Purpose: Available multiattribute utility instruments have not been validated for very young children and there is significant variation in methods for valuing child health. Parents are commonly used as proxies to obtain values for infant health states. Little is known about how using proxies influences the magnitude of expressed preference, particularly parents as proxies for child health.

Methods: We reviewed data from 3 preference studies that valued health for young children using the time trade-off method, fielded between 2001-2006 via telephone and in-person. Respondents included parents of affected children (n=178), parents of healthy children (n=161), and members of the community (n=303). Health states valued included mild pneumococcal illness, invasive pneumococcal disease, vaccination and related adverse events, influenza and complications, and severe metabolic disorders. Responses were categorized as unusable (excluded from analysis) or usable. Unusable responses included (1) refusals, (2) illogical responses, and (3) unable to comprehend (as determined by the interviewer). Unusable responses were analyzed by respondent type and reason for exclusion. Respondent comments, when made, were reviewed for relevance to the valuation task.

Results: Unusable responses occurred as follows: refusals were highest for community members (16%) and lowest for parents of affected children (5%); illogical responses were lowest for community samples (8%) and highest for parents of affected children (44%); and inability to comprehend the task was similar for all respondent types (3-4%). Reasons for refusal included a difficulty in placing value on health, a belief in God, and rejection of the premise (“these questions are stupid”). Illogical responses were accompanied by comments regarding guilt (“I would do anything to keep my child healthy”) and an unwillingness to trade time for mild decrements in health (“Kids are sick all the time”).

Conclusions: Level of participation in preference surveys regarding children's health states may vary by respondent type. While parents of ill children may be more willing to participate, they may have more difficulty answering time trade-off questions than community members. In particular, factors specific to the caretaker role may enter the valuation process when parents are used as proxies, causing an over- or underestimation of values. Alternative proxies or a household or parent-child model of utility that includes family spillover effects should be considered when valuing children's health.