia, and meningitis) and 1 vaccine-related adverse event (fever and fussiness after vaccine). Interviews were conducted with 2 groups of respondents: 1) parents of children who had experienced 1 or more of the outcomes described in the survey (n = 101) and 2) a US community sample (n = 109). The 30-minute telephone interview used modified time trade-off questions and willingness-to-pay questions. Values from the interview were incorporated in an existing decision-analytic model that simulated the cost-effectiveness and cost-benefit of pneumococcal conjugate vaccine in a hypothetical cohort of newborns.
Results. Among parents, the median amount of time that respondents said that they would be willing to trade to avoid diseases ranged from 0 days for otitis media to 1 year for severe pneumonia and 2 years for meningitis. Among the US community sample, the median amounts of time traded were higher, ranging from 7 days for otitis media to 3 years for meningitis. Median willingness-to-pay amounts varied from $100 to prevent 1 episode of otitis media and $500 to reduce the risk of meningitis from 21 in 100 000 to 6 in 100 000 and were similar between parents and community members. Incorporating time trade-off amounts into the existing economic model for pneumococcal conjugate vaccine resulted in cost-effectiveness ratios <$10 000 per quality-adjusted life year at a vaccine cost of $58 per dose.
Conclusions. Both parents and community members assign relatively high values to preventing meningitis, pneumonia, and complex otitis media. When the value of preventing pneumococcal diseases is incorporated into economic analyses, pneumococcal conjugate vaccine has a cost-effectiveness ratio in the range of other widely used health interventions.
Pneumococcal conjugate vaccine, introduced in 1999 at a private-sector price of $232 for the 4-dose series, is the highest priced vaccine ever recommended for routine use in US infants. The original cost-effectiveness analysis of this vaccine measured health benefits in terms of deaths prevented by averting meningitis and estimated that the vaccine would cost $110 000 per life-year saved.1 This cost-effectiveness ratio made the vaccination program seem worse than other health interventions. However, the denominator of life-years saved did not give the vaccine credit for preventing other morbidity, including otitis media, pneumonia, and nonfatal meningitis.
Scant data exist on how to place quantitative values on the benefits of disease prevention by pneumococcal conjugate vaccine. The 2 most widely used approaches to valuing health benefits are preference assessment and willingness-to-pay elicitation. A few studies have attempted to estimate preferences for otitis media, but these have important limitations, including small and atypical samples.2–4 One study elicited willingness-to-pay amounts for otitis media,5 but comparable information for pneumonia and meningitis is lacking.
The problem of how to give vaccines credit for preventing morbidity in addition to mortality is likely to become increasingly important as the country considers new vaccination programs. Financial support for vaccination against influenza in young children, hepatitis A in high-risk areas, and pertussis in adolescents and adults would largely prevent morbidity rather than death from these diseases. In addition, there is debate about whether it is better to use the values of parents, who have children and may have directly experienced the diseases of interest, or those of a community-based sample, which may better reflect societal preferences.6 Thus, testing ways of giving credit for preventing morbidity and evaluating how this may change the appraisal of a vaccine’s cost-effectiveness hold lessons that are generalizable to future vaccination programs. This study’s objectives were to 1) measure preferences and willingness-to-pay amounts for selected diseases that are prevented by pneumococcal conjugate vaccine, 2) compare these values between parents and other adult community members, and 3) evaluate how adding these values to an existing economic model may change the results of the analysis.
Two groups of respondents were asked to participate in the study. The first group consisted of parents of children who had received diagnoses of conditions that are prevented by pneumococcal conjugate vaccine (n = 101). Children who had experienced otitis media (n = 97), pneumonia (n = 46), or meningitis (n = 2) within the last 3 years were identified through medical records of Harvard Vanguard Medical Associates, a large group practice in New England. Membership data were then used to identify contact information for the parents of these children. The second group of respondents was a national community sample identified through random-digit dialing and matched to US population characteristics (n = 109). Procedures were approved by the Harvard Pilgrim Health Care Human Subjects Committee and the Centers for Disease Control and Prevention’s Institutional Review Board.
