A, Time from treatment start (combined intranasal and inhaled zanamivir vs placebo) to antibiotic prescription for a respiratory tract complication in 1134 patients with laboratory-confirmed influenza illness. Placebo recipients, n = 447; zanamivir recipients, n = 687. B, Time from treatment start (inhaled zanamivir vs placebo) to antibiotic prescription for a respiratory tract complication in 1572 patients with laboratory-confirmed influenza illness. Placebo recipients, n = 765; zanamivir recipients, n = 807.
Kaiser L, Keene ON, Hammond JMJ, Elliott M, Hayden FG. Impact of Zanamivir on Antibiotic Use for Respiratory Events Following Acute Influenza in Adolescents and Adults. Arch Intern Med. 2000;160(21):3234-3240. doi:10.1001/archinte.160.21.3234
Influenza infections commonly lead to respiratory tract complications that result in antibiotic treatment.
To determine frequency of respiratory events leading to antibiotic use following influenza illness in adolescents and adults, and to assess whether treatment with topical zanamivir prevents these complications.
Meta-analysis of 7 randomized, double-blind, placebo-controlled trials; 3815 mainly healthy adolescents and adults (mean age, 34 years) with an influenzalike illness of less than 2 days' duration were randomly assigned to receive combined inhaled and intranasal zanamivir, inhaled zanamivir, or corresponding placebos. Twelve percent of enrolled subjects were high-risk patients. The main outcome was the incidence of respiratory events leading to antibiotic prescriptions in patients with proven influenza.
Influenza infections were laboratory confirmed in 2499 (66%) of 3815 patients (influenza A in 88% and B in 12%). Placebo recipients developed a respiratory event leading to antibiotic use in 17% of cases, mainly for acute bronchitis or acute sinusitis. Among zanamivir-treated patients (n = 1494%) the incidence of respiratory events leading to the use of antimicrobials was 11% (relative risk [RR] compared with placebo, 0.69; 95% confidence interval [CI], 0.57-0.84). Intranasal and inhaled zanamivir seemed to reduce the number of upper (RR, 0.59; 95% CI, 0.36-0.97) and lower respiratory tract events (RR, 0.64; 95% CI, 0.38-1.08). Inhaled zanamivir reduced the number of lower respiratory tract events (RR, 0.60; 95% CI, 0.42-0.85), but the reduction in the number of upper respiratory tract events was not statistically significant (RR, 0.90; 95% CI, 0.63-1.27).
Respiratory complications or worsening of symptoms leading to antibiotic use occurred in about 17% of adolescents or adults with influenza infection. Early treatment of influenza illness with zanamivir reduced the number of these antibiotic prescriptions.
UPPER AND lower respiratory tract events leading to antibiotic prescriptions are common complications of influenza virus infection.1 Influenza leads to alteration of nasal patency, eustachian tube dysfunction, middle ear pressure abnormalities, and accumulation of fluids in the middle ear and/or paranasal sinuses, events that promote secondary bacterial infections. The causal role of influenza virus infection in development of otitis media or acute sinusitis has been demonstrated in experimental infection of animals and humans.2- 4 In addition, lower respiratory complications, particularly bronchitis and pneumonia, are well-recognized consequences of influenza illness. This association has been confirmed again in recent reports involving elderly persons5- 7 and younger adults.8 However, the type and frequency of respiratory complications and associated antibiotic use following proven influenza illness in adults living in the community has not been prospectively assessed in large populations. Furthermore, it is unknown whether treatment of influenza A illness with the M2 inhibitors amantadine and rimantadine reduces complications.9
Zanamivir is an influenza virus–specific neuraminidase inhibitor. When delivered by inhalation with or without intranasal administration, it is effective in the treatment of influenza illness in adults.10- 12 In some of these trials, a lower rate of antibiotic prescriptions occurred among zanamivir recipients than among placebo recipients.10 To determine whether early treatment of influenza illness with zanamivir could prevent respiratory tract complications or worsening of symptoms leading to antibiotic prescriptions, we analyzed the incidence of such events in available clinical trials that tested the efficacy of zanamivir in the treatment of influenza in mainly healthy adolescents and adults residing in the community. In addition, we analyzed the relationship between route of administration and the types of complications.
