TECHNOLOGY REPORT. Issue 13 January Economic Evaluation of Zanamivir for the Treatment of Influenza

Transcription

1 TECHNOLOGY REPORT Issue 13 January 2001 Economic Evaluation of Zanamivir for the Treatment of Influenza

2 Publications can be requested from: CCOHTA Green Valley Crescent Ottawa, Ontario, Canada, K2C 3V4 Tel. (613) Fax. (613) or download from CCOHTA s web site: Cite as: Brady B, McAuley L, Shukla VK. Economic evaluation of zanamivir for the treatment of influenza. Ottawa: Canadian Coordinating Office for Health Technology Assessment; Technology report no 13. Reproduction of this document for non-commercial purposes is permitted provided appropriate credit is given to CCOHTA. Legal Deposit National Library of Canada ISBN (print) ISBN (online) Publications Mail Agreement Number:

3 Canadian Coordinating Office for Health Technology Assessment Economic Evaluation of Zanamivir for the Treatment of Influenza Bruce Brady, M.A., M.Sc. 1 Laura McAuley, M.Sc. 1 Vijay K. Shukla, B.Pharm., Ph.D. 1 January Canadian Coordinating Office for Health Technology Assessment, Ottawa, Ontario

4 REVIEWERS These individuals kindly provided comments on this report. CCOHTA takes sole responsibility for the final form and content. CCOHTA Scientific Advisory Panel Dr. Gina Bravo Sherbrooke University Geriatric Institute Sherbrooke, Quebec Mr. Doug Coyle Loeb health Research Institute Ottawa Civic Hospital Ottawa, Ontario External Reviewers Dr. Allison McGeer Mount Sinai Hospital Toronto, Ontario Mr. Craig Mitton University of Calgary Calgary, Alberta Dr. Laura Park St. Michael s Hospital Toronto, Ontario Dr. Rod Taylor Department of Public Health & Epidemiology University of Birmingham Edgbaston, United Kingdom ACKNOWLEDGEMENTS This report drew heavily upon a document 1,2 commissioned on behalf of the National Institute for Clinical Excellence (NICE) in the United Kingdom, particularly with regard to the systematic review and meta-analysis of clinical trials examining the efficacy of zanamivir for the treatment of influenza. In addition, the present report made use of various aspects of NICE s economic evaluation of the drug and the economic model used in the NICE analysis. CCOHTA is very grateful to the main author of the NICE document, Dr. Amanda Burls (Department of Public Health & Epidemiology, University of Birmingham, Edgbaston, UK), and NICE for their full cooperation in making their document, and additional information, available so that this report could be produced in a timely manner. Ms. Becky Skidmore and Ms. Annie Hall provided expertise in the area of information science. Feedback on this report was obtained from Glaxo SmithKline (GSK), however, the conclusions of the report reflect the opinions of CCOHTA, and not GSK. CCOHTA appreciates the cooperation of GSK in providing this feedback in a shorter time frame than is usually given for industry. i

5 EXECUTIVE SUMMARY Introduction The objective of this report is to assess the cost-effectiveness of zanamivir for treating influenza in adults. Cost-effectiveness is assessed for the general population as well as for those at-risk of developing complications related to influenza. In Canada, the morbidity and mortality due to influenza is substantial, but they are difficult to estimate due to difficulties with data collection and the under-reporting of influenza-related complications. Zanamivir (Relenza ) is one of a new class of antiviral agents called neuraminidase inhibitors. It has been approved by Health Canada for the treatment of influenza in patients 12 years of age and older, who have been symptomatic for no more than 48 hours. Oseltamivir (Tamiflu ) is another neuraminidase inhibitor approved in Canada for the treatment of influenza. The National Institute of Clinical Excellence (NICE) in the United Kingdom commissioned a recently completed systematic review of the clinical efficacy of zanamivir for the treatment of influenza in adults. 1,2 In order for the present report to be available in a timely manner, the NICE report was used as the basis for data on clinical efficacy, and, as such, the analysis here is restricted to zanamivir. The NICE meta-analysis of the results from the clinical trials showed that for the confirmed influenza cases in the zanamivir group compared to the placebo group, there is: a mean reduction in time to alleviation of symptoms of 1.4 days (95% CI, 0.8 to 1.9) for the general population and 1.7 days (95% CI, 0 to 3.4) for the high-risk population; a mean reduction in time to resumption of normal activities of 0.9 days (95% CI, 0.1 to 1.6) for the general population; a mean reduction in antibiotic use of 28% (95% CI, 9% to 43%) in the general population, and 24% in the high-risk population; and no higher prevalence of adverse events. Methods This economic evaluation uses a decision analytic model to derive the results on cost-effectiveness. The perspective is primarily that of a Ministry of Health or other government payer, although a societal perspective is also considered. Evidence on the efficacy and safety of zanamivir is reviewed, and health outcomes, resource use and costs associated with treating influenza using zanamivir are compared with using over-the-counter medications for symptomatic relief. Health outcomes include time to alleviation of symptoms, risk of complications, and time to return to normal activities. Data for disease epidemiology, unit costs and resource use are derived from published studies and databases, and are based on estimates for Canada, where available. Clinical data for a sub-group analysis of high-risk patients includes results from a recently completed trial. Results of the model are presented in terms of incremental cost per symptom day avoided and quality-adjusted life years gained. A number of sensitivities are performed, including reductions in influenza-related hospitalization using the reduction in antibiotic use for the zanamivir group (as shown in the trials) as a proxy. ii

