Abstract

Background.

Oseltamivir is widely used for the treatment of influenza. Previous systematic reviews suggest that they reduce complications, but had significant methodologic limitations.

Objective.

To assess the effect of oseltamivir on duration of symptoms, complications and hospitalizations in adults.

Methods.

We searched Medline without time or language restrictions, and trial registries maintained by the manufacturer. We included published and unpublished randomized double-blinded, placebo-controlled trials of oseltamivir in adults with suspected influenza that reported duration of symptoms, complications or hospitalizations. We abstracted data regarding study quality, the duration of symptoms and rates of complications and hospitalization.

Results.

Three published and eight unpublished studies met our inclusion criteria. For the intention-to-treat (ITT) population, the mean reduction in the duration of symptoms was 20.7 hours [95% confidence interval (CI) 13.3 to 28.0 hours]. Two large unpublished studies in the elderly and in adults with chronic disease did not find a significant reduction in the symptom duration. There was no difference in the likelihood of hospitalization in the ITT population (33/2633 patients for oseltamivir versus 20/1694 for placebo). The rate of complications in the intention-to-treat infected (ITTI) population was reduced when acute bronchitis was included (−2.8%, 95% CI −0.6 to −4.9), but not when it was excluded. The risk of pneumonia was reduced in the ITTI population (−0.9%, 95% CI −0.1 to −1.7) but not in the ITT population.

Conclusions.

There is no evidence that oseltamivir reduces the likelihood of hospitalization, pneumonia or the combined outcome of pneumonia, otitis media and sinusitis in the ITT population.

Introduction

Oseltamivir is widely used for the treatment of acute influenza. When taken within 36 hours of the onset of symptoms, it has a modest benefit in terms of symptom duration based on published studies.1–3 As a result, a recent study found that an antiviral medication was prescribed for 58% of patients with suspected influenza, but often inappropriately (i.e. initiated beyond 2 days of symptom onset).4

Nine randomized controlled trials (RCTs) sponsored by the manufacturer have never been published in full, including an RCT with 1447 healthy adults, an RCT with 401 adults who had a chronic cardiac or respiratory disease and three RCTs with a total of 736 adults over 65 years of age. The failure to publish these data is concerning, since patients agreed to participate in studies of what was then an experimental drug to advance medical knowledge.

An important area of controversy is the extent to which osteltamivir prevents important complications of influenza such as pneumonia, hospitalization and death, especially in older patients and those with chronic disease. While previous systematic reviews have concluded that oseltamivir reduces the risk of complications,5,6 the authors considered acute bronchitis as a complication that required an antibiotic, something that is inconsistent with current practice. Also, the impact of oseltamivir on the duration of illness has not been fully reported for the elderly or chronically ill. As a result, the clinical picture of the effectiveness of oseltamivir in adults is incomplete. The goal of this systematic review was to determine the effect of oseltamivir on the duration of illness and prevention of serious complications and hospitalizations in adults using both published and unpublished data, by a group of authors with no ties to the manufacturer and considering an appropriate definition for complications requiring antibiotics.

Methods

We performed a systematic review of the literature to identify randomized, double-blinded, placebo- controlled trials comparing oseltamivir to placebo in adults. Our study protocol followed the guidelines for reporting recommended by the Preferred Reporting Items for Systematic Reviews Meta-Analyses statement.

Data sources and searches

We used the following initial search strategy in MEDLINE, limited to RCTs and systematic reviews:

 
Influenza [Ti/Ab} AND oseltamivir [Ti/Ab]

This returned a total of 60 articles. We also reviewed the reference lists of any studies that met our initial review, the five systematic reviews returned by our search, the meta-analysis in the Cochrane database of systematic reviews1 and the Cochrane Controlled Trials Register (71 studies). A total of three published studies met our inclusion criteria.2,3,7

Several published RCTs identified by our search merit further discussion. A study published in Russian was reviewed by a native Russian speaker but did not provide usable data for our analysis.8 We also excluded two studies that had an open-label design and did not mask patients or outcome assessors.9,10 We identified one case of duplicate publication.11,12 Both of these studies were excluded because they only reported median duration of symptoms, and did not report the mean duration of symptoms or the rate of hospitalization or complications in the treatment and placebo groups. The authors do state that there was no difference in the likelihood of complications but do not report absolute numbers.

We requested and obtained permission from Roche to download clinical study reports for 10 published and unpublished RCTs from the site https://tamiflu.clinit.net/. These reports ranged in length from 199 to 1004 pages and provided detailed data the study design and results. Other than a single email providing permission to access the data for research purposes, we have had no further contact with the sponsor. We also identified one additional RCT in a search of the Roche trials database (http://www.rochetrials.com/trialResults.action). This study (WV16277) randomized 451 adults to oseltamivir or placebo, and stratified them by vaccination status and the presence or absence of chronic obstructive lung disease. They found no difference between groups for the primary outcome of cough and fever alleviated 36 hours after the onset of treatment. Because they did not report the overall duration of symptoms, complications or hospitalizations we were unable to use it in our analysis. Table 1 provides detailed information about the three published and eight unpublished RCTs that met our inclusion criteria.

