Varicella Vaccine, Cost-effectiveness Analyses, and Vaccination Policy

Abstract

HAS VARICELLA VACCINATION BEEN AS COSTeffective as anticipated when it was universally recommended for children 10 years ago? The answer is complicated and illuminates an evolving controversy regarding the role of cost-effectiveness analyses in deliberations over national vaccine recommendations. In 1994, Lieu and colleagues estimated the costeffectiveness of a universal childhood varicella vaccination program. Projecting a national vaccination rate of 97% by the sixth year of the program, they estimated a 94% reduction in varicella incidence, accompanied by an 89% reduction in direct medical costs. Accounting as well for reductions in the quantity of parents’ and other adults’ lost work time attributable to varicella (ie, the “indirect” benefits of vaccination), Lieu et al projected that a varicella vaccination program would save more than $5 for every $1 spent. These projections of reduced varicella disease burden and direct medical costs with varicella vaccination appear prescient, even though they were based on optimistic vaccination rates. In this issue of JAMA, Zhou and colleagues present data from a national group of privately insured individuals in 1994-2002 that indicate reductions of nearly 90% in hospitalization rates for varicella, nearly 60% in ambulatory visit rates, and 74% in costs borne by health care payers— accomplished in an era when slightly more than 80% of US toddlers had been vaccinated against chickenpox. Cost reductions might have been even more substantial if Zhou et al had sufficient data to analyze pharmaceutical expenditures for antiviral agents that Lieu et al considered. Although the study by Zhou et al is based solely on trends among insured individuals, the magnitude of their findings corresponds well to patterns of varicella incidence and mortality collected without regard to payer and to varicellarelated hospitalization trends and associated charges collected on insured and uninsured children and adults. In addition, these data support the strong evidence of herd immunity against varicella in the vaccination era, with adults experiencing the benefits of children’s vaccination, and infants—who are not eligible for vaccination—showing the greatest reductions of all child age groups. Nonetheless, these findings do not conclusively confirm that childhood varicella vaccination is as cost-effective as originally anticipated, for several reasons. First, the cost of the vaccine has increased more than $10 per dose in inflationadjusted terms since 1995 (the current public sector price per dose is $52.25), although an increase of this magnitude was not anticipated to change the cost-effectiveness dramatically. Second, the national varicella vaccine recommendation prompted states to measure and react to varicella as a reportable vaccine-preventable illness. The costs of such monitoring and of responding to outbreaks of varicella (eg, in day care or school settings) may be substantial and were not included in the original analysis by Lieu et al. Third, and perhaps most important, there is great uncertainty about the extent to which parents and other adults experienced reductions in lost work time attributable to varicella. As with other childhood and adolescent vaccines that have recently been recommended (eg, pneumococcal and meningococcal conjugate vaccines), indirect cost savings with varicella vaccine were expected to be larger than savings in direct medical costs. With varicella vaccine, the ratio of indirect cost savings to direct cost savings was expected to be nearly 5:1. Despite the importance of indirect cost quantification to the evaluation of vaccination programs, there are no established mechanisms to measure such effects. Therefore, it is possible that the increasing value of work time over the past decade may have increased savings beyond what Lieu et al anticipated. Conversely, increasing unemployment may have allowed more parents to care for their ill children at home without missing work, effectively reducing savings attributable to the program. Both of these changes may have occurred disproportionately in specific sectors of the economy (eg, wage increases in the software industry and unemployment increases in the manufacturing industry) that may have different likelihoods of having young children in the household. These changes are plausible, but their magnitude remains unknown. Pointing out current limitations in measuring indirect benefits of vaccination programs may seem methodologically nitpicky, but it is more than an academic exercise. The contem-