6 Results Base Case results show an incremental cost per quality-adjusted life year (QALY) gained of $195,000 to $235,000 ($194 to $234 per symptom day avoided) when the diagnostic rate is low (e.g. 14%). When the diagnostic rate is higher (e.g. 35%), the Base Case results are $77,000 to $95,000 per QALY ($77 to $93 per symptom day avoided). The results fall below $50,000 if there are significant reductions in hospitalization due to treatment with zanamivir combined with a high rate of diagnostic accuracy. From a societal perspective, most of the societal costs associated with influenza fall outside the healthcare system, i.e. on those who are ill and their caregivers. Discussion Assessing the cost-effective prescribing of zanamivir is complicated by several factors: although treatment with zanamivir has been shown to reduce antibiotic use, there is an absence of evidence that it would reduce serious influenza-related complications resulting in hospitalization; the difficulty in distinguishing influenza from other causes of influenza-like illnesses; the difficulty in identifying which patients are truly at risk of developing complications and could benefit from therapy; and uncertainty about how many patients will present to physicians soon enough after the onset of symptoms so that they are treated with zanamivir within the first 48 hours. In addition, there have been recent warnings about the decline in respiratory function which may occur in those with underlying chronic pulmonary conditions. Given the large number of assumptions and incomplete information on zanamivir efficacy, as well as other factors, this study may not constitute the final word in determining the cost-effectiveness of zanamivir for treating influenza. Further analysis may be warranted when additional data on the efficacy and effectiveness of the drug, particularly in high-risk groups, becomes available. Analysis using other antiviral agents as a comparator would also be useful. Conclusion From the perspective of a government healthcare payer, the evidence suggests that it is not costeffective to prescribe zanamivir for the treatment of influenza in those who are not at risk of influenza-related complications. Zanamivir could be cost-effective in high-risk groups, if the accuracy of diagnosing influenza is relatively high and if significant hospitalizations can be prevented - but the evidence for this is inconclusive at this time. A cause for concern is the likelihood of primary care consultations from those with little risk of complications who would not ordinarily seek treatment. Finally, if zanamivir were to be covered by a drug plan, additional budgetary resources would have to be allocated for the purchase of the drug since it is unlikely that treating high-risk patients with zanamivir will be a cost-saving strategy. iii

7 TABLE OF CONTENTS LIST OF TABLES AND FIGURES...vii ABBREVIATIONS...viii 1 INTRODUCTION Study Objectives Background Influenza in Canada The burden of influenza in Canada: Prevention of Influenza Vaccination: Treatment of Influenza Amantadine: Neuraminidase inhibitors: CLINICAL EFFICACY OF ZANAMIVIR The NICE Systematic Review Time to alleviation of symptoms: Time to loss of fever: Time to resumption of normal activities: Antibiotic use: Hospitalization: Adverse events: Recent Developments Post-marketing adverse events: Compliance: Review of Economic Evaluations Methods Literature Search Methodology Economic Evaluation Overview Model Structure Parameter Values and Assumptions Health outcomes: Diagnostic accuracy: Resource use: Unit costs: Other parameters: Societal resources:...23 v

8 5 Results Expected Costs General population Base Case: High-risk population Base Case: Societal expected cost: Incremental Results General population- incremental results for Base Case: General population - sensitivity analysis High-risk population - incremental results for Base Case: High-risk population sensitivity analysis: Societal perspective incremental results: Summary of Incremental Results DISCUSSION Study Limitations Parameter value uncertainty: Methodological and modeling issues: Issues not addressed in this study include: Risks and Uncertainties Uncertainty about reducing serious complications: Impact on provision of primary care: Estimating aggregate budget impact: Improving diagnostic accuracy: Comparison with other antiviral agents: Ongoing Research CONCLUSION REFERENCES...47 APPENDIX 1: Databases Searched and Strategies...52 APPENDIX 2: Tables and Figures...55 vi

9 LIST OF TABLES AND FIGURES Table 1 Distribution of Laboratory Confirmed Cases of Influenza by Age Group in Canada...3 Table 2 Hospitalization Rate due to Influenza...4 Table 3 Features of Antiviral Agents for Treatment of Influenza...6 Table 4 Previous Economic Evaluations of Zanamivir for Treatment of Influenza...55 Table 5 Parameter Values for Base Cases...56 Table 6 Table 7 Table 8 Utility Values for Health States...17 Recommendations of Prescription Only Medication (POM) for Clinical Diagnosis of Influenza (All Ages)...19 Value of a Day of Productivity...23 Table 9 Mean Expected Costs for Base Cases...25 Table 10 Incremental Results for Base Cases...27 Table 11 Table 12 Table 13 Incremental Cost-Effectiveness Results for General Population...57 Incremental Cost-Effectiveness Results for High-Risk Population...58 Comparison of Incremental Results for Government Payer and Societal Perspectives...33 Table 14 Range of Incremental Cost-Effectiveness Ratios...34 Table 15 Ongoing Treatment Trials of High-Risk Patients using Zanamivir...45 Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Canadian Age-Standardized ILI Rate by Report Week, to Treatment Pathways for Influenza: Zanamivir vs. Standard Treatment...59 Decision Tree Prior to Treatment of Influenza and Influenza-like Illnesses...60 Incremental Cost per QALY vs. Diagnostic Accuracy...61 Incremental Cost per QALY: Base Cases vs. Percentage of Newcomers...62 High-Risk Population: Absolute Percentage Decrease in Hospitalization...63 vii

