Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Tamiflu: What is the NNT to prevent one flu death?
- A Rouse (9 November 2005)
-
Explaining the wisdom of stockpiling oseltamivir
- James R Smith, Regina Dutkowski (11 November 2005)
-
Oseltamivir: NNT to prevent hospital admissions
- Carl Heneghan, Richard T Mayon-White (15 December 2005)
-
The evidence of Oseltamivir (Tamiflu®) in reducing influenza mortality
- Martin Sprenger, Christoph Pammer (1 March 2006)
|
|
|||
|
A Rouse, consultant in public health heart of birmingham PCT B16 9PA
Send response to journal:
|
I, like Professor Joe Collier, would like to know the basis of Secretary of State for Health Patricia Hewitt’s claim that “…the acquisition of 14 million doses of the antiviral drug oseltamivir (Tamiflu) would reduce the number of (flu) deaths in Britain.” (1) Who has given her this advice? Would they also tell us the NNT? 1. East Asia is most at risk of human flu epidemic, experts say. Michael Day http://bmj.bmjjournals.com/cgi/content/full/331/7522/921/DC1 Competing interests: None declared |
|||
|
|
|||
|
James R Smith, Medical Leader F. Hoffman-La Roche Ltd, Basel, Switzerland CH-4070, Regina Dutkowski
Send response to journal:
|
We would like to correct some misunderstandings recently reported in the BMJ regarding the benefits of oseltamivir (Tamiflu®).1 A question was asked on what data supports an improvement in mortality by oseltamivir. Two datasets were recently presented at the second European Influenza Conference (ESWI) in Malta (www.eswi.org). A large retrospective cohort study of patients with influenza-like illness (n=176,001) taken from a US health database showed that oseltamivir (75 mg twice-daily, n=39,202) significantly reduced the risks of pneumonia by 32% (p<0.001) and of death by 91% (p<0.05) [Nordstrom et al, 2005]. In Canadian patients with laboratory-confirmed influenza (A or B) requiring hospital admission (<15–>64 years; n=359), oseltamivir reduced the risk of death by 68% [McGeer et al, 2005]. Thus, treatment with oseltamivir statistically and meaningfully reduces the risk of death in patients of all ages and from all walks of life, infected with influenza A or B. The article also reported an opinion that ‘oseltamivir does not prevent infection with the flu virus, and that at best it would reduce the severity of illness’. In fact, based on clinical trial data, Tamiflu® was approved by the European Medicines Agency (EMEA) for preventing influenza in adults and adolescents aged ≥13 years. Additionally, a recent large-scale study examined the effectiveness of oseltamivir (75 mg once- daily) in protecting family members who had come into contact with an influenza-infected individual: oseltamivir protected 80% of contacts from influenza infection.2 Furthermore, while oseltamivir has certainly been proven to reduce duration of illness by 3 days, it also reduces the risk of influenza-associated hospitalisations, bronchitis, pneumonia and associated antibiotic use.3 Given its beneficial effects on the treatment and prevention of influenza, the prevention of complications and hospitalisations, and on overall patient mortality, there is definite wisdom in stockpiling oseltamivir as part of governmental pandemic preparedness plans. References 1. Cole A. Experts question wisdom of stockpiling oseltamivir. BMJ 2005; 331: 1041. 2. Hayden FG, Belshe R, Villanueva C, et al. Management of influenza in households: a prospective, randomized comparison of oseltamivir treatment with or without postexposure prophylaxis. J Infect Dis 2004; 189: 440–9. 3. 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. Competing interests: None declared |
|||
|
|
|||
|
Carl Heneghan, Clinical Research Fellow Centre for Evidence Based Medicine, Department of Primary Health Care, University of Oxford, OX3 7LF, Richard T Mayon-White
Send response to journal:
|
EDITOR – The efficacy of oseltamivir in influenza A has been questioned not only by doctors in the BMJ 1, but also by patients consulting their GPs. To some of our patients, the key question was the avoidance of complications requiring hospital admission. We addressed the issues in our departmental evidence-based journal club. The best answer that we could find was in the analysis of 10 trials published by Kaiser et al in 2003, 2 a paper found by a search using PubMed clinical queries using the RCT filter. In otherwise healthy adults under 65 years old with influenza like illness, 10/1418 patients in the oseltamivir treated group required hospital admission as opposed to 7/940 in the placebo group. This results in a non significant Number Needed to Treat (NNT) = 2500. For patients at increased risk of complications with influenza like illness, the absolute risk reduction was 0.8% resulting in a non significant NNT = 85. When the analysis is repeated on confirmed cases of influenza the NNT for otherwise healthy adults translates into a non- significant NNT = 200. In at-risk patients, the absolute risk reduction was 1.6% [95% CI, -0.5 to 3.7%] a non-significant NNT =63 (p value of 0.17, as quoted by Kaiser et al). Significance is achieved when the results are pooled for the overall population with confirmed influenza: as stated by Grove 3 the NNT is = 97 [95% CI 52-726]. Using the critical appraisal techniques that we teach to medical students, we identified several weaknesses in the evidence: the paper pools the analysis of 10 trials conducted with different subjects (4 with people aged 65+ years, 3 with younger adults, 2 with adults with chronic obstructive airways disease, and one with adults of all ages). Randomisation appears to be unbalanced; producing 2023 people in the oseltamivir group and 1541in the placebo group. A meta-analysis would have been the appropriate method to pool the data. The type of influenza was predominantly A H3 N2 and the results might not be applicable to other subtypes. Our conclusion was that oseltamivir had a small impact on hospitalisations, but the evidence was unsatisfactory. We have not found any more recent randