This review assessed the use of oseltamivir and zanamivir for the treatment of influenza A and B. The authors concluded that both drugs appeared to reduce the duration of influenza illness, but that further research was required in certain populations. The review contains several potential sources of bias, as the authors acknowledged. The clinical importance of the small benefit from drug treatment is unknown.

This abstract addresses the use of neuraminidase inhibitors for the treatment of influenza A and B. Systematic reviews pertaining to the prophylactic use of neuraminidase inhibitors for the prevention of influenza A and B, and the use of amantadine hydrochloride for the treatment and prevention of influenza A, are summarised in other DARE abstracts.

The authors' objective was to assess the effectiveness of two neuraminidase inhibitors, oseltamivir and zanamivir, for the treatment of influenza A and B.

MEDLINE, EMBASE and the Science Citation Index (all from inception to December 2001), and PubMed and HEED were searched; the search terms were reported. The search findings were checked against 14 named registers and online databases, and the contents and archives of 17 named journals were also searched (further details were given in the report). In addition, the reference lists of identified articles, a named textbook, two relevant reports from the National Institute for Clinical Excellence, and personal databases were checked. Details of additional studies were sought from two pharmaceutical companies. Information on the studies had to have been available in English.

Study designs of evaluations included in the review

Double-blind, randomised controlled trials (RCTs) were eligible for inclusion.

Specific interventions included in the review

Studies of oseltamivir or zanamivir were eligible for inclusion. Studies had to use the drug formulations submitted for licensing approval. The included studies compared zanamivir (10 mg inhaled twice daily with or without 6.4 mg intranasal) or oseltamivir (75 or 150 mg twice daily) with placebo. Only comparisons of zanamivir 10 mg inhaled twice and oseltamivir 75 mg twice daily were included in the analysis. Where stated, the duration of treatment was 5 days.

Participants included in the review

Studies of patients who had contracted, or were suspected of having contracted, naturally occurring influenza were eligible for inclusion. Studies of patients who had been deliberately infected with experimental influenza in the course of a clinical trial were excluded from the review. The review classified patients into four groups: 'healthy' individuals aged 12 to 65 years; 'high-risk' individuals aged 12 years or older; 'healthy' children aged 12 or younger; and 'high-risk' children aged 12 or younger. The review defined 'high-risk' as patients of any age with concurrent disease of sufficient severity to require regular medical follow-up, or otherwise healthy patients aged 65 years or older. Across the studies, between 0 and 46% of the patients had been vaccinated.

Outcomes assessed in the review

Studies that assessed at least one of the following outcomes were eligible for inclusion: time to alleviation of symptoms; time to alleviation of major influenza symptoms; time to eradication of major signs and symptoms; time to return to normal activities; the number of days symptoms were scored as 'none' or 'mild'; complications requiring the use of antibiotics; treatment-related adverse events; and hospitalisations. The included studies used different definitions for time to symptom alleviation. Most of the studies rated symptoms as absent, mild, moderate or severe for up to 21 days (further details were given in the paper). Where stated, the duration of follow-up ranged from 5 to 28 days.

How were decisions on the relevance of primary studies made?

The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.

The authors assigned a quality score, using the Jadad scale to assess randomisation, blinding and the reporting of withdrawals. The scale ranged from 1 (lowest) to 5 (highest). Studies with a score of 4 or 5 were regarded as high quality. The authors did not state how the papers were assessed for quality, or how many reviewers performed the quality assessment.

Data were extracted for the review using a data extraction form, but the authors did not state how many reviewers performed the data extraction.

For time-to-event data (time to alleviation of symptoms and time to return to normal activities) the median and its standard error were extracted or calculated for each individual study. Where insufficient data were available, additional data were requested from the manufacturer. Studies used different methods to calculate medians (some allowed for censored data and others did not). In studies of zanamivir, time-to-event was measured in days and rounded up to the nearest half-day. In studies of oseltamivir, time-to-event was measured in hours and rounded up to the nearest hour. Where studies included patients from more than one group, stratified data were requested from the manufacturer. Data from the individual studies were extracted on an intention-to-treat (ITT) basis and for patients with clinically confirmed influenza (influenza positive). For treatment-related adverse events, hospitalisations and complications requiring the use of antibiotic data on the occurrence of each event, or the proportion of patients experiencing the event, were extracted from each study.