10 Abbreviations CCOHTA CDC CIHI COPD GP HRQL ICER ILI IPP ITT NICE NIs OTC POM QALY WMD Canadian Coordinating Office for Health Technology Assessment Centers for Disease Control and Prevention (USA) Canadian Institute for Health Information chronic obstructive pulmonary disease general practitioner health-related quality of life incremental cost-effectiveness ratio influenza-like illness influenza positive population intent-to-treat National Institute of Clinical Excellence (United Kingdom) neuraminidase inhibitors over-the-counter prescription only medication quality-adjusted life year weighted median difference viii

11 1 INTRODUCTION 1.1 Study Objectives The objective of this report is to assess the cost-effectiveness of zanamivir for treating influenza in adults. Cost-effectiveness is assessed for the general population as well as for those at-risk of developing complications related to influenza. The National Institute of Clinical Excellence (NICE) in the United Kingdom commissioned a recently completed systematic review of the clinical efficacy of zanamivir for the treatment of influenza in adults. 1,2 In order for the present report to be available in a timely manner, the NICE report was used with their permission as the basis for the background and data on clinical efficacy. 1.2 Background The influenza virus causes an acute respiratory illness affecting both the upper and lower respiratory tracts. The virus spreads quickly through water droplets released from coughing and sneezing. The incubation period is one to three days. Infected people can be contagious before they begin to exhibit symptoms and adults can remain infectious for one week or less (two weeks or less in children). 1 Clinical symptoms include cough, fever, headache, muscle pain and weakness. Acute illness usually persists for three to seven days. Complications of influenza that can become especially serious include pneumonia, exacerbations of chronic respiratory disease and bronchiolitis in children. In addition, a range of non-respiratory symptoms and complications can arise from influenza, including nausea and vomiting, diarrhoea, febrile convulsions, Reye's syndrome, Goodpasture s Syndrome, encephalopathy, transverse myelitis, pericarditis and myocarditis. 1 Influenza can be difficult to diagnose since other illnesses with similar symptoms are caused by a number of other organisms, including adenoviruses, rhinovirus, respiratory syncytial virus, parainfluenza virus and bacterial infections. 1 Health Canada defines these influenza-like illnesses (ILIs) as acute febrile respiratory illness (fever and / or chills) characterized by one or more of the following: cough, sore throat, arthralgia (joint pain), myalgia (muscle pain), or prostration which in the opinion of the attending physician could be due to influenza virus. 3 The majority of people presenting with an ILI do not have influenza (see subsection 4.4.2) and laboratory testing is required for a definitive diagnosis. There are three known serotypes; influenza A, B, and C, each with different antigenic characteristics of the nucleoprotein and matrix antigens. 1 Almost all influenza illness is caused by influenza A and B, with more severe illness caused by influenza A. 4 Two glycoproteins on the influenza virus lipid membrane act as powerful antigens: neuraminidase (N antigen) and hemagglutinin (H antigen). Hemagglutinin facilitates the entry of the virus into cells of the respiratory epithelium, while neuraminidase facilitates the release of newly produced viral particles (virions) from infected cells. 1 The long-term epidemiological success of influenza viruses is primarily due to variations in antigenic properties that take place in these two surface glycoproteins. 1

12 1.3 Influenza in Canada In Canada influenza season runs from October to April or May. There is usually an annual outbreak, which begins abruptly, peaks over a 2 to 3 week period, generally lasts for 2 to 3 months and often subsides almost as rapidly as it began. During average outbreaks, overall attack rates are estimated to be 10% to 20%. 5-7 However, certain susceptible populations such as school children or nursing home residents may experience attack rates of 40% to 50%. 6 The spread of influenza through a community typically causes large increases in medical visits for fever and respiratory problems. Due to the lack of availability of a quick, reliable, and inexpensive diagnostic test, patients presenting at doctor s offices with ILI are not routinely tested for influenza infection. However, depending upon the patients condition and doctor s judgment, respiratory samples may be sent for laboratory testing. According to the information available from FluWatch (a national surveillance program), 14% of the total respiratory samples received by Canadian laboratories during the influenza season tested positive for influenza, peaking at 25% in early January Not all samples are from patients suffering from ILI, and it is not currently possible to determine the proportion of ILI sufferers who are influenza positive The burden of influenza in Canada: Influenza is one of the major causes of mortality and morbidity in Canada. 9 Epidemiological studies consistently find a relationship between the presence of influenza virus in the community and increased health services utilization, increased rates of hospital admission and increased mortality rates. 6 All these factors contribute to the burden of illness. According to the FluWatch data from 1996 to 1999, general practitioner (GP) consultations for ILI accounted for 3% to 6% of the total patient visits during the influenza season. 10 During the peak of influenza season the rate varied from 8% to 12%. Figure 1 shows the Canadian age-standardized ILI rates as a proportion of physician office visits (total number of patients diagnosed with ILI / total number of patients seen by physicians X 1000) for the last four years. Figure 1: Canadian Age-Standardized ILI Rate by Report Week, to Rate per 1,000 patient visits *Note: this figure was reproduced with permission from Health Canada. 2