Table 1

Characteristics of included studies

Study Site Data collection Population Group ITT ITTI (% of ITT) Mean age in years (SD) 
M76001 USA 24 December 1998 to 19 February 1999  Adults age 13 to 80 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 482 361 (74.9%) 34.9 (14.5) 
75mg bid 965 702 (72.7%) 35.4 (14.3) 
Nicholson et al.2 Europe, China and Canada 12 December 1997 to 18 April 1998  Adults age 18 to 65 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 235 161 (68.5%) 37.4 (11.9) 
75mg bid 241 158 (65.6%) 38.2 (11.1) 
150mg bid 243 156 (64.2%) 36.7 (11.8) 
Treanor et al.3 USA 23 December 1997 to 20 April 1998 Adults age 18 to 65 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 204 129 (63.2%) 32.4 (10.2) 
75mg bid 206 124 (60.2%) 32.3 (10.8) 
150mg bid 205 121 (59.0%) 33.1 (9.8) 
WV15707 Australia, South Africa and South America 21 February 1998 to 16 November 1998 Elderly adults over 65 years of age with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo  6 (66.7%) 71.1 (6.2) 
75mg bid  17  6 (35.3%) 71.9 (5.1) 
WV15730 Australia and South Africa 1 July 1998 to 21 September 1998 Adults age 18 to 65 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo  27  19 (70.4%) 36.4 (12.1) 
75mg bid  31  19 (61.3%) 34.1 (9.9) 
WV15812 and WV15872 WV15812: USA, Canada and Europe; WV15872: Australia, New Zealand and South Africa WV15812: 5 January 1999 to 12 April 1999; WV15872: 2 June 1999 to 1 October 1999 Adults age 13+ with chronic cardiac or respiratory disease and ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 202 133 (65.8%) 49.9 (17.4) 
75mg bid 199 118 (59.3%) 53.7 (18.0) 
WV15819, WV15876 and WV15989 Europe, USA, Canada, Israel, South Africa, New Zealand and Australia 15 January 1999 to 23 March 2000 Elderly adults age 65 or older with ILI (fever ≥ 37.5°C + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 376 254 (67.6%) 72.9 (6.7) 
75mg bid 360 223 (61.9%) 73.0 (6.1) 
Kashiwagi et al.7  Japan 1998 to 2000 Adults age 16 to 80 with ILI (fever > 38°C + two of the following symptoms: cough, nasal symptoms, sore throat, sweating, fatigue, headache, myalgia) presenting within 36 hours. Only seven patients age 70 or older. Placebo 162 130 (80.2%) 33.6 (13.9) 
75mg bid 154 122 (79.2%) 35.5 (14.6) 
Study Site Data collection Population Group ITT ITTI (% of ITT) Mean age in years (SD) 
M76001 USA 24 December 1998 to 19 February 1999  Adults age 13 to 80 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 482 361 (74.9%) 34.9 (14.5) 
75mg bid 965 702 (72.7%) 35.4 (14.3) 
Nicholson et al.2 Europe, China and Canada 12 December 1997 to 18 April 1998  Adults age 18 to 65 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 235 161 (68.5%) 37.4 (11.9) 
75mg bid 241 158 (65.6%) 38.2 (11.1) 
150mg bid 243 156 (64.2%) 36.7 (11.8) 
Treanor et al.3 USA 23 December 1997 to 20 April 1998 Adults age 18 to 65 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 204 129 (63.2%) 32.4 (10.2) 
75mg bid 206 124 (60.2%) 32.3 (10.8) 
150mg bid 205 121 (59.0%) 33.1 (9.8) 
WV15707 Australia, South Africa and South America 21 February 1998 to 16 November 1998 Elderly adults over 65 years of age with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo  6 (66.7%) 71.1 (6.2) 
75mg bid  17  6 (35.3%) 71.9 (5.1) 
WV15730 Australia and South Africa 1 July 1998 to 21 September 1998 Adults age 18 to 65 with ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo  27  19 (70.4%) 36.4 (12.1) 
75mg bid  31  19 (61.3%) 34.1 (9.9) 
WV15812 and WV15872 WV15812: USA, Canada and Europe; WV15872: Australia, New Zealand and South Africa WV15812: 5 January 1999 to 12 April 1999; WV15872: 2 June 1999 to 1 October 1999 Adults age 13+ with chronic cardiac or respiratory disease and ILI (fever + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 202 133 (65.8%) 49.9 (17.4) 
75mg bid 199 118 (59.3%) 53.7 (18.0) 
WV15819, WV15876 and WV15989 Europe, USA, Canada, Israel, South Africa, New Zealand and Australia 15 January 1999 to 23 March 2000 Elderly adults age 65 or older with ILI (fever ≥ 37.5°C + one respiratory and one constitutional symptom) presenting within 36 hours. Placebo 376 254 (67.6%) 72.9 (6.7) 
75mg bid 360 223 (61.9%) 73.0 (6.1) 
Kashiwagi et al.7  Japan 1998 to 2000 Adults age 16 to 80 with ILI (fever > 38°C + two of the following symptoms: cough, nasal symptoms, sore throat, sweating, fatigue, headache, myalgia) presenting within 36 hours. Only seven patients age 70 or older. Placebo 162 130 (80.2%) 33.6 (13.9) 
75mg bid 154 122 (79.2%) 35.5 (14.6) 

ILI, influenza-like illness.