All phase II and phase III studies testing the efficacy of combined inhaled and intranasal zanamivir or inhaled zanamivir alone in the treatment of community-acquired influenza illness were considered. Our analysis included only double-blind, placebo-controlled, randomized studies performed before or during the 1997-1998 winter season and in which incidence of antibiotic prescriptions was monitored. A total of 11 phase II and/or III studies for treatment of influenza were performed; 2 pilot studies using nebulized zanamivir and 2 studies performed exclusively in Japan with different designs and databases were excluded from our analysis. Thus, 4 phase II studies11- 13 and 3 phase III studies10,14,15 representing 93% of patients included in all phase II and III studies met our inclusion criteria and were included in this analysis.
Patients enrolled in these studies were at least 12 years old and had an influenzalike illness with symptom duration shorter than 2 calendar days during periods of influenza virus circulation in the community. Subjects enrolled were mainly healthy adolescents and adults; only 12% were considered high-risk subjects (defined as subjects 65 years or older or those with a chronic illness, including cardiopulmonary conditions and diabetes). An influenzalike illness was defined by the presence of fever or feverishness with at least 2 additional symptoms (cough, sore throat, headache, or myalgia). Three studies required that patients have a temperature of 37.8°C or higher (≥37.2°C for patients 65 years or older). For other studies, the subjective experience of feverishness was sufficient. At entry, a medical history was obtained, a physical examination was performed, and patients recorded their symptoms on a diary card. Influenza infection was documented by viral culture, antigen detection, and/or reverse transcription polymerase chain reaction for viral RNA on nasopharyngeal specimens and paired acute and convalescent serum hemagglutination-inhibition antibody assays. A positive result on at least 1 test was considered a laboratory-confirmed influenza illness. Patients were required to return for a posttreatment visit 5 days after starting treatment and for an end-of-study visit after 21 or 28 days. In all studies, investigators collected data prospectively on antibiotic use. No specific instructions were given for antibiotic indications or type, and antibiotics were prescribed when deemed necessary by investigators. For the 3 phase III trials and 2 of the phase II trials, antibiotic use for complications of influenza was prospectively defined as a secondary end point. For 1 trial, the investigators specifically recorded whether the patient was prescribed an antibiotic for an influenza complication. For the remaining trial, the indications for use of all antibiotics were reviewed and classified as influenza complications or other illness by a physician unaware of treatment assignments.
Treatment in phase II trials consisted of combined intranasal (6.4 mg, 2 or 4 times daily) and inhaled zanamivir (10 mg, 2 or 4 times daily) or inhaled zanamivir alone for 5 days. To maintain the blinding in these trials, placebo and zanamivir recipients had both intranasal sprays and drug powder in inhalation. Patients enrolled in the treatment groups in phase III trials received 10 mg of inhaled zanamivir twice daily for 5 days or corresponding placebo. The respective numbers of placebo recipients and zanamivir-treated patients enrolled in each study (as well as the route of administration) are given in Table 1.
The primary end point for this analysis was the occurrence of at least 1 of the following events during the follow-up period in a patient with a laboratory-confirmed influenza illness: (1) any upper or lower respiratory tract event leading to an antibiotic prescription; (2) worsening of initial symptoms leading to antibiotic prescription; or (3) an antibiotic prescribed to prevent bacterial complications. For the purpose of this analysis respiratory events were categorized under blinded conditions based on the information available in the case report forms and the recorded adverse experience reports. Upper respiratory events were defined as acute sinusitis (including complications of sinusitis and purulent and persistent rhinitis), pharyngitis (including streptococcal pharyngitis, tonsillitis, and pharyngeal abscess) or ear infections (including otitis media and mastoiditis). Lower respiratory events were defined as acute bronchitis (including tracheobronchitis, productive cough with colored sputum, or persistent cough), or pneumonia (including bronchopneumonia). When patients experienced 2 or more events, the events were included in the tabulations of complication type.