13 There are, on average, at least seven million office visits annually due to cold and flu symptoms in Ontario, with physician billings totalling over $200 million annually. 11 Influenza also leads to significant societal costs, with the cost falling mainly on those with the illnesses, as well as their employers and caregivers. 12 In the US, influenza is estimated to have a direct cost of US $1 to US $3 billion annually and over US $10 billion when indirect costs are included. 13 Studies for other countries (e.g. Italy, 14 France and Germany 12 ), also indicate that 80% to more than 90% of societal costs are indirect costs, mostly in the form of absenteeism from work and reduced productivity. 15 It is estimated that each influenza case results in an average of 3 to 7 days of work lost, 16 and that half of those with influenza continue to work though probably with lower productivity. 12 Infection with influenza type A is more common than with influenza type B. During the influenza season in Canada, 85% of the influenza viruses isolated in laboratories were identified as influenza type A. 4 Influenza infection is more common in young children under 5 years old and seniors than in other age groups, as shown in Table 1. Table 1: Distribution of Laboratory Confirmed Cases of Influenza by Age Group in Canada. Influenza Season Total Lab Percentage (%) of cases in different age groups Confirmed Cases <5 yr <5-9 yr yr yr yr yr >64 yr Canadian Institute for Health Information (CIHI) data from 1995 to 1998 (Table 2) indicate that influenza and influenza-related pneumonias are responsible for approximately 2200 to 4000 hospitalizations annually, 35% to 50% of which occur in the elderly (65+ years old). 18 The average length of stay in hospital for influenza in the elderly is 6-7 days compared to 4-5 days for the general population. However, the figures in Table 2 relate to hospitalizations in which influenza and influenza-related pneumonias are recorded as the most responsible diagnosis for admission, and likely represent only a minority of actual hospitalizations due to influenza. It is known that undercounting of hospitalizations and mortality due to influenza occurs when influenza and influenza-related pneumonia are considered a less serious diagnosis and are overridden by a more serious one. 19 Estimates of hospitalizations attributed to influenza are given in subsection Increased mortality due to influenza is caused not only by pneumonia but also from coexisting cardiopulmonary conditions and other chronic diseases that can be exacerbated by influenza. 6 Estimates of deaths per year in Canada are attributed to influenza and its complications. 4 However, the quantification of deaths caused by influenza is complicated by the fact that death certificates may fail to list influenza as a primary underlying, or contributory, cause of death because a laboratory diagnosis is not made. 6 3

14 Table 2: Hospitalization Rate due to Influenza Total/Weighted Average All ages: Total flu cases ,745 % of all hospital cases 0.16% 0.09% 0.10% 0.11% 0.11% Average length of stay years: Total flu cases % of all hospital cases 0.26% 0.12% 0.13% 0.14% 0.16% Average length of stay yr cases as % of all flu cases 49.9% 40.0% 39.6% 36.2% 42.2% 65+ yr cases as % of all flu bed days 60.8% Peak week: % of all hospital cases 1.0% 0.36% 0.59% 0.36% % 65+ hospital cases 1.9% 0.69% 1.0% 0.5% Despite hospitalization and the burden of excess deaths, influenza remains predominantly a primary care based problem. According to the 1998/1999 National Population Health Survey, 20 29% of the respondents reported having sore throat, cold or flu in the previous month. When they developed symptoms, most people (61%) treated themselves or ignored their condition (20%). About 20% of those with colds or flu sought medical care (at a clinic, community health centre, physician s office or emergency room), with about half of these seeking medical care when they initially experienced symptoms and the other half seeking care if symptoms persisted. Of those reporting self-care, 72% took over-the-counter (OTC) medication and 61% reduced activities and rested. Further, a study in Ontario found that about 35% of those with cold or flu symptoms saw a healthcare professional. 11 In 1998, the National Advisory Committee on Epidemiology undertook a priority setting exercise in order to establish which communicable diseases should be monitored at the national level. 21 The ranking was based on 10 criteria to establish the importance of each disease for surveillance purposes. Overall, influenza ranked third of the 43 diseases considered, with only HIV and AIDS placing higher. In terms of individual criteria, influenza scored high on potential to drive public health policy, and medium on socio-economic burden defined as global cost per case. 4