How were the studies combined?

The studies were combined using a random-effects meta-analysis where possible. A pooled median with 95% confidence intervals (CIs) was calculated separately for the difference between the licensed dose of zanamivir or oseltamivir and placebo for time-to-event outcomes. Separate meta-analyses were conducted for each of the four subgroups using the ITT population and the influenza-positive patients.

Previous pooled analyses of individual patient data were used for complication end points, as these contained more complete data than were available to the reviewers. A meta-analysis of complication end points was also undertaken using only the data available to the reviewers. A narrative summary was presented where meta-analysis was not possible. The reviewers requested that pharmaceutical companies reanalyse data for total populations and for influenza-positive patients. The review reported two estimates of the pooled median time-to-event for each study where data were sufficient; one used published data and made no allowance for censored data (published analysis), while the other was the median provided on request from the pharmaceutical companies (published reanalysis).

How were differences between studies investigated?

Sensitivity analyses were conducted to assess the effects of including only data published in peer-reviewed journals or high-quality trials. Sources of clinical heterogeneity among studies were discussed in the paper.

Eighteen RCTs were included in the review. Nine RCTs reported in 11 studies compared zanamivir (10 mg inhaled twice daily) with placebo (n=3,552). Nine RCTs compared oseltamivir (75 mg twice daily) with placebo (n=2,740).

For zanamivir, three RCTs scored 5 on the Jadad scale, two RCTs scored 4, two RCTs scored 3 and two RCTs scored 2. For oseltamivir, three RCTs scored 5 on the Jadad scale, three RCTs scored 4 and three RCTs scored 2.

Zanamivir (10 mg twice daily) versus placebo.

Healthy patients aged 12 to 65 years: zanamivir significantly reduced the time to alleviation of symptoms compared with placebo (median difference 0.78 days, 95% CI: -1.31, -0.26), based on ITT data from 1,947 patients in five RCTs. Similarly, zanamivir significantly reduced time to alleviation of symptoms in the influenza-positive subgroup (median difference 1.26 days, 95% CI: -1.93, -0.59), based on 1,327 patients in five RCTs.

High-risk subgroup aged 12 years or older: zanamivir reduced the time to alleviation of symptoms compared with placebo, but the difference was not statistically significant (median difference 0.93 days, 95% CI: -1.90, 0.05), based on ITT data from 763 patients in five RCTs. Zanamivir significantly reduced the time to alleviation of symptoms in the influenza-positive subgroup compared with placebo (median difference 1.99 days, 95% CI: -3.08, -0.90), based on 484 patients in five RCTs.

Children aged 5 to 12 years (1 RCT, n=435): zanamivir significantly reduced the time to symptom alleviation compared with placebo in the normally healthy subgroup (median difference 1.0 days, 95% CI: -1.5, -0.5), and a non significant benefit was found in a small (n=36) high-risk subgroup (median difference 2.0 days, 95% CI: -6.9, 2.9). The results were similar for the influenza-positive subgroup (healthy: median difference -1.0 days, 95% CI: -1.6, -0.4; high-risk: median difference -3.8 days; 95% CI: -7.6, 0.1).

All patients regardless of age or risk status (7 RCTs, n=3,181): zanamivir significantly reduced the time to symptom alleviation compared with placebo (median difference 0.94 days, 95% CI: -1.23, -0.65). Similarly, zanamivir significantly reduced the time to symptom alleviation compared with placebo in the influenza-positive subgroup (median difference 1.26 days, 95% CI: -1.61, -0.90).

Oseltamivir (75 mg twice daily) versus placebo.

Healthy patients aged 12 to 65 years: oseltamivir significantly reduced the time to alleviation of symptoms compared with placebo (median difference 20.69 hours, 95% CI: -33.97, -4.41), based on ITT data for 937 patients in 3 RCTs. Similarly, oseltamivir significantly reduced the time to alleviation of symptoms compared with placebo in the influenza-positive subgroup (median difference 33.1 hours, 95% CI: -47.1, -19.1), based on 605 patients in 3 RCTs.