Prospective participants from the parent sample were initially contacted with a letter explaining the purpose and requirements of the study, which included an “opt-out” postcard. If they did not return the opt-out postcard, then parents were contacted by telephone and invited to participate in the survey. Prospective participants for the community sample were identified using random-digit dialing. In the initial telephone call for both study samples, the interviewer explained the study’s purpose and how the respondent had been selected and made clear that participation was voluntary and not participating would not affect the family’s medical care in any way. When the respondent agreed to participate, an interview time was scheduled for at least 1 week later to allow time for respondents to receive survey materials by mail. Respondents were interviewed between May and September 2001.
Participation in the study consisted of completion of a 30-minute telephone survey. Before the telephone interview, respondents were mailed materials to which to refer during the interview. These included a booklet with written descriptions of the outcomes that can be prevented by pneumococcal conjugate vaccine. Visual aids adapted from Corso et al7 were also included to improve respondents’ understanding of risk reductions in the willingness-to-pay questions.
The 30-minute closed-ended interview included time trade-off and willingness-to-pay questions for 6 hypothetical health states and 1 vaccine adverse event. The complete list of health states and their descriptions are provided in Appendix 1. Time trade-off questions were asked as the amount of the respondent’s life that the respondent would be willing to give up to prevent a certain health state in their child or a hypothetical child (Fig 1). The time trade-off amounts elicited for this study are preference-based measures but are not utilities, which are typically scaled between 1.0, representing perfect health, and 0.0, representing a health state judged equivalent to being dead. Instead, the time trade-off amounts are lower for less severe health states (respondents willing to give up less time) and higher for more severe health states. Respondents were also asked for the maximum amount of money that they would be willing to pay for a specific risk reduction that reflected vaccine effectiveness for each condition (Fig 1). Willingness to pay was measured by asking the respondent whether he or she was willing to pay an opening bid. Depending on the answer to the first bid, the respondent was asked either a higher or lower second bid (dichotomous-choice double-bounded question), which was then followed by a final open-ended question about their maximum willingness to pay. Respondents were randomized to 4 different groups of bids to minimize anchoring bias. Data on sociodemographic characteristics, as well as whether the respondent’s child had experienced any of the described conditions, were also collected.
Analysis of Survey Data
Summary statistics for time trade-off and willingness-to-pay amounts, including medians, standard deviations, standard errors, and 25th and 75th percentiles, were reported separately for patient and community respondents. Differences between the parent and community samples for time trade-off and willingness-to-pay responses were evaluated using the Kolmogorov-Smirnov test.8 Medians were reported instead of means because the distributions were skewed toward 0, especially for the less severe health states.
For the time trade-off questions, we also calculated an alternative set of time trade-off amounts using a different assumption for when time was traded off from the respondents’ lives. The undiscounted responses assume that the respondent was trading off life today. Because it was clear from respondents’ comments that many considered this time to be traded off from the end of their life, we calculated an alternative set of time trade-off amounts (discounted response) that assumes that all time was traded off from the end of the respondent’s life and discounted over the difference between the participant’s age and life expectancy using a rate of time preference of 3% per year.
The effect on overall time trade-off and willingness-to-pay amounts of respondent characteristics were evaluated using estimated random effects from the generalized linear mixed models version of Poisson regression.9 Dependent variables included in the model were age, sex, race/ethnicity (white: yes/no), education (college or more: yes/no), marital status (married: yes/no), income (>3 times poverty level: yes/no), and health status (good or better: yes/no, whether the respondent has young children, and whether a child has experienced any of the outcomes described in the survey). For willingness-to-pay amounts, we also evaluated the effect of the opening bid.
Incorporation of Survey Data in Economic Model
The effect of including values for changes in morbidity effects was evaluated by incorporating time trade-off and willingness-to-pay amounts into an existing economic model for pneumococcal conjugate vaccine.1 Our time trade-off method is different from that often used for chronic health states, in which respondents choose between years of life with the condition and years without it, because our study evaluated temporary health states. Respondents were asked for the amount of time that they were willing to give up to prevent a specific temporary health state. These values were then included directly in the model. For example, if a respondent was willing to trade off 7 days to prevent simple otitis media, then this represented a 1-time reduction of 7 days in quality of life (or a 1-time loss of 0.02 quality-adjusted life years [QALYs]). Because the valuation explicitly instructed respondents to include reduction in both the child and the parent’s quality of life, it is possible for the value to exceed the length of the temporary health state.