Study selection

We included both published studies and unpublished randomized, double-blinded, placebo-controlled trials that compared oseltamivir with placebo in patients with suspected or confirmed influenza. We excluded studies that artificially infected persons with influenza virus, studies in children under the age of 12 years, studies that did not report any of the outcomes of interest and studies that did not mask both patients and outcome assessors to treatment assignment. Exclusion of the latter is especially important when the outcomes are subjective (self-assessment of symptoms) or require a subjective decision (did this patient have a complication, or should I hospitalize this patient). While we did not explicitly exclude pregnant patients from the systematic review, pregnant participants were excluded by most of the reporting articles.

The decision to include unpublished studies was based on the significant number of large, well-designed, yet unpublished studies on the topic. While unpublished trials have not undergone formal peer review, this ‘gray literature’ has been used by the Cochrane library and others in systematic reviews.13 Unpublished data tends to have a lower estimate of effect than published research. This may be due to suppression of research by sponsors because it does not show a beneficial treatment effect,14,15 or an editorial preference for studies showing a statistically significant treatment effect.13 While there is concern that unpublished studies are less methodologically strong than published studies,16 in this case, the unpublished studies that we identified were nearly identical in terms of study design to the published studies.

Data extraction and quality assessment

All three of the authors reviewed all of the abstracts returned by the initial search. Any article identified by at least one of the reviewers as possibly meeting the inclusion criteria based on the abstract was reviewed in full by all three authors. A final decision regarding inclusion was made by consensus. Two authors then abstracted data regarding study design and quality, the intervention, duration of symptoms, complications, hospitalizations and deaths in parallel. They then compared the data abstracted and resolved discrepancies by discussion with the third author. Quality of included studies was evaluated using the Jadad criteria.

Outcomes

The primary outcomes were the differences between oseltamivir and placebo groups regarding the mean duration of symptoms, the likelihood of complications and the likelihood of hospitalization. Most of the trials (including all of those sponsored by the manufacturer) were consistent in their reporting of influenza complications, and defined a complication requiring an antibiotic as otitis media, acute bronchitis, pneumonia and sinusitis. However, the results of systematic reviews and the recommendations of evidence-based practice guidelines do not support the use of antibiotics for acute bronchitis.17–19 We therefore report two composite outcomes for complications, one that includes acute bronchitis and one that does not.

The number of hospitalizations was ascertained from published reports or from the detailed description of complications in the clinical study reports. We considered both hospitalizations for any cause and only those hospitalizations that we felt were related either to influenza or adverse effects of the drug (i.e. due to respiratory complications, pneumonia, sepsis, nausea, vomiting or dehydration).

Data synthesis and analysis

We evaluated the heterogeneity of continuous and dichotomous data using the I2 statistic, with 0% indicating that the variance was entirely within studies (homogeneity) and 100% indicating that the variance was entirely between studies (heterogeneity). For the weighted mean difference between treatment and placebo groups regarding duration of symptoms we calculated both fixed effects (inverse variance) and random effects (DerSimonian–Laird) models using the METAN module in Stata version 11.1 (StataCorp., College Station, TX). As there was no difference in results between the models (I2 = 0%), we report only the fixed effects results. For dichotomous data, we used a random effects model to determine summary estimates of the risk difference and its 95% confidence interval (CI), again using the METAN module in Stata. Two studies2,3 evaluated both 75 and 150mg doses of oseltamivir in comparison with placebo. We treated these as separate studies for the outcome of mean duration of symptoms, but not for complications and hospitalizations in order to avoid double counting of those outcomes in the placebo group.

Results

We identified a total of 11 studies with 4769 patients that met our inclusion criteria (Table 1). Results from three unpublished studies in older adults (WV15819, WV15876 and WV15989) and two unpublished studies of adults with chronic respiratory or cardiac disease (WV15812 and WV 15872) were reported in aggregate by the manufacturer because they had nearly identical methods. Three other studies were unpublished (M76001, WV15707 and WV15730), and three had been previously published.2,3,7

We analyzed and report data separately for intention-to-treat (ITT) and intention-to-treat infected (ITTI) populations. The ITTI population includes only patients confirmed to have influenza by a laboratory-based reference standard test. The ITT population includes patients with suspected influenza and is a better reflection of the patients seen in real-world clinical practice, where reference laboratory confirmation is not readily available.

