Objectives

Cost-utility analysis in which health benefits are quantified in terms of quality-adjusted life-years (QALYs) has now become the standard type of cost-effectiveness analysis. These studies are potentially influential in determining the extent of funding for particular pediatric interventions, and so their methodologic quality is extremely important. The objective of this study was twofold: first, to critically appraise published cost-utility analyses of interventions in child and adolescent health care in terms of the methods used to derive QALYs and, second, to discuss unresolved methodologic issues that are pertinent to the measurement of QALYs in pediatric populations.

Methods

A comprehensive search using computerized databases (including Medline, Embase, Econlit, and databases specific to economic evaluation), Web searches, and citation tracking was undertaken to identify cost-utility studies of interventions that were aimed at those who were younger than 16 years and published before April 2004. The methods of individual studies were compared with the recognized published guidelines of the US Panel on Cost-Effectiveness in Health and Medicine and the National Institute for Clinical Excellence in England and Wales, which recommend the use of a generic health state classification system (eg, Health Utility Index, EuroQol-5D), a choice-based valuation method (eg, standard gamble or time trade-off) and preferences of the general public in estimating QALYs. Studies therefore were categorized and evaluated according to the methods used to describe the health state, the valuation technique, and source of preferences.

Results

Fifty-four studies were reviewed, 34 (63%) of which were published in the past 5 years. A generic health status classification instrument was used in 22 (35%) cases; the remainder developed study-specific health state descriptions or elicited preferences directly from patients or proxies. In 3 (5%) cases, sources were unclear. Preference weights were elicited using choice-based techniques in 28 (42%) cases, either as tariffs for health status classification instruments (17 cases) or by directly valuing health state descriptions or patient health (11 cases). Preferences of the general public were used in only 23 (37%) cases. Four studies aggregated QALYs for mother/child or parents/child pairs without giving any theoretical justification. Although there was an increasing tendency for studies to use generic health status classification instruments, choice-based methods, and preferences of the general public, the majority of studies still did not adhere to these standard recommendations even in the period between January 2000 and March 2004. Despite increasing standardization in the methods advocated for economic evaluation over the past 10 years, there remains extensive variation in the actual methods used by researchers to calculate QALYs for children and adolescents. It is unclear whether these results suggest poor practice or a set of positive (or reactive) choices made by analysts in a methodologically uncertain area in which specific guidance is lacking regarding how to address the complexities of pediatric outcomes within the QALY framework. Many aspects of QALY measurement in children are not yet fully developed. In particular, there is (1) a lack of appropriate health state classification instruments that take account of the dynamics of child development, (2) a lack of health state classification instruments for use in children and infants who are younger than 5 years, and (3) the need to understand fully the role of proxies for measuring and valuing child health. Additional research efforts are also required to develop methods that account for the health benefits of parents or caregivers of the child and to consider the implications of combining different forms of utility measurement in childhood and adulthood.

Conclusions

Although variations from standard recommendations may be attributable to poor practice among researchers who are either unaware of these recommendations or choose not to follow them, they could equally be the result of attempts to make research more rigorous and more defensible than it might be if the standard recommendations were followed. There are 4 potential approaches to conducting cost-utility analysis in pediatric populations: (1) the explicit development of a generic instrument designed to be applicable across both child and adult populations (likely to be difficult in practice), (2) insistence on use of a generic instrument developed for adults, (3) the use of generic instruments specifically developed for children without being concerned about comparability with interventions aimed at adults, and (4) abandoning attempts to use single outcome measures that combine mortality with quality weights. In the absence of a clear way forward, it is suggested that an expert panel be convened to debate and further consider these potential solutions and recommendations for best practice and future research. In the interim, comparisons of the relative cost-effectiveness reported as cost per QALY gained across interventions for different diseases and populations should be treated with extreme caution.