This model used probabilities from published and unpublished data as well as appropriate costs to estimate the projected outcomes of routine pneumococcal conjugate vaccination versus no vaccination of infants. The types of costs included were 1) direct medical (estimated on the basis of national and local data) and 2) direct nonmedical (eg, out-of-pocket expenses for parking or transportation to doctor visits). Other nonmedical “costs,” including pain, anxiety, inconvenience, and time costs, were represented in the model by including either time trade-off or willingness-to-pay values from the survey. More detailed information on modeling assumptions are available at www.dacp.org/pneumopaper.html.
Base-case cost-effectiveness and cost-benefit results were calculated using median time trade-off and willingness-to-pay amounts. Sensitivity analyses used alternative assumptions for median time trade-off amounts (25th and 75th percentiles), vaccine costs (0.5–2.0 times the base-case assumption for actual vaccine costs of $232), medical costs (0.5–2.0 times the base-case assumptions), and disease incidence (0.5–2.0 times the base-case assumptions). Additional sensitivity analyses evaluated the effects of using the discounted, rather than undiscounted, time trade-off amounts on cost-effectiveness results.
For the parent sample, 391 letters were mailed to parents to invite them to participate in the survey. Of those, 24% could not be contacted to schedule an interview. Of the 298 remaining potential participants, 36% agreed to participate and completed an interview, 9% scheduled an interview but refused to participate at the time of the interview, 8% scheduled an interview but could not be contacted to complete the interview, and 47% refused to participate by either returning an opt-out card or by declining to schedule an appointment at the time of the follow-up telephone call. The parent sample (mostly mothers) included 96% whose children had experienced otitis media, 46% whose children experienced pneumonia, and 2% whose children had experienced meningitis. The parent sample tended to have higher levels of educational achievement and incomes than the community sample (Table 1). There were also more women and married respondents in the parent sample compared with the community sample.
For the national community sample, 1069 people were contacted and 20% agreed to participate in the survey. Of those who agreed to participate, 53% completed an interview (for an overall response rate of 10%), 26% initially agreed to be interviewed but were unable to be contacted to complete the interview, 19% refused, and 1% were excluded at the time of the interview because they could not remember agreeing to participate. Members of the community sample were similar to the US population sample in terms of age, race, and income but contained a higher proportion of women and were somewhat less likely to be married when compared with the general US population.10,11
Six (6%) observations were excluded from the parent sample and 3 (3%) observations were excluded from the community sample on the basis of interviewer assessment of the survey as invalid because the respondent was unable to understand the questions. In similar types of studies, 5% to 10% of respondents demonstrated difficulty answering similar preference questions.12,13
Time Trade-off Amounts
The median time trade-off amounts (days or years that respondents were willing to give up to avoid pneumococcal-related health states) ranged from 0 days (otitis media) to 2 years (meningitis) for the parent sample (Table 2). More than half of all parents were not willing to trade off any time to avoid simple otitis or fever and fussiness, and the median was only 1 day for moderate pneumonia (Table 2). Time trade-off amounts generally increased with increasing severity of the health state. Time trade-off amounts were higher for community respondents than for parents for most outcomes but only significantly different for simple otitis media, moderate pneumonia, and fever and fussiness after vaccination. The median amount of time community respondents were willing to trade off ranged from 7 days for otitis media to 3 years for meningitis (Table 2).
In an alternative analysis of time trade-off amounts, we assumed that respondents considered time traded off to occur at the end of their life and discounted these amounts appropriately (Table 2). For both parent and community samples, these discounted time trade-off amounts were approximately half of the undiscounted time trade-off amounts.
Willingness to Pay
An analysis of ranks indicated that willingness-to-pay responses were not significantly different for the 2 samples except for 1 health state, fever and fussiness after vaccination (Table 3). For example, both parents and community respondents were willing to pay a median of $400 to prevent 1 episode of severe pneumonia and $500 to reduce the risk of meningitis (Table 3). Median willingness to pay for a simplified description of the 4-dose series of pneumococcal conjugate vaccine was $250 for parents and $300 for community respondents.