The incidence of the first respiratory event leading to antibiotic use was compared between treatments using a Mantel-Haenszel test, with the analysis stratified by study.16 For patients experiencing more than 1 event, only the first event was considered in this analysis. Statistical tests were performed at the 2-sided 5% level of significance, and corresponding estimates of relative risks and 95% confidence intervals (CIs) were also calculated. Time to first complication was also compared using a log-rank test.
The effect of inhaled and intranasal zanamivir treatment was compared with placebo across trials that included a treatment arm of combined inhaled and intranasal drug. Similarly, inhaled zanamivir alone was compared with placebo in trials that included a treatment arm of inhaled zanamivir alone.
Interaction tests were performed to assess whether the efficacy of zanamivir varied across subgroups. These tests compared all placebo patients with all zanamivir patients using an exact test for homogeneity of odds ratios (ORs%) across strata.17
Seven clinical trials enrolling 3815 patients met our inclusion criteria and were analyzed. These studies were performed during the fall and winter seasons between 1994 and 1998 in North America, Europe, or in the Southern Hemisphere. Six of the 7 trials included a treatment arm of inhaled zanamivir, while 4 trials included a treatment arm of combined inhaled and intranasal drug. Three of the 7 trials included both of these treatment arms (Table 1). Among the 3815 patients enrolled, 2499 (66%) had a laboratory-confirmed influenza infection; 687 (27%) were randomized to receive inhaled and intranasal zanamivir; 807 (32%), inhaled zanamivir; and 1005 (40%), placebo. The mean age of patients was 34 years; 50% were male, and 12% were considered high-risk patients (suffering from a chronic illness or older than 65 years). The mean duration of influenza illness from the onset of symptoms to enrollment was 29 hours.
Among the 2499 patients with laboratory-confirmed influenza illness, respiratory events leading to antibiotic use occurred in 172 (17%) of 1005 placebo recipients. Twenty-five percent of these patients experienced their first events within 3 days, 50% within 5 days, and 75% within 10 days. Of the placebo recipients who presented a respiratory event leading to antibiotic precriptions, 43% had acute bronchitis, 25% acute sinusitis, 10% pharyngitis, 9% ear infections, and 8% pneumonia. An additional 10% of patients received antibiotics for worsening of initial symptoms (without a specific diagnosis%) or ostensibly to prevent bacterial complications.
Respiratory events leading to antibiotic use occurred in 167 (11%) of 1494 zanamivir recipients with confirmed influenza illness (RR compared with 17% placebo (0.69; 95% CI, 0.57-0.84; P<.001). This difference corresponds to a 31% reduction of antibiotic prescriptions in zanamivir recipients. Among the 1316 patients with influenzalike illness, but without confirmed influenza infection, the incidence of respiratory events leading to antibiotic prescriptions was 15% in both placebo (80/519) and zanamivir (120/797) recipients (RR, 1.0; 95% CI, 0.77-1.30; P>.9). The effect in zanamivir recipients was significantly associated with the presence of a laboratory-confirmed influenza infection (test of interaction between the influenza-positive and influenza-negative population, P = .02).
Among 339 patients with confirmed influenza and respiratory events leading to antibiotic prescriptions, 8 (2.4%) were hospitalized after initiation of study drug treatment. Five of these 8 hospitalizations were related to a respiratory event: pneumonia in 4 cases, and an acute sinusitis with pleuritis in 1. Four of these 5 patients were placebo recipients and 1 received inhaled zanamivir. The other 3 hospitalizations were for non–influenza-related events (appendicitis in 2 cases, postantibiotic colitis in 1).
