Recently, a 26-valent GAS vaccine candidate was developed that may provide protection against pharyngitis, invasive disease and rheumatic fever [8,9]. Such a vaccine may reduce the burden of GAS disease by up to 85% in the U.S [4,10,11]. However, parental preferences regarding the prevention of disease vs. the risk for minor vaccine adverse events should be explicitly considered before recommending widespread use of a GAS vaccine. This is particularly relevant as concerns about vaccine safety have become prominent, and the number of parents refusing to vaccinate their children continues to grow as perceptions about the risks of vaccination may outweigh perceived benefits [12-15].

Estimating preferences for childhood vaccination programs has been challenging for several reasons. First, parents often serve as proxy respondents for young children, raising concerns about how to distinguish the child's well-being from the parent's altruism [16-18]. However, since parents are responsible for decision-making about vaccinating their own children, and young children are often unable to provide quantitative assessments of their preferences, the choice of parents as a proxy may be appropriate. Second, while many of the older childhood vaccines have been focused on preventing chronic disability and death (e.g. neurologic disability and death after Haemophilus influenzae type b infection, paralysis and death after polio infection), newer vaccines may target health conditions that are temporary in duration (e.g. otitis media and bacteremia in pneumococcal infection) [19,20]. Health states that are described as short-term may be valued differently than the same health states as chronic conditions [21-23]. Third, off-the-shelf utilities do not exist for many of the short-term health conditions, necessitating a formal assessment of preferences when implementing a new vaccination program. Finally, although many childhood vaccination programs have historically been found to be cost-saving, the higher prices associated with several newer vaccines and rising emphasis on concerns about vaccine safety require explicit examination of the costs and health consequences of new programs [12-15]. An empirical assessment of the willingness-to-pay for a quality-adjusted life year gained may provide important information for decision-makers in the context of national vaccine policy.

Prior studies have estimated parental WTP to avoid minor vaccine adverse events such as local or systemic reactions. A 1999 study [20] reported a median parental WTP of $10 to $25 to reduce an infant's pain and emotional distress from childhood vaccination, while a study in 2001 reported a WTP of $25 to prevent fever and fussiness in young children after pneumococcal conjugate vaccine administration [25]. Another study conducted in 2002 examined parental WTP to avoid local and systemic reactions in adolescents, and found median estimates of $3 and $13, respectively [21]. Parents in our study reported slightly higher WTP values to prevent local ($30) and systemic ($50) reactions after a GAS vaccine, which may reflect differences in health state descriptions across studies, different considerations by parents depending on the age of the child (infant vs. toddler vs. adolescent), differences in the socioeconomic status of our population, inflation, or changes in the overall societal context regarding vaccine safety.

Empirically calculating the implied WTP per QALY may provide insight into the true societal WTP for gains in health, which may be preferred to using the persistent benchmark of $50,000 to $100,000 per QALY saved [26]. Although the standard approach in cost-effectiveness analysis relies on benchmarks for high-value interventions using the same threshold value for the WTP per QALY for all interventions, we observe substantial differences in the WTP per QALY to avoid different health states. In our study, parents were willing to pay more per incremental health gain to avoid vaccine adverse events (~$60,000 per QALY) compared to avoiding health states associated with GAS disease (~$18,000 to $36,000 per QALY). If these differences relate to true variability in the relative importance parents place on different types of outcomes, after controlling for the duration and severity of these outcomes, an important implication is that increased attention should focus on minimizing potential complications in healthy individuals. While it has previously been shown that treatment interventions are strongly preferred by society to preventive interventions [27,28], we are not aware of any studies that have explicitly compared preferences regarding vaccine adverse events vs. disease prevention.

Our findings that parents have a greater WTP per QALY for preventing vaccine adverse events compared to disease may be indicative of how individuals experience regret. An action, such as vaccinating a child, resulting in a potential adverse event may generate more regret than an inaction (i.e. refusing to vaccinate a child), even if a child becomes ill with a preventable disease [29-31]. This phenomenon is often characterized in terms of the distinction between "acts of commission" and "acts of omission", which is particularly relevant in the case of vaccination [32]. Parents may feel more guilt over agreeing to give a vaccine to their child that might cause harm, particularly in the short term, when compared to not vaccinating their child who by random chance develops disease. This may be reinforced by the changing perception of the risk-benefit balance by society, where fewer individuals have direct experience with vaccine-preventable diseases, furthering the intuitive response by some parents to focus more on vaccine safety and concerns about harming their child [33]. Further exploration of how regret for errors of commission and omission may influence parental preferences in vaccination programs is needed, particularly as new vaccines are recommended for use.

Our study has several limitations. First, our study population was relatively small and limited to parents of children who have experienced GAS pharyngitis. Consideration should be given to obtaining community values regarding GAS vaccination and disease [34]. Second, parents may not have had a complete understanding of the implications of these health states since our descriptions were brief and interviews were conducted by phone. As with any TTO, since parents were trading time from the end of their life, they may have assumed that they were trading time from a worse health state than their present condition and potentially have biased our TTO disutility estimates upward [35]. Third, anchoring bias may have occurred for our WTP and TTO estimates since we presented individuals with an initial opening bid that may have affected subsequent responses, although we did attempt to minimize this by randomizing among 3 different starting bids [36]. Fourth, missing or incomplete responses may have biased our WTP and TTO estimate in either direction. In a secondary analysis, however, our findings did not change significantly with the inclusion of predicted estimates for these individuals based on their characteristics. Fifth, WTP per QALY was inferred rather than directly elicited. Additionally, the pattern of declining WTP per QALY estimates for more severe health states may be due in part to the insensitivity to scale in WTP [37,38]. Finally, another key limitation of this study is that information was not available regarding parental refusal or deferral on any of their child's vaccines, thus we could not validate the WTP per QALY estimates with actual changes in behavior patterns.

Funding/Support: This study was supported by the Agency for Healthcare Research and Quality, US Department of Health and Human Services Grant, 5 K08 HS013908-04 (to GML).