Having a household income >3 times the poverty level was associated with higher willingness-to-pay amounts, but no other respondent characteristics were found to affect responses. The coefficients on the income and education variables were also significant for time trade-off amounts, showing higher responses for respondents with higher incomes and lower responses for those with higher education levels (data not shown).
Inclusion of Preferences/Willingness to Pay in Existing Economic Models
Incorporating median time trade-off amounts into an existing economic model resulted in a cost-effectiveness ratio of <$5000/QALY for base-case assumptions using both parent and community values (Table 4). Three-way sensitivity analyses varying time trade-off amounts, disease incidence rates, and medical costs all resulted in cost-effectiveness ratios <$100 000/QALY except for when the lower 25th percentile time trade-off amounts from the parent sample were used. Cost-effectiveness ratios were sensitive to vaccine costs and changed almost proportionately with vaccine cost (data not shown). For example, varying the vaccine cost from the base-case assumption of $58/dose to $100/dose caused the cost-effectiveness ratios using community preferences to change from $2400/QALY to $4600/QALY. Using discounted time trade-off amounts resulted in slightly higher but still favorable cost-effectiveness ratios approximately 60% to 80% higher than the base-case cost-effectiveness results using median time trade-off amounts.
Although a key benefit of the vaccine is the reduction in morbidity and mortality associated with meningitis, it was the inclusion of the change in health-related quality of life associated with complex otitis media, which has a higher incidence than meningitis, that caused the greatest change in the cost-effectiveness ratio (Table 5). Giving credit for the quality-of-life benefits of preventing meningitis, bacteremia, and pneumonia combined resulted in a reduction in the $/QALY saved to $23 000, whereas giving credit for preventing otitis media alone resulted in reduction of the $/QALY saved to $4500.
Cost-benefit analyses using median willingness-to-pay amounts resulted in positive net benefits (signifying that implementing a vaccination program would be preferable to having no vaccination program) for both parent and community responses. Three-way sensitivity analyses varying willingness-to-pay amounts, disease incidence rates, and medical costs all resulted in positive net benefits. Using higher vaccine costs still resulted in net benefits unless 25th percentile willingness-to-pay values were used and vaccine costs were >4 times the costs assumed in the base-case analysis (data not shown).
This study was initiated amid concern about the high price of pneumococcal conjugate vaccine relative to previously recommended childhood vaccines. These findings indicate that despite its relatively high cost, routine pneumococcal conjugate vaccination of children is economically favorable when community or parent values for health benefits of the vaccine are considered in addition to traditional direct and opportunity costs typically included in such models. Most previously recommended vaccines have lower costs and result in medical cost savings. However, most health care interventions do not achieve cost savings but are termed “cost-effective” because they provide benefits at a cost deemed acceptable by policy makers.14 A review of cost-utility analyses for clinical preventive services reported a median cost-effectiveness ratio of $14 000/QALY (1997 dollars).15 The current study suggests that pneumococcal conjugate vaccine is relatively cost-effective from a societal perspective, even though it requires some investment per gain in life-years or QALYs. The finding that vaccination is economically desirable is robust in that we reached this conclusion both in the primary analysis that used cost-effectiveness as its metric and using a widely accepted alternative method, cost-benefit analysis.
Both parents and community members associate a significant loss in quality of life with conditions prevented by pneumococcal conjugate vaccine, except for relatively mild health states of simple otitis media and “moderate” pneumonia (not requiring hospitalization). Compared with parents, community members placed higher values on preventing pneumococcal-associated outcomes according to time trade-off responses. It is often the case that respondents who are more familiar with a health condition will rate the condition as less severe.6 Given these differences by respondent type, the issues of whose responses are used (patient versus community) should be considered when incorporating quality-of-life adjustments into economic analyses. The US Public Health Service Panel on Cost-effectiveness in Health and Medicine recommends the use of community ratings for health states, but community-based ratings can suffer from problems: response rates for these surveys tend to be low (as it is for our survey), and the health scenarios are even more hypothetical than for patients.6 However, if resources are available only to measure in 1 type of respondent, then community values should be used to ensure comparability across studies. In the present study, both sets of values yielded favorable cost-effectiveness results, but vaccination seemed slightly better using community members’ values than parents’ values.