In the 4 trials that included combined inhaled and intranasal drug therapy, the incidence of respiratory events leading to antibiotic prescription was 15% in placebo recipients compared with 9% in zanamivir recipients (P = .003) (Figure 1A). In the 6 trials that included a treatment arm of inhaled zanamivir alone, this incidence of complications was 18% in placebo recipients and 13% in zanamivir recipients (P = .006). Corresponding reductions were seen in time to first complication (Figure 1, P = .001 for combined inhaled and intranasal compared with placebo; P = .005 for inhaled alone compared with placebo). Of note, the rise in events around day 6 coincided with the patients' posttreatment visits.
Combined intranasal and inhaled zanamivir seemed to reduce both upper (RR, 0.59; 95% CI, 0.36-0.97) and lower respiratory tract events (RR, 0.64; 95%CI, 0.38-1.08) (Table 2). Inhaled zanamivir treatment reduced lower respiratory tract illness (RR, 0.60; 95% CI, 0.42-0.85), but the reduction in upper respiratory tract events was smaller and not statistically significant (RR 0.90; 95% CI, 0.63-1.27) (Table 3).
We assessed whether the size of the effect was consistent across subgroups (Table 4 and Table 5). The largest differences in efficacy were for influenza subtype and patient sex. However, the P values for interaction tests comparing all zanamivir recipients with corresponding placebo recipients were P = .41 for female vs male patients and P = .72 for influenza A infection vs influenza B. Neither of these values was significant, indicating that in the studied population zanamivir effect was not limited to a subgroup of patients.
Antibiotics prescribed were penicillins in 41% of cases, macrolides in 25%, cephalosporins in 21%, tetracyclines in 9%, sulfonamides in 4%, and other (including quinolones) in 6%. There were no differences in the pattern of specific antibiotic use between zanamivir and placebo recipients.
Our results showed that in a largely healthy population of ambulatory adolescents and adults developing acute influenza, the rate of physician-diagnosed respiratory events leading to antibiotic prescriptions was about 17%. Approximately half of these events occurred within the 6 days after the onset of influenza symptoms and were most commonly acute bronchitis and acute sinusitis. Our results also indicate that in mainly healthy adolescent and adult patients, treatment of laboratory-confirmed influenza illness with either combined intranasal and inhaled or inhaled zanamivir alone reduces respiratory events leading to antibiotic use; antibiotic prescriptions were 28% lower in inhaled zanamivir recipients compared with placebo recipients.
This study represents the first demonstration in a large population of patients with influenza that early antiviral therapy can reduce physician-diagnosed complications and associated antibiotic use. Of note, in our population the mean duration after the onset of symptoms was approximately 29 hours, and all patients were included within 2 days. Therefore, efficacy has only been established when zanamivir is administered early in the course of the disease. Also of note, only a minority (12%) of our subjects had comorbidities or were older than 65 years; further studies are necessary in such high-risk populations.
The beneficial effect of inhaled zanamivir was observed mainly in patients presenting with symptoms consistent with lower respiratory tract events. This suggests that zanamivir effect was mainly directed to the lower respiratory tract. The addition of intranasal zanamivir treatment to inhaled zanamivir therapy was not associated with additional overall clinical benefit. However, intranasal zanamivir treatment reduces viral replication in the nasopharynx, whereas inhaled zanamivir does not12; and we observed a lower incidence of upper respiratory tract events in subjects receiving intranasal zanamivir.
In previous studies exploring the role of zanamivir given intranasally in experimental human influenza, the incidence of middle ear abnormalities was decreased in zanamivir recipients.18 The rate of otitis media was low in our adult population, and no specific effect on this complication occurred. However, otitis media is a frequent event and the leading cause of antibiotic prescriptions in young children.19 Whether the addition of intranasal zanamivir could be associated with additional benefit should be investigated in appropriate studies comparing intranasal plus inhaled zanamivir to inhaled zanamivir treatment alone in young children.
