HEALTH UTILITIES IN PEDIATRIC COST-UTILITY STUDIES

Purpose: Cost-utility analyses can be powerful tools to inform society about effective use of health care resources, but pediatric health utility values are rarely assessed in a rigorous manner. Would rigorously measured pediatric utility values, obtained by both standard gamble and time trade off method in 4016 parent interviews, change the results of published pediatric cost–utility studies?

Method: We searched Medline, EMBASE, EconLit, Health Technology Assessment Database, Cochrane Database on Systematic Reviews, Database of Abstracts of Reviews of Effects and the CEA Registry at Tufts Medical Center from the beginning of each database until September 15, 2009. We included published pediatric cost-utility studies of children <18 years of age in which QALY were used as a measure of effectiveness and were health utilities for one of the following conditions had been used: ADHD, asthma, gastroenteritis, hearing loss, metal retardation, otitis media, seizure disorder and vision loss.  We determined the source and value of the used health utilities and whether the article conclusion favored the intervention. We further determined if results of univariate threshold sensitivity analyses on the used health utilities were presented, and when they were, if a change to our equivalent condition specific utility value would change the result.

Result: We identified 589 abstracts in our literature search and 39 articles met inclusion criteria.  Most analyses (85%) favored the studied intervention. The majority of studies used lower utilities values than those we had obtained (69%).  Only 18 articles presented results of univariate sensitivity analysis on used utilities, and the tested ranges were narrow.  Six articles presented sensitivity analyses that spanned over our equivalent health utility.  In four of the six articles the use of our utility value would change the main conclusion of the study. For the 33 articles were either no univariate threshold sensitivity analyses result was presented or our utility values fell outside the range tested, we determined that a change to our equivalent utility value would go against the study conclusion in 11 articles.

Conclusion: It appears to be a bias for pediatric cost-utility studies favoring the studied intervention. In addition, results of univariate sensitivity analyses of used health utilities are often left out, and when presented, the tested ranges are too narrow making comparison of similar studies challenging.