15 1.4 Prevention of Influenza Vaccination: Vaccination has traditionally been recommended for high-risk populations and individuals with close contact to high-risk populations. The National Advisory Committee on Immunization 22 recommends that the following people at high risk of developing serious complications from influenza be vaccinated: those age 65 and over; those with chronic cardiac and pulmonary conditions (such as bronchitis, emphysema, cystic fibrosis and asthma) severe enough to require regular medical care; those with chronic conditions such as diabetes mellitus and other metabolic diseases, kidney disease, cancer, immune suppression or deficiency, HIV, anemia and hemoglobinopathy; children and adolescents with conditions that require treatment with acetylsalicylic acid (ASA) therapy for long periods; and those who reside in nursing homes and chronic care facilities. The efficacy of immunization depends on the age and immunocompetence of the vaccinated person, and on the match between viral strains in the vaccine and those of the circulating virus. 23 Vaccination with inactivated parenteral vaccines can reduce the number of serologically confirmed cases of influenza A by 68% (95% CI, 49% to 79%) and the number of clinical cases of ILI by 24% (95% CI, 15% to 32%) Treatment of Influenza The usual recommendation for people with ILI who are otherwise healthy is to rest at home and use OTC medications to relieve symptoms. Until 1999 the antiviral amantadine was the only treatment for influenza and normally only used in certain restricted situations. In 1999 zanamivir and oseltamivir were launched in Canada as other antiviral options for the treatment of influenza. Table 3 below is a summary of the various features of these three antivirals for the treatment of influenza Amantadine: Amantadine is an antiviral agent licensed in Canada for the prevention and treatment of influenza. It acts by blocking the M2 protein ion channel after the virion has invaded the host cell. It interferes with the production of the enzymes involved in viral replication and assembly. 13 A recent Cochrane review 25 showed that amantadine reduced the duration of fever by one day (95% CI, 0.7 to 1.3) when given orally to adults with influenza A within 48 hours of symptom onset. However, it was associated with significant central nervous system and gastrointestinal adverse events. There is concern that resistance to amantadine may emerge. Resistance is easily produced in the laboratory and resistant strains have been recovered after six days of illness in children receiving treatment and in family members receiving post-exposure prophylaxis. 1 However, amantadine is an important adjunct in some situations. 22,23,26-28 The National Advisory Committee on Immunization recommends that amantadine prophylaxis be offered in certain situations (e.g. to high-risk residents when influenza A outbreaks occur in long term care facilities) 5

16 but that it should not replace annual vaccinations. 22 It may also be considered for treatment of those at high risk of complications who are thought to have influenza A. Table 3: Features of Antiviral Agents for Treatment of Influenza Zanamivir Oseltamivir Amantadine Approved for treatment 12 years & over 18 years & over Adults & children Class / mechanism of action Neuraminidase inhibitor Neuraminidase inhibitor Influenza viruses inhibited A & B A & B A only Blocks activity of viral M2 protein Usual adult dosage 10 mg bid x 5 days 75 mg bid x 5 days 100 mg bid or 10 ml bid x 5 days (may vary with age & renal function) Route of administration Oral inhalation (inhaler device) Oral (capsule) Oral (capsule or syrup) Most common adverse effects Upper respiratory tract GI symptoms CNS & GI symptoms symptoms (bronchospasm & reduced FEV in COPD) (nausea & vomiting) Viral resistance None known None known Possible rapid emergence of resistance Average cost (5 days)* $35 $42 $0.81 (syrup) $10.40 (capsule) * Average cost (5 days): zanamivir and oseltamivir are based on the manufacturer s wholesale price; 29 amantadine is based on the Ontario Drug Benefit Formulary price for Symmetrel syrup (10 ml bid) 30 and a listed wholesale price for Symmetrel capsules (100 mg bid). 31 Note that the pharmacy mark-up and dispensing fees are not included in these costs Neuraminidase inhibitors: As mentioned above, neuraminidase aids both in the entry of viral particles into the target cell and the subsequent release of virions. Neuraminidase inhibitors (NIs) are designed to prevent both the uptake and release of influenza A and B virions by blocking the activity of neuraminidase. 1 It is also possible that NIs could prove to be useful for prevention as well as for treatment of infected individuals. 27,32,33 To date, three NIs are in phase III clinical trials or beyond, including: inhaled zanamivir (formerly known as GG167) developed by Glaxo Wellcome and marketed under the trade name Relenza ; oral oseltamivir (formerly known as RO or GS 4104) co-developed by Gilead Sciences Inc and Hoffman La Roche Ltd and marketed under the trade name Tamiflu ; and oral RWJ developed by BioCryst Pharmaceuticals and Johnson & Johnson, and currently undergoing phase III trials in North America. 1 Currently zanamivir and oseltamivir are available in Canada, having been approved for sale in November 1999 and December 1999, respectively. 34 Both products are indicated for the treatment of uncomplicated acute illness due to influenza infection when symptoms have been present for no 6

17 more than 48 hours. 35 Zanamivir can be used in patients 12 years and older. The recommended dose of zanamivir is 10 mg (two inhalations of 5 mg blister) twice daily (approximately 12 hours apart) for 5 days. 35 It is administered as a dry powder using an inhalation device. There is no data to support the safety and efficacy of zanamivir when started more than 48 hours after symptom onset. 35 Oseltamivir is for adult use only (although studies of oseltamivir for the treatment of children have been done, Hoffman LaRoche has not yet made a regulatory submission for this indication). 1 7

