Antiviral drugs for the treatment of influenza: a systematic review and economic evaluation

Objectives: To evaluate the clinical effectiveness (including adverse events) and cost-effectiveness of antivirals for the treatment of naturally acquired influenza for 'at-risk' and otherwise healthy populations. Data sources: Eleven electronic databases (MEDLINE, EMBASE, Cumulative Index to Nursing and Allied Health Literature, Pascal, Science Citation Index, BIOSIS, Latin American and Caribbean Health Sciences, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Database of Abstracts of Reviews of Effects, and Health Technology Assessment Database) were searched from October 2001 to November 2007. A supplementary search was undertaken in June 2008 for information relating to drug resistance during the 2007-8 influenza season. Review methods: Systematic reviews of the evidence on the clinical effectiveness and cost-effectiveness of antivirals for the treatment of influenza were undertaken. Twenty-nine randomised controlled trials comparing antivirals with each other, placebo, or best symptomatic care were included in the evaluation of clinical effectiveness in patients presenting with an influenza-like illness (ILI). Primary outcomes were measures of symptom duration (median time to alleviation of symptoms and median time to return to normal activity). Incidence of complications, mortality, hospitalisations, antibiotic use (as a surrogate for complications) and adverse events was also assessed. In addition, an independent decision model was developed to evaluate the cost-effectiveness of antiviral treatment from the perspective of the UK NHS. Results: Amantadine was excluded at an early stage, owing to a lack of any new trials that met the inclusion criteria and the limitations of the existing evidence. The review therefore focused on the neuraminiclase inhibitors (NIs) oseltarrivir and zanarnivir, both of which were found to be effective in reducing symptom duration (zanamavir by 0.5- 1.0 days and oseltarnivir by 0.5-1.5 days). However, the effect sizes were often small and unlikely to be clinically significant in many cases, particularly in healthy adults. For the at-risk subgroups, effect sizes for differences in symptom duration were generally larger, and potentially more clinically significant, than those seen in healthy adults (median duration of symptoms reduced by 1-2 days with zanarnivir and 0.50-0.75 days with oseltarnivir). However, there was greater uncertainty around these results, with estimates often failing to reach statistical significance. The most consistent data and strongest evidence related to antibiotic use, with both zanarnivir and oseltarnivir resulting in statistically significant reductions in antibiotic use. In general, the estimates from the cost-effectiveness model were more favourable in at-risk populations (including adults and children with comorbid conditions and the elderly) compared with otherwise healthy populations. Zanamivir was the optimal NI treatment in each of the at-risk populations considered, and oseltarnivir was optimal for healthy populations (both adults and children). Conclusions: The clinical effectiveness data for population subgroups used to inform the multiparameter evidence synthesis and costeffectiveness modelling were, in places, limited and this should be borne in mind when interpreting the findings of this review. Trials were often not designed to determine clinical effectiveness in population subgroups and hence, although the direction of effect was clear, estimates of differences in symptom duration tended to be subject to greater uncertainty in subgroups. Despite some concerns, the use of Nis in at-risk populations appeared to be a cost-effective approach for the treatment of influenza. Well-designed observational studies might also be considered to evaluate the clinical course of influenza in terms of complications, hospitalisation, mortality and quality of life, as well as the impact of Nis.