Quality of included studies

The quality of included studies is summarized in Table 2. In general, studies were very well designed with adequate description of randomization, allocation concealment, masking and follow-up. The primary limitation was a failure by three studies to report primary outcomes for both ITT and ITTI populations (WV15707, WV15730 and WV15812). All of the included studies had a Jadad score of 5 out of 5 points (randomized, blinded, information about withdrawals provided, appropriate method for randomization and appropriate method for blinding).20

Table 2

Quality of included studies

Study Randomized Method of randomization described Investigator masked Participant masked Allocation concealed Description of dropouts and withdrawals <80% dropouts or withdrawals ITT results reported for primary outcomes 
M76001 
WV15670/Nicholson et al.2 
WV15671/Treanor et al.3 
WV15707 
WV15730 
WV15812 and WV158872 
WV15819, WV15876 and WV15989 
Kashiwagi et al.7 
Study Randomized Method of randomization described Investigator masked Participant masked Allocation concealed Description of dropouts and withdrawals <80% dropouts or withdrawals ITT results reported for primary outcomes 
M76001 
WV15670/Nicholson et al.2 
WV15671/Treanor et al.3 
WV15707 
WV15730 
WV15812 and WV158872 
WV15819, WV15876 and WV15989 
Kashiwagi et al.7 

Duration of symptoms

The duration of symptoms was reported by five studies with 3833 patients for the ITT population and seven studies with 2690 patients for the ITTI population. The mean reduction in the duration of symptoms was 20.7 hours for the ITT population (95% CI 13.3 to 28.0 hours), and is summarized in Figure 1a. The mean reduction in the duration of symptoms was 25.4 hours for the ITTI population (95% CI 17.2 to 33.5 hours), and is shown in Figure 1b. There was excellent homogeneity (I2 = 0%) for both populations.

Figure 1

Forest plot of mean reduction in the duration of symptoms between oseltamivir and placebo groups, stratified by published versus unpublished studies in the intention-to-treat population (a) and intention-to-treat infected (b) populations.

Figure 1

Forest plot of mean reduction in the duration of symptoms between oseltamivir and placebo groups, stratified by published versus unpublished studies in the intention-to-treat population (a) and intention-to-treat infected (b) populations.

Both published studies found statistically significant reductions in the duration of symptoms for both 75 and 150mg doses of oseltamivir in the ITTI analysis (Fig. 1b). Among the unpublished studies, two were quite small and had very broad CIs (WV15707 and WV15730), while one (WV76001) was the largest of all studies with 1547 patients. Two RCTs appeared to be adequately powered but did not find a statistically significant reduction in the duration of symptoms. Study WV15819 included 736 adults over age 65 years, and found only a 7.4 hour mean reduction in symptoms in the ITT population and 15.8 hours in the ITTI population (neither was statistically significant). Study WV15812 included 401 adults with at least one chronic cardiac or respiratory condition, and found a non-significant 1.4 hour reduction in the duration of symptoms in the ITTI population (data for ITT were not reported).

Complications and hospitalizations

Data regarding complications and hospitalizations are also reported separately for ITT and ITTI populations in Table 3. Hospitalizations were only reported for the ITT population, and were rare: 33/2633 patients in the oseltamivir group and 20/1694 patients in the placebo group. There was no significant difference between oseltamivir and placebo groups regarding the likelihood of any hospitalization (risk difference [RD] 0.1%, 95% CI −0.5% to 0.6%). We also looked separately at only those hospitalizations due to respiratory complications, sepsis or dehydration (a possible side effect of oseltamivir) and again found no difference between groups in the ITT population (RD 0.0%, 95% CI −0.5% to 0.4%).