18 2 CLINICAL EFFICACY OF ZANAMIVIR 2.1 The NICE Systematic Review In the United Kingdom, the National Institute of Clinical Excellence (NICE) has recently completed a systematic review and meta-analysis of the clinical efficacy of zanamivir for the treatment of influenza in adults. 1 The NICE report considers both high-risk as well as general-population patients, and an update was produced to incorporate the results of a recently completed trial of zanamivir in high-risk patients. For their review, NICE considered all randomized, or quasi-randomized trials of zanamivir versus placebo or other therapy for the treatment of healthy or high-risk adults (at least 75% of participants were over the age of 12 years) with influenza. High-risk was defined as individuals 65 years of age or older; or patients with chronic respiratory disease, including asthma, chronic heart disease, chronic renal disease, diabetes mellitus, or immunosuppression (due to illness or treatment including asplenia or splenic dysfunction). Studies were included if they reported results on one or more of the outcomes of interest: time to alleviation of symptoms; time to become afibrile; time to return to normal daily activities; and secondary illnesses. All data were included regardless of publication status. However, only studies reporting complete or near complete outcomes for all recruited patients were included. The group identified 52 different original studies. Of those, eight met the inclusion criteria and had sufficient data to be used in the systematic review and meta-analysis. The authors report a rigorous methodology for the identification through to the data abstraction of relevant studies. When data of interest could not be found in the published report, the investigators contacted the study authors and Glaxo Wellcome for details. In some cases the report authors imputed the variability of the outcome estimates. All included studies were high quality (i.e. double-blind, small loss to follow-up and analysis on an intention-to-treat basis) randomized controlled trials published between 1994 and Two were North American trials, and another two were North American and European trials. Of the remaining four, two were European, and two were conducted in the Southern Hemisphere. These trials included data from 4136 patients of which 238 were high-risk. All included trials were supported by Glaxo Wellcome. An analysis of the trials by NICE suggested that publication bias was not a concern. A ninth, recently completed trial became available after the initial report had been prepared. This trial included 525 patients considered to be high-risk with asthma or chronic obstructive lung disease. NICE prepared an update of their review for the high-risk population to capture these data. 2 The primary outcome for all trials was time to alleviation of symptoms. Other outcomes that were considered included time to alleviation of fever, time to resumption of normal activities, antibiotic use, and hospitalization. All outcomes were considered for populations on an intent-to-treat (ITT) basis. Outcomes were also reported for the subset of the populations with confirmed influenza (influenza positive population - IPP). The IPP analysis showed more benefit from treatment with zanamivir for all outcomes considered than did the ITT analysis, as would be expected given that 8

19 this is the population for which zanamivir is indicated. Summarized below are NICE s findings for treatment with 10 mg bid inhaled zanamivir (licensed dose) compared with placebo Time to alleviation of symptoms: This outcome was defined as the absence of fever (<37.8 C) and feverishness, mild or no headache, and the absence of cough, muscular pain (myalgia) and sore throat for 24 hours as scored by patient diaries. Results were presented in median number of days due to skewness of the data. NICE reports concern about the use of median values, and the risk they may pose for the exaggeration of small differences. A sample of 1498 influenza positive patients from six studies showed a statistically significant reduction in the time to alleviation of symptoms with zanamivir treatment compared to placebo of 1.4 days (95% CI, 0.8 to 1.9). Although zanamivir was marginally less effective in the ITT analysis, a significant reduction in time to alleviation of symptoms was noted (weighted median difference (WMD) 1.0 days, 95% CI, 0.4 to 1.7). The reduction in the time to alleviation of symptoms was even greater for the influenza positive high-risk population (n= 484) at 1.7 days; however, NICE reports that this reduction was not statistically significant (95% CI, to +3.4). Glaxo Wellcome reports the reduction in time to alleviation of symptoms as significant in the largest trial in this pooled analysis. NICE estimated the variance for this trial (as it was not included in the publication) and found non-significant results. NICE acknowledges that Glaxo Wellcome, who has access to the full data set, may have a more accurate measure of variance. Given this trial accounts for 313 of the 484 patients (65%), the accuracy of the values will impact on the results of the pooled analysis. Without this trial, NICE reports a reduction in time to alleviation of symptoms of 1.9 days (95% CI, -0.5 to +4.3) for the influenza positive high-risk population, though this reduction was not statistically significant Time to loss of fever: In these studies fever was reported to generally last between 2 and 2.5 days. Zanamivir reduced the time to become afebrile by a half day compared to placebo in the IPP (n= 1244) (95% CI, 0.2 to 0.8). A similar trend occurred in the ITT analysis, however the results are not statistically significant. No results are reported for the high-risk population Time to resumption of normal activities: Resumption of normal activities was defined within a trial as the ability of a patient to carry out normal daily activity for two consecutive days. In the IPP (n= 1341), treatment with zanamivir resulted in a statistically significant reduction of 0.9 days (95% CI, 0.1 to 1.6) in return to normal activities in comparison to placebo. For this population, patients were able to resume normal activities in about six rather than about seven days. The results from ITT analysis showed a similar trend, but were not significant (WMD 0.5 days, 95% CI, -0.4 to +1.5). No results are reported for the high-risk population. 9