The amount of time traded off and willingness to pay to prevent some pneumococcal-associated health states reported by both parents and community respondents could be considered to be high for some health states. For example, both parents and community members were willing to trade off 1 year to prevent severe pneumonia in a 1-year-old child, which in our survey was described as requiring a short hospital stay but with no long-term sequelae. First, it is important to note that our time trade-off and willingness-to-pay questions explicitly instructed parents to include their time spent caring for a sick child and stress associated with each described health state that would cause values to be higher. Also, there is some evidence that parents are willing to pay more to prevent a health state in a child versus the identical situation in an adult.16 Because our time trade-off question differs from that commonly used for chronic health states and loss of quality of life for both parent and child are explicitly included, the time trade-off amounts presented here are not directly comparable to utility values from generic utility instruments for measuring reductions in quality of life for chronic health states, such as the Health Utilities Index17 or the EQ-5D.18 A study by Bennett et al19 reported parents’ utilities for bacterial meningitis as being less severe than in our study, but these used the standard-gamble method and did not explicitly include parents’ loss in quality of life.
We asked respondents to value the risk reductions associated with the vaccine because this is the benefit that they would be purchasing by paying for the vaccine. Despite our use of visual aids to improve respondents’ understanding of the risk reductions involved in the willingness-to-pay questions, the results reveal a lack of sensitivity to scope. A respondent is considered not sensitive to scope when his or her increase in willingness to pay is not proportional to the amount of the increased risk. For example, a person should be willing to pay approximately double the amount if the risk is also doubled. In our study, however, respondents were willing to pay more for the reduction in meningitis risk alone ($500 in both samples) than for the vaccine description that included all of the risk reductions including meningitis ($250 in the parent sample and $300 in the community sample). This lack of sensitivity to scope has been previously reported and underscores the limitations of asking questions that include small probabilities that can be difficult for respondents to answer.20 For those who are interested in including these values in an economic analysis, a better approach may be to ask questions about avoiding an episode of illness, rather than the risk of illness, and then including the probabilities in the decision model. This method, however, may result in more conservative willingness-to-pay responses because willingness to pay to avoid a risk will not be captured if respondents are on average risk averse.21 Excluding risk from our study questions would likely have resulted in smaller willingness-to-pay amounts that would in turn have made the results of the cost-benefit analysis less favorable.
Preference measurement for childhood health states presents a number of challenges. First, parents must serve as proxy respondents because children younger than 12 years are unable to understand the relatively abstract questions required.22 A general problem with proxy respondents is that they tend to rate an illness as worse than the patient themselves.23 In this study, such a bias would have resulted in more favorable economic analyses of the vaccine than if we had been able to use the patients’ (in this case children’s) own values. One study that compared parents’ and adolescents’ ratings for neonatal outcomes showed the opposite: adolescents rated the health states as worse than their parents did.24 If this were the case in our study, then the result would have been less favorable economic analyses. Other studies have compared values from parents and health professionals and found differences in valuations between the groups.25–27 Our recommendation is that parents are the more appropriate proxy respondent as they are closer to the patient and that parents’ loss in quality of life should also be measured and included.
Second, it is difficult for parents to exclude values for benefits to themselves as caregivers when they are placing values on children’s health states. In our study, we explicitly asked parents to include caregiver time and parental loss in quality of life when valuing the health states. Thus, the preferences that we collected are not directly comparable with those of other preference studies, which do not typically include these components. We believe that benefits to parents are an important component of valuing childhood health states because parents make many, if not all, of the health care decisions, especially for young children, and it is likely that they include their own values in these decisions. During pretesting, respondents expressed difficulty in separating benefits to themselves from the valuation of the child’s health state.
Third, parental “guilt” may inflate the responses given: some respondents made statements indicating that they would be considered to be “bad” parents if they were not willing to trade off any time from their life or pay to prevent a health state, even simple otitis media, in their child. A number of respondents said that they would be willing to trade all of the rest of their own life or pay all that they had to prevent relatively mild health states. This could result in unrealistically high values or could simply represent that parents value the children’s health more than their own. Our study findings show, however, that most parents are not willing to trade off more time or pay more than community respondents (who were less likely to have children). More research needs to be undertaken to refine methods for measuring preferences for children’s health states.