High-risk subgroup aged 12 or over: oseltamivir reduced the time to alleviation of symptoms compared with placebo, but the difference was not statistically significant (median difference 8.33 hours, 95% CI: -33.69, 17.03), based on ITT data for 733 patients in five RCTs. Similarly, no significant difference was found in influenza-positive patients (median difference 10.91 hours, 95% CI: -45.04, 23.22), based on 728 patients in five RCTs.

Children aged 5 to 12 years (1 RCT, n=669): oseltamivir significantly reduced the time to symptom alleviation compared with placebo in the normally healthy subgroup (median difference 20.9 hours, 95% CI: -35.7, -6.1) and in the influenza-positive subgroup (median difference 35.8 hours, 95% CI: -53.3, -18.2). The RCT did not include any high-risk children.

All patients regardless of age or risk status (9 RCTs, n=2,740): oseltamivir significantly reduced the time to symptom alleviation compared with placebo (median difference 19.15 hours, 95% CI: -28.36, 9.94). Similarly, oseltamivir significantly reduced the time to symptom alleviation compared with placebo in the influenza-positive subgroup (median difference 31.98 hours, 95% CI: -42.40, -21.56).

Sensitivity analyses. For 'healthy' participants, the magnitude and direction of effect were similar after excluding studies scoring less than 4 on the Jadad scale. For 'high-risk' patients, the results from the two RCTs scoring 4 or 5 on the Jadad scale were unchanged from the main meta-analysis.

Adverse events. The definition of adverse events varied among studies for both drugs, and adverse events were difficult to separate from complications. Data on adverse events were not reported in the review.

Data on time to return to normal activities, and complications requiring antibiotics were presented in the report.

A review of the cost-effectiveness of oseltamivir or zanamivir was conducted. The base-case analysis found the cost-effectiveness of these agents to be relatively unfavourable. Uncertainty analyses found probabilities between 55 and 60% and 50 and 68% of a cost per quality-adjusted life-year below £30,000 for oseltamivir and zanamivir, respectively, for the four subgroups evaluated. The authors concluded that cost-effectiveness varied markedly between the target populations.

Oseltamivir 75 mg twice daily for 5 days and zanamivir 10 mg twice daily, both taken for 5 days, appeared to reduce the duration of influenza illness. However, further research is required in certain high-risk populations.

The review question was clear and the inclusion criteria appeared appropriate. The search was extensive and attempts were made to limit publication bias. However, only English publications were included, thus introducing language bias. The methods used to select studies, assess quality and extract the data were not described, so it is not known whether any efforts were made to reduce errors and bias. Relevant information was presented for studies where possible, but details of a number of studies could not be published because of commercial confidentiality. In addition, data from several identified studies were unavailable.

The studies were combined using meta-analyses, and forest plots that allowed an examination of statistical heterogeneity were presented for most analyses. For some studies, the reviewers were dependent upon summary results from analyses conducted by pharmaceutical companies. The decision to pool medians of time-to-event data had limitations, as the authors acknowledged. The use of hazard ratios to combine studies of time-to-event data would have been more appropriate, as it would have allowed for a more complete representation of the data. The use of medians means that information could be missing from the results, although the authors did state that pooling medians was a non-standard practice.

The authors reported some of the potential sources of bias, such as different definition used for efficacy outcomes, the use of different scales for time-to-events for the two drugs, the existence of unpublished studies for which data could not be obtained, and inconsistent and incomplete reporting of adverse events. The review showed small but statistically significant benefits for the use of neuraminidase inhibitors, although the clinical importance of this is unknown. The several potential sources of bias in the included studies and methods of analysis, as well as limitations in the reporting of the review process, weaken the evidence presented in the review.

Practice: The authors did not state any implications for practice.

Research: The authors stated that further research is required to directly compare oseltamivir, zanamivir and amantadine; to assess the effects of these drugs in high-risk individuals; and to assess their effects on hospitalisations, deaths and quality of life.

NHS R&D Health Technology Assessment (HTA) Programme, project number 01/47/01.

Cooper NJ, Sutton AJ, Abrams KR, Wailoo A, Turner DA, Nicholson KG. Effectiveness of neuraminidase inhibitors in treatment and prevention of influenza A and B: systematic review and meta-analyses of randomised controlled trials. BMJ 2003;326:1235-9.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.