Table 3

Rates of complications and hospitalizations

Study Risk difference in % (95% CI)a Weight 
INTENTION-TO-TREAT POPULATIONb 
Any hospitalization 
 Kashiwagi et al.7 4.7 (0.0 to 9.4) 1.2 
 M76001 0.1 (−0.8 to −0.10) 35.1 
 Nicholson et al.2 −0.2 (−1.1 to 0.7) 32.0 
 Treanor et al.3 0.2 (−1.0 to 1.5) 17.2 
 WV15707 0.7 (−0.25 to 0.26) 0.04 
 WV15730 0.0 (−6.5 to 6.5) 0.6 
 WV15812 −1.0 (−4.0 to 2.1) 3.0 
 WV15819 0.3 (−1.3 to 1.9) 10.8 
I2 = 0.0%  Summary effect: 0.1 (−0.5 to 0.6) 100.0 
Hospitalization for sepsis, respiratory complication, or dehydration 
 M76001 0.2 (−0.6 to 1.0) 35.3 
 Nicholson et al.2 −0.2 (−1.1 to 0.7) 23.7 
 Treanor et al.3 0.2 (−0.6 to 1.1) 25.6 
 WV15730 0.0 (−6.5 to 6.5) 0.5 
 WV15812 −1.0 (−3.7 to 1.8) 2.7 
 WV15819 −0.5 (−1.8 to 0.8) 12.1 
I2 = 0.0%  Summary effect: 0.0 (−0.5 to 0.4) 100.0 
Pneumonia 
 M76001 −0.8 (−0.2 to 0.3) 40.5 
 Nicholson et al.2 0.0 (−1.0 to 1.0) 46.8 
 WV15707 −11.1 (−34.5 to 12.3) 0.2 
 WV15812 −2.0 (−4.7 to 0.8) 12.5 
I2 = 31.0%  Summary effect: −0.6 (−1.7 to 0.4) 100.0 
INTENTION-TO-TREAT INFECTED POPULATIONc 
Pneumonia   
 M76001 −1.7 (−3.1 to −0.2) 32.8 
 Nicholson et al.2 −0.6 (−2.2 to 0.9) 30.2 
 Treanor et al.3 −0.8 (−2.7 to 1.1) 19.4 
 WV15730 −5.3 (−18.7 to 8.2) 0.4 
 WV15812 0.2 (−2.9 to 3.3) 7.4 
 WV15819 −0.1 (−2.8 to 2.6) 9.8 
I2 = 0.0%  Summary effect: −0.9 (−1.7 to −0.1) 100.0 
Complications requiring an antibioticd 
 M76001 −1.7 (−5.2 to 1.8) 37.9 
 Nicholson et al.2 −2.7 (−6.5 to 1.0) 33.2 
 Treanor et al.3 −6.0 (−12.0 to 0.1) 12.8 
 WV15730 5.3 (−11.8 to 22.3) 1.6 
 WV15812 −0.7 (−10.8 to 9.2) 4.6 
 WV15819 −4.8 (−11.7 to 2.0) 9.9 
I2 = 0.0%  Summary effect: −2.8 (−4.9 to −0.6) 100.0 
Complications requiring an antibiotic, excluding acute bronchitis 
 M76001 0.7 (−2.0 to 3.5) 34.5 
 Nicholson et al.2 −2.1 (−5.4 to 1.1) 24.9 
 Treanor et al.3 −2.2 (−6.7 to 2.3) 12.9 
 WV15730 5.3 (−11.8 to 22.3) 0.9 
 WV15812 3.3 (−3.3 to 9.9) 6.0 
 WV15819 0.9 (−2.6 to 4.5) 20.8 
I2 = 0.0%  Summary effect: −0.1 (−1.7 to 1.5) 100.0 
Study Risk difference in % (95% CI)a Weight 
INTENTION-TO-TREAT POPULATIONb 
Any hospitalization 
 Kashiwagi et al.7 4.7 (0.0 to 9.4) 1.2 
 M76001 0.1 (−0.8 to −0.10) 35.1 
 Nicholson et al.2 −0.2 (−1.1 to 0.7) 32.0 
 Treanor et al.3 0.2 (−1.0 to 1.5) 17.2 
 WV15707 0.7 (−0.25 to 0.26) 0.04 
 WV15730 0.0 (−6.5 to 6.5) 0.6 
 WV15812 −1.0 (−4.0 to 2.1) 3.0 
 WV15819 0.3 (−1.3 to 1.9) 10.8 
I2 = 0.0%  Summary effect: 0.1 (−0.5 to 0.6) 100.0 
Hospitalization for sepsis, respiratory complication, or dehydration 
 M76001 0.2 (−0.6 to 1.0) 35.3 
 Nicholson et al.2 −0.2 (−1.1 to 0.7) 23.7 
 Treanor et al.3 0.2 (−0.6 to 1.1) 25.6 
 WV15730 0.0 (−6.5 to 6.5) 0.5 
 WV15812 −1.0 (−3.7 to 1.8) 2.7 
 WV15819 −0.5 (−1.8 to 0.8) 12.1 
I2 = 0.0%  Summary effect: 0.0 (−0.5 to 0.4) 100.0 
Pneumonia 
 M76001 −0.8 (−0.2 to 0.3) 40.5 
 Nicholson et al.2 0.0 (−1.0 to 1.0) 46.8 
 WV15707 −11.1 (−34.5 to 12.3) 0.2 
 WV15812 −2.0 (−4.7 to 0.8) 12.5 
I2 = 31.0%  Summary effect: −0.6 (−1.7 to 0.4) 100.0 
INTENTION-TO-TREAT INFECTED POPULATIONc 
Pneumonia   
 M76001 −1.7 (−3.1 to −0.2) 32.8 
 Nicholson et al.2 −0.6 (−2.2 to 0.9) 30.2 
 Treanor et al.3 −0.8 (−2.7 to 1.1) 19.4 
 WV15730 −5.3 (−18.7 to 8.2) 0.4 
 WV15812 0.2 (−2.9 to 3.3) 7.4 
 WV15819 −0.1 (−2.8 to 2.6) 9.8 
I2 = 0.0%  Summary effect: −0.9 (−1.7 to −0.1) 100.0 
Complications requiring an antibioticd 
 M76001 −1.7 (−5.2 to 1.8) 37.9 
 Nicholson et al.2 −2.7 (−6.5 to 1.0) 33.2 
 Treanor et al.3 −6.0 (−12.0 to 0.1) 12.8 
 WV15730 5.3 (−11.8 to 22.3) 1.6 
 WV15812 −0.7 (−10.8 to 9.2) 4.6 
 WV15819 −4.8 (−11.7 to 2.0) 9.9 
I2 = 0.0%  Summary effect: −2.8 (−4.9 to −0.6) 100.0 
Complications requiring an antibiotic, excluding acute bronchitis 
 M76001 0.7 (−2.0 to 3.5) 34.5 
 Nicholson et al.2 −2.1 (−5.4 to 1.1) 24.9 
 Treanor et al.3 −2.2 (−6.7 to 2.3) 12.9 
 WV15730 5.3 (−11.8 to 22.3) 0.9 
 WV15812 3.3 (−3.3 to 9.9) 6.0 
 WV15819 0.9 (−2.6 to 4.5) 20.8 
I2 = 0.0%  Summary effect: −0.1 (−1.7 to 1.5) 100.0 