Our conclusions have some limitations. The diagnosis of respiratory complications was based mainly on clinical examinations, and we did not provide additional bacteriological or radiological data. It is possible that in some cases, symptoms leading to antibiotic prescriptions were attributable to persisting influenza rather than bacterial complications. We cannot precisely state the proportion of respiratory complications of viral origin alone or of bacterial origin alone or in combination with viral. In addition, the conclusions on the effect of route of administration on the different types of respiratory complications (ie, upper or lower respiratory tract events) are limited by the lack of accuracy of the clinical diagnosis. However, these limitations are a common dilemma in daily practice, where additional investigations in patients with respiratory diseases are limited and antibiotics are generally prescribed on the basis of clinical findings only. Of note, our results are also supported by recent trials performed with another neuraminidase inhibitor, oral oseltamivir (or GS4104), in patients with community-acquired influenza.20,21 In these studies, a significant reduction of respiratory complications leading to antibiotic prescriptions also occurred. Another limitation of our study is that we enrolled a limited number of high-risk persons.
It has been well documented that influenza virus infections initiate production of various proinflammatory cytokines, including interferon α, interferon γ, interleukin (IL) 6, and IL-8 . These cytokines are implicated in recruitment of inflammatory cells and contribute to disease expression. Their role in the pathogenesis of secondary complications is not well understood, but in adults experimentally infected with influenza A virus, IL-6, IL-8, and tumor necrosis factor α peak in nasal washes on day 4 or 5 after the initial infection.22 Recent studies found that intravenous zanamivir or oral oseltamivir administration significantly reduces viral replication, proinflammatory cytokine levels in nasal lavages, and illness during experimental human influenza.23,24 Oral rimantadine therapy also decreased virus titers, levels of IL-8, and severity of symptoms.25 These results suggest that early antiviral treatment of influenza infection decreases levels of inflammatory mediators and thus could decrease complications promoted by these mediators. However, no direct studies characterizing the effect of antiviral therapy on such mediators in natural influenza have been reported.
Findings of subgroup analyses do not suggest that the beneficial effect of zanamivir on respiratory events leading to antibiotic use was limited to a subgroup of patients. A trend toward a greater effect was observed in patients with influenza A infection compared with those with influenza B, but the number of patients with influenza B was smaller, which limits our conclusions. The higher incidence of antibiotic prescriptions initially observed in female patients compared with males was not confirmed by further analysis. In trials performed with zanamivir, no sex differences occurred in other measures of clinical efficacy. This finding is likely attributable to chance, although differences in utilization of health care resources, physician diagnosis, or actual risk of complication could account for the difference.
Our study is the first comprehensive survey of influenza complications in previously healthy adolescent and adult patients with laboratory-documented influenza illness. We determined that approximately 1 of 6 of the placebo recipients eventually received an antibiotic prescription, mainly for secondary events such as acute bronchitis or acute sinusitis, and less often for worsening of initial symptoms. Half of these apparent complications occurred within approximately 6 days after the onset of symptoms. Our results showed that early therapy with neuraminidase inhibitors reduces clinical events leading to antibiotic prescriptions following this viral respiratory infection.
Accepted for publication June 6, 2000.
Dr Kaiser is recipient of a grant from the Division of Infectious Diseases, University Hospital, Geneva, Switzerland. Dr Kaiser's current address is Division of Infectious Diseases, University Hospital of Geneva, Rue Micheli-du-Crest, 1211 Geneva 14, Switzerland (e-mail: email@example.com).
Presented in part at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, Calif, September 28, 1999.
Reprints: Frederick G. Hayden, MD, University of Virginia School of Medicine, Department of Internal Medicine, Division of Epidemiology and Virology, Box 800473, Charlottesville, VA 22908.