20 2.1.4 Antibiotic use: For the general population group that were influenza-positive, NICE reports that 4.6% fewer patients treated with zanamivir required antibiotics for any indication compared with placebo (13.1% vs 17.7%). A meta-analysis conducted by Glaxo Wellcome researchers using seven of the included studies indicates that the rate of respiratory events requiring antibiotic use was 18.2% in the placebo group compared to 13.0% in the zanamivir group, indicating that patients treated with zanamivir are 0.72 times as likely to require antibiotic treatment as those treated with placebo (i.e. a relative risk of 0.72, 95% CI, 0.57 to 0.91). For the high-risk sub-group, 70 patients out of 391 (17.9%) treated with zanamivir required antibiotics compared to 97 out of 412 (23.5%) in the placebo group. This indicates that high-risk patients treated with zanamivir are 0.76 times as likely to require antibiotic treatment as those treated with placebo (i.e. a relative risk of 0.76) Hospitalization: Hospitalization rates were extremely low. For the general population, zanamivir treatment had no impact on hospitalization when compared with placebo. Less than 1% of patients were hospitalized in either arm. No results are reported for the high-risk population Adverse events: Adverse events were mainly influenza-like symptoms and were no more prevalent in the zanamivir group than the placebo group. 2.2 Recent Developments Post-marketing adverse events: There have been reports of adverse events associated with the use of zanamivir since the drug has begun to be marketed: The Food and Drug Administration in the USA (January, 2000) and Glaxo Wellcome Canada (July 11, 2000) have advised caution when prescribing zanamivir to patients with underlying asthma or chronic obstructive pulmonary disease (COPD). Reports of deterioration of respiratory function (e.g. bronchospasm) in these patients have been received, and they have advised that administration of the drug should be stopped if this occurs. 34 The Canadian Adverse Drug Reaction Monitoring Program reported that during the 8-month period prior to June 30, 2000, sixteen domestic reports of suspected adverse reactions to zanamivir were received, six of which were classified as serious and unexpected. 34 Of the six reports with serious adverse reactions, one man with a history of cardiomyopathy and taking concomitant medications was admitted to hospital with acute renal failure about two days after taking zanamivir, and died the following day. The five remaining people experiencing serious adverse reactions had varied medical histories and reported different types of reactions. At the time that report was released, two of the five had recovered from 10

21 the reaction, one had not yet recovered, and two had unknown outcomes. However, as yet no proven causal association has been established between using zanamivir and these six patients with serious adverse reactions Compliance: Comments have been made about the potential difficulty some people may have in using the inhaler device, 28,33,36 although evidence for this is limited. However, one study by Lee 37 reported that 26% of residents (some with dementia) in a long-term care facility had at least some difficulty coordinating the use of the inhalation device. Also, a telephone survey of influenza patients treated with zanamivir indicated that about 90% of patients and clinicians found the inhaler device somewhat or very easy to use

22 3 REVIEW OF ECONOMIC EVALUATIONS The report commissioned by NICE 1,2 contained an economic evaluation of both the general and high-risk populations in the UK. In addition, CCOHTA s literature search produced one other economic evaluation that assessed zanamivir for treating influenza, that by Mauskopf et al 39 which assessed the cost-effectiveness of treating a high-risk population in Australia. (Following the completion of this report, another economic evaluation of zanamivir was retrieved - see subsection 6.2.5). 40 The NICE and Mauskopf evaluations followed a similar methodology and framework of analysis. The treatment pathways in their decision analytic models were similar. Both evaluations took the perspective of a government healthcare payer, and used no active intervention (symptomatic relief only with OTC medications) as the comparator. Each study used data on unit costing, resource utilization and practice patterns for their relevant country, where possible. In both studies, resource use items included zanamivir prescription, physician visits, antibacterial use and hospitalization. The impact of adverse reactions associated with zanamivir and the implications of treatment compliance were ignored in both evaluations. There were also some key differences between the two economic evaluations. The NICE evaluation used the meta-analysis of clinical trials as the basis for the efficacy data in their model, whereas the efficacy data in the Mauskopf study was based on sub-group analysis of a small number of high-risk individuals from one trial only (the MIST study). The Mauskopf study uses a 70% prevalence of influenza among ILI patients based on the data in the one trial, whereas the NICE evaluation used figures of 14% and 34% in order to reflect more closely the real clinical situation in the UK. In addition, the Mauskopf study extrapolated the reduction in antibiotic use to a reduction in GP follow-up visits. On the other hand, the NICE evaluation extrapolated the antibiotic reduction to lower hospitalization and death for the high-risk population, but did not do so for the general population. Some of the similarities and differences between the two evaluations are shown in Table 4 (Appendix 2). Comparing the results in Canadian dollars using the exchange rates on December 15, 2000 (Canadian $1.00 = Australia$ = UK 0.446), the NICE evaluation shows the cost per qualityadjusted life year (QALY) for the general population varies from C$85,000 at a higher diagnostic accuracy (34%) to C$353,000 at a low diagnostic accuracy (14%). The ICER for the high-risk population in the NICE evaluation ranges from C$21,000 to C$70,000, depending on assumptions about the diagnostic accuracy, reduction in hospitalization and utility value gain for no influenza. The ICER for the high-risk population in the Mauskopf evaluation is about C$9,600 per QALY (C$12 per symptom day avoided), which is considerably lower than in the NICE evaluation. This large difference between the results in the NICE and Mauskopf evaluations is mainly due to differences in diagnostic accuracy (14% and 34% vs 70%), the cost of zanamivir (C$54 vs C$37), utility gain from not having influenza symptoms (0.28 vs 0.44), and symptom days avoided due to treatment with zanamivir (1.4 days vs 2.5 days). 12