aA positive risk difference means that there were more complications in the treatment group, while a negative value means that there were fewer complications in the treatment group.

bIntention-to-treat population.

cIntention-to-treat infected population.

dDefined by original study investigators as pneumonia, acute bronchitis, otitis media and sinusitis.

Pneumonia was less common among patients receiving oseltamivir than placebo in the ITTI population (18/1092 placebo versus 9/1654 oseltamivir, RD −0.9%, 95% CI −1.7% to −0.1%, number needed to treat [NNT] = 111). However, a significant reduction in the likelihood of pneumonia was not observed among patients in the ITT population, a group that is more representative of the type of patients treated with oseltamivir in clinical practice (15/1694 placebo versus 10/2623 oseltamivir, RD −0.6%, 95% CI −1.7% to 0.4%).

The authors of the original studies considered otitis media, sinusitis, pneumonia and bronchitis to be complications requiring an antibiotic. The composite outcome of any of these complications was significantly less common among patients receiving oseltamivir than placebo in the ITTI population (RD −2.8%, 95% CI −4.9% to −0.6%, NNT = 36). However, if one excludes acute bronchitis as an outcome requiring antibiotics, there is no difference between groups in the likelihood of the combined outcome of otitis media, sinusitis or pneumonia (RD −0.1%, 95% CI −1.7% to 1.5%). Data were not reported for these outcomes in the ITT population, where presumably there would be even less benefit. Only one death was reported in any of the studies. It occurred in a placebo group patient in study WV15812, which recruited patients with a chronic cardiac or respiratory condition.

Time to treatment

To date, we could only identify a single trial that stratified patients by the time from onset of symptoms. Study MV76001 was entitled ‘A randomized, double-blind, placebo-controlled, multicenter study of efficacy based on the time to treatment of influenza infection with the neuraminidase inhibitor Ro 64-0796.’ The report described the mean duration of symptoms for patients with treatment initiation within 24 hours of the onset of symptoms as well as for the entire group that presented between 0 and 36 hours of onset. While data for patients presenting between 24 and 36 hours after the onset of symptoms were not directly reported, they could be calculated from the above information. The reduction in symptoms was 28.8 hours for patients presenting within 24 hours, compared with 14.8 hours for those presenting between 24 and 36 hours.

Discussion

A number of previous systematic reviews have attempted to address the question of whether oseltamivir provides a clinically meaningful benefit to patients. One systematic review received access to the unpublished data by the manufacturer (Hoffman-LaRoche), and performed simple pooling to conclude that oseltamivir reduces the likelihood of lower respiratory tract complications requiring antibiotics.5 However, simple pooling is likely to be biased by differences in the risk of complications and does not represent a statistically valid approach to meta-analysis. In addition, five of the six authors were employees or paid consultants of the manufacturer. This type of direct involvement by the sponsor has been shown to increase the likelihood of bias and the reporting of results favourable to the manufacturer.21 A second systematic review was sponsored by the manufacturer, and the senior author was one of its employees. This study also reported the results of a pooled analysis of data from published and unpublished trials and found a 24-hour reduction in the median duration of symptoms.22

A third systematic review did not use unpublished data, combined results for oseltamivir and zanamivir, and concluded that overall complications were reduced.23 This review did not exclude non-blinded studies. A fourth systematic review found a modest but statistically significant reduction in the duration of symptoms but not in the likelihood of complications. However, this review did not include the unpublished studies, including studies of patients with co-morbidities and older patients.1 Finally, the fifth systematic review was commissioned by the manufacturer, although they did not provide financial support to the investigators.6 This brief report included bronchitis as a complication requiring an antibiotic, which is not consistent with the best available evidence that acute bronchitis does not require an antibiotic,17–19 and did not report data regarding hospitalizations.