The results of this study will assist in evaluations and decision making about the use of pneumococcal conjugate vaccine and other vaccines that prevent meningitis, pneumonia, and otitis media. These findings demonstrate that including the value of preventing morbidity may dramatically improve the cost-effectiveness ratio of a vaccine compared with including only mortality. When only mortality was included, this vaccine’s cost-effectiveness ratio of $110 000 per life-year saved (at the manufacturer’s list price of $58 per dose) was higher than many accepted preventive health interventions.1 Once the value of preventing morbidity from pneumococcal disease was included in the analysis, the cost-effectiveness and cost-benefit results for this vaccine were favorable under a variety of assumptions.
Financial support for this study was provided by the National Immunization Program, Centers for Disease Control and Prevention, via a contract with the Association of Community Health Plans.
We thank Phaedra Corso and Kimberly Thompson for helpful comments on the survey and visual aids. We also express our thanks to the contributors to the original economic analysis of pneumococcal conjugate vaccine: Steven Black, Jay Butler, Jerome Klein, Robert Breiman, Mark Miller, and Henry Shinefield.
- Received January 29, 2003.
- Accepted May 14, 2003.
- Reprint requests to (L.A.P.) Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, 133 Brookline Ave, 6th Fl, Boston, MA 02215. E-mail:
- ↵Lieu TA, Ray GT. Summary of updates and revisions to: “Projected cost-effectiveness of pneumococcal conjugate vaccination of healthy infants and young children” (Lieu TA, Ray GT, Black SB, et al. JAMA2000;283 :1460– 1468). Presentation to CDC Advisory Review Panel, Atlanta, GA, May 29, 2000. See also www.dacp.org/pneumopaper.html
- ↵Sorum PC. Measuring patient preferences by willingness to pay to avoid: the case of acute otitis media. Med Decis Making.1999;19 :27– 37
- ↵Gold MR, Siegel JE, Russell, et al. Cost-Effectiveness in Health and Medicine. New York, NY: Oxford University Press; 1996
- ↵Sheskin DJ. Handbook of Parametric and Nonparametric Statistical Procedures. 2nd ed. New York, NY: Chapman and Hall; 2000:319–329
- ↵Breslow NE, Clayton DG. Approximate inference in generalized linear mixed models. J Am Stat Assoc.1993;88 :25
- ↵1998 USA Statistics in Brief—Law, Education, Communications, Transportation, Housing. Available at: www.census.gov/statab/www/part2. Accessed February 8, 2000
- ↵USA Statistics in Brief. Available at: www.census.gov/statab/www/part1-part4. Accessed April 2, 2000
- ↵Salkeld G, Cameron ID, Cumming RG, et al. Quality of life related to fear of falling and hip fracture in older women: a time trade off study. BMJ.2000;320 :341– 346
- ↵Johannesson M, Weinstein MC. Designing and conducting cost-benefit analyses. In: Spilker B, ed. Quality of Life and Pharmacoeconomics in Clinical Trials. 2nd ed. Philadelphia, PA: Lippincott-Raven Publishers; 1996:1085–1092
- ↵Drummond MF, O’Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 2nd ed. New York, NY: Oxford University Press; 1999:219–221
- ↵Juniper EF, Guyatt GH, Feeny DH, Griffith LE, Ferrie PJ. Minimum skills required by children to complete health-related quality of life instruments for asthma: comparison of measurement properties. Eur Respir J.1997;10 :2285– 2294
- ↵Saigal S, Rosenbaum P, Hoult L, et al. Conceptual and methodological issues in assessing health-related quality of life in children and adolescents: illustration from studies of extremely low birthweight survivors. In: Drotar D, ed. Measuring Health-Related Quality of Life in Children and Adolescents: Implications for Research and Practice. Mahwah, NJ: Lawrence Erlbaum Associates; 1998:151–169
- Kramer MS, Etezadi-Amoli J, Ciampi A, et al. Parents’ versus physicians’ values for clinical outcomes in young febrile children. Pediatrics.1994;93 :697– 702
- Copyright © 2004 by the American Academy of Pediatrics