23 4 Methods 4.1 Literature Search Methodology Published literature was obtained by searching a number of databases (see Appendix 1). Two search strategies were developed: one clinical strategy to update the recently published NICE report 41 and another strategy to focus solely on the economic aspects of influenza and influenza treatment in Canada. On the DIALOG system, MEDLINE, EMBASE, HealthSTAR and SciSearch were searched. Database alerts/updates were established for the clinical search on Adis LMS Drug Alerts, Adis Newsletters, Current Contents Search, EMBASE, MEDLINE, Pharmaceutical News Index (PNI ) and SciSearch, and on Current Contents Search, EMBASE and MEDLINE for the economic search. The Cochrane Library on CD-ROM, Issue 4, 2000 was searched for clinical information. Web sites of health technology assessment and near technology assessment agencies were also searched, as were specialized databases, such as the University of York NHS Centre for Reviews and Dissemination. Extensive data were obtained from the Canadian Institute for Health Information (CIHI) Discharge Database for the years 1995 to 1998 regarding hospital admissions and hospitalizations documented with a most responsible diagnosis of influenza. IMS Health provided data from the Canadian Disease and Therapeutic Index database and the Health Statistics Division of Statistics Canada provided detailed data from the 1998/1999 National Population Health Survey. Data from Health Canada s FluWatch project were also acquired. These searches were supplemented by hand searching of selected journals and documents in the CCOHTA library collection and the bibliographies of selected papers. 4.2 Economic Evaluation Overview The economic evaluation uses a decision analytic model that compares the health outcomes, resource consumption and costs associated with treating influenza using zanamivir to symptomatic relief only. The approach and framework of analysis used in this report is similar to that in the studies by NICE and Mauskopf et al. The perspective is primarily that of a Ministry of Health or other government healthcare payer in Canada, although a societal perspective is also analyzed. Two types of economic evaluation are performed: a cost-effectiveness analysis and a cost-utility analysis. Results are reported in terms of incremental cost per symptom day avoided and incremental cost per quality-adjusted life year (QALY) gained. The cost-effectiveness of prescribing zanamivir for the general population is assessed, and a sub-group analysis of the population at-risk of influenzarelated complications is also done. Evidence on the efficacy and safety of zanamivir, for both the general and high-risk populations, was taken from the NICE meta-analysis of trials, as described in section 2. Health outcomes included time to alleviation of symptoms, time to return to normal activities, and antibiotic use. Data for disease epidemiology, resource use and unit costs were based on estimates for Canada, where available. 13

24 Below is a summary of the main features of the evaluation: 1. Analytical perspective primarily that of a government healthcare payer in Canada, so that only the direct medical costs borne within the healthcare system are considered. Less extensive analysis of the societal perspective has also been included. 2. Patients the general population over 12 years old, presenting to a physician with an influenza-like illness (ILI) within the recommended 48 hours of symptom onset. The general population includes a small proportion of high-risk patients. A sub-group analysis of the high-risk population is also performed. It is assumed that patients using zanamivir begin treatment within the 48-hour time frame, as there is no data to support the safety and efficacy of zanamivir when started beyond 48 hours. 35 However, the impact of patients being treated later than 48 hours is examined in sensitivity analysis. It is assumed that patients who are prescribed zanamivir are eligible for full coverage of the drug s cost under a government drug benefit plan. 3. Time horizon about 4 weeks, which takes into consideration the length of the illness. This time period captures all the relevant major clinical and economic outcomes or events (including hospitalization due to complications) and permits the treatment strategies to be fully evaluated. Some of the sensitivities of the high-risk population consider a longer time horizon to capture any deaths related to influenza. 4. Treatment comparator current standard treatment for influenza does not involve active medical intervention in most cases; those with influenza-related complications may be prescribed antibacterials or hospitalized (if serious). Although other antiviral agents, notably amantidine and oseltamivir, are approved in Canada for treatment of influenza, the former has limitations for general use (e.g. no activity against influenza B, problems with sideeffects, risk of developing resistance - see subsection 1.5.1), and is not part of standard clinical care for influenza in primary care practice. (An exception to this is the use of amantadine for treating high-risk patients in long-term care facilities). Also, there have been no head-to-head trials of amantadine, oseltamivir and / or zanamivir. In the analysis of the societal perspective, it is assumed that a proportion of patients purchase OTC medications for symptomatic relief (e.g. antipyretics, decongestants, cough medication). 5. Health outcomes - includes time to alleviation of clinical symptoms, time to return to normal activities, and the risk of complications requiring treatment, as derived from the metaanalysis of the clinical trials (refer to section 2). Influenza-related complications may be treated by prescribed medication (mainly antibacterials) or hospitalization. Changes in health-related quality of life (HRQL), as measured by utilities, are assigned to various health states: no influenza; influenza and not hospitalized; influenza and hospitalized for serious complications. In the sub-group analysis, death is considered in some of the sensitivities. 6. Resource use - from a government payer perspective, resources consumed during treatment include: (i) zanamivir, (ii) physician visits for diagnosis and prescription, (iii) prescribed medications (mainly antibacterials) to treat complications, and (iv) hospitalization to treat more serious complications. The societal perspective also includes direct patient costs and indirect costs to society: (v) the purchase by patients of OTC medications for symptomatic relief, and (vi) productivity losses resulting from absenteeism from work due to influenza. 14