Our study is unique in that we used all available data, including published and unpublished studies; we excluded open-label (unblinded) studies; we did not include acute bronchitis as a complication requiring an antibiotic; we report data regarding hospitalizations and complications; and we received no support from and have no relationship with the manufacturer. We feel that this approach provides the most comprehensive and objective assessment possible.

We found that oseltamivir provides only a small reduction in the duration of symptoms, approximately 21 hours for patients in the ITT population presenting a mean of 24 hours after the onset of symptoms. This is lower than the published estimates of an approximately 30 hour reduction based on the ITTI populations.2,3 There was a non-significant trend towards a smaller reduction in the duration of symptoms for unpublished studies compared with published studies (18.4 versus 31.0 hours, P = 0.12). In addition, two unpublished studies found no statistically significant reduction in duration of symptoms in elderly (WV15812) and chronically ill (WV15819) patients.

We found no evidence from the ITT population that oseltamivir reduces the likelihood of any hospitalization or hospitalization due to respiratory complications, sepsis or dehydration. While there was a small reduction in the number of patients diagnosed with a complication requiring an antibiotic, this benefit disappeared when we excluded patients diagnosed with acute bronchitis, a complication that does not require use of an antibiotic based on the best available evidence.17–19 There was a small reduction in the likelihood of pneumonia in the patients with PCR-confirmed influenza (NNT = 111), but this reduction was not statistically significant in the ITT population. This modest benefit must be balanced against the drug’s cost of over $100 for a 5-day course (http://www.drugstore.com), clinically significant adverse effects such as nausea and vomiting in at least 10% of patients,2 and concerns about drug resistance with increased use.

We are concerned about the failure to publish the results of large, adequately powered RCTs in peer-reviewed journals. We debated whether to include these unpublished studies in our meta-analysis, but because they were generally well designed, because we had access to the final reports of the manufacturer, and because we felt they deserved to be made more publicly available, we decided to include them. Most importantly, the patients who agreed to participate in a study using an experimental drug deserve that the results of those studies reach the broadest audience possible.

Physicians widely believe that oseltamivir prevents hospitalizations and complications, particularly in more vulnerable patients such as the elderly and patients with chronic cardiac or respiratory co-morbidities. However, we found no evidence that oseltamivir reduces the likelihood of either of these important clinical outcomes. In fact, study WV15812, which included 401 patients with at least one chronic cardiac or respiratory condition, found no reduction in the mean duration of symptoms; neither did study WV15819, which included 733 elderly patients. Further studies are needed in vulnerable populations to determine whether or not these patients actually derive clinically meaningful benefit from oseltamivir.

Limitations of our study include limited reporting of complications in the ITT population. In addition, the existing literature includes only a small number of hospitalizations and serious complications, which limited the power of comparisons between groups regarding these outcomes. We were not able to include data for at least one trial (WV16277) because it was only incompletely reported on the manufacturer’s clinical trials database. Finally, we feel that the use of unpublished was justified for the reasons enumerated above.

In summary, oseltamivir reduces the duration of symptom among patients in the ITT population by approximately 21 hours. There is no evidence that it reduces the likelihood of hospitalization or complications requiring antibiotics in the ITT or ITTI populations, and only a slight reduction in the risk of pneumonia in the ITTI population.

Declaration

Funding: MC and JS were supported by graduate student funding from the Institute for Evidence-Based Health Professions Education at the University of Georgia (http://ebp.uga.edu).

Ethical approval: The study was reviewed and approved by the Human Subjects Office of the University of Georgia.

Conflict of interest: none.

Acknowledgements

The authors would like to thank Dr Ming Zhang, Dr Ryusuke Sawada and Ms Svetlana Chirina for their assistance with the review of manuscripts in Chinese, Japanese and Russian languages, respectively.

References

1
Jefferson
T
Jones
M
Doshi
P
Del Mar
C
.
Neuraminidase inhibitors for preventing and treating influenza in healthy adults: systematic review and meta-analysis
.
BMJ
 
2009
;
339
:
b5106
.
2
Nicholson
KG
Aoki
FY
Osterhaus
ADME
et al
Efficacy and safety of oseltamivir in treatment of acute influenza: a randomised controlled trial
.
Lancet
 
2000
;
355
:
1845
50
.
3
Treanor
JJ
Hayden
FG
Vrooman
PS
et al
Efficacy and safety of the oral neuraminidase inhibitor oseltamivir in treating acute influenza: a randomized controlled trial
.
JAMA
 
2000
;
283
:
1016
24
.
4
Linder
JA
Nieva
HR
Blumentals
WA
.
Antiviral and antibiotic prescribing for influenza in primary care
.
J Gen Intern Med
 
2009
;
24
:
504
10
.
5
Kaiser
L
Wat
C
Mills
T
et al
Impact of oseltamivir treatment on influenza-related lower respiratory tract complications and hospitalizations
.
Arch Intern Med
 
2003
;
163
:
1667
72
.
6
Hernan
MA
Lipsitch
M
.
Oseltamivir and risk of lower respiratory tract complications in patients with flu symptoms: a meta-analysis of eleven randomized clinical trials
.
Clin Infect Dis
 
2011
;
53
:
277
9
.
7
Kashiwagi
S
Kudoh
S
Watanabe
A
Yoshimura
I
.
Clinical efficacy and safety of the selective oral neuraminidase inhibitor oseltamivir in treating acute influenza--placebo-controlled double-blind multicenter phase III trial [in Japanese]
.
Kansenshogaku Zasshi
 
2000
;
74
:
1044
61
.
8
Kolubukhina
V
.
Efficacy of ozeltamivir (Tamiflu) in adult influenza on the epidemic rise of morbidity in Russia in the 2006–2007 season [article in Russian]
.
Voprosy virusologii
 
2008
;
53
(
4
):
23
26
.
9
Deng
WW
Li
QY
Zhong
NS
.
“Oseltamivir in the treatment of suspected influenza patients” Study Group
.
A multicenter study of efficacy and safety of oseltamivir in the treatment of suspected influenza patients [in Chinese]
.
Zhonghua Yi Xue Za Zhi
 
2004
;
84
:
2132
6
.
10
Lin
JT
Yu
XZ
Cui
DJ
et al
A multicentre, randomized, controlled trial of oseltamivir in the treatment of influenza in a high-risk Chinese population
.
Curr Med Res Opin
 
2006
;
22
(
1
):
75
82
.
11
Li
L
.
A multicenter study of efficacy and safety of oseltamivir in treatment of naturally acquired influenza [in Chinese]
.
Chin J Int Med
 
2001
;
40
:
838
42
.
12
Li
L
Cai
B
Wang
M
Zhu
Y
.
A double-blind, randomized, placebo-controlled multicenter study of oseltamivir phosphate for treatment of influenza infection in China
.
Chin Med J
 
2003
;
116
:
44
8
.
13
Hopewell
S
McDonald
S
Clarke
M
Egger
M
.
Grey literature in meta-analyses of randomized trials of health care interventions
.
Cochrane Database of Systematic Reviews
 
2007
, Issue
2
. Art. No.:
MR000010
.
14
Blumenthal
D
Campbell
EG
Anderson
MS
Causino
N
Louis
KS
.
Withholding research results in academic life science: evidence from a national survey of faculty
.
JAMA
 
1997
;
277
:
1224
8
.
15
Bodenheimer
T
.
Uneasy alliance: clinical investigators and the pharmaceutical industry
.
N Engl J Med
 
2000
;
342
:
1539
44
.
16
Egger
M
Juni
P
Bartlett
C
et al
How important are comprehensive literature searches and assessment of trial quality in systematic reviews? Empirical study
.
Health Technol Assesst
 
2003
;
7
:
1
76
.
17
Gonzales
R
Bartlett
JG
Besser
RE
et al
for the American Academy of Family Physicians, American College of Physicians–American Society of Internal Medicine, Centers for Disease Control and Prevention, Infectious Diseases Society of America
.
Principles of appropriate antibiotic use for treatment of uncomplicated acute bronchitis: background
.
Ann Intern Med
 
2001
;
134
:
521
9
.
18
Snow
V
Mottur-Pilson
C
Gonzales
R
.
Principles of appropriate antibiotic use for treatment of acute bronchitis in adults
.
Ann Intern Med
 
2001
;
134
:
518
20
.
19
Evans
AT
Husain
S
Durairaj
L
et al
Azithromycin for acute bronchitis: a randomised, double-blind, controlled trial
.
Lancet
 
2002
;
359
:
1648
54
.
20
Jadad
AR
Moore
RA
Carroll
D
et al
Assessing the quality of reports of randomized clinical trials: is blinding necessary?
Control Clin Trials
 
1996
;
17
:
1
12
.
21
Lexchin
J
Bero
LA
Djulbegovic
B
Clark
O
.
Pharmaceutical industry sponsorship and research outcome and quality: systematic review
.
BMJ
 
2003
;
326
:
1167
70
.
22
Singh
S
Barghoorn
J
Bagdonas
A
et al
Clinical benefits with oseltamivir in treating influenza in adult populations: results of a pooled and subgroup analysis
.
Clin Drug Investig
 
2003
;
23
:
561
9
.
23
Falagas
ME
Koletsi
PK
Vouloumanou
EK
et al
Effectiveness and safety of neuraminidase inhibitors in reducing influenza complications: a meta-analysis of randomized controlled trials
.
J Antimicrob Chemother
 
2010
;
65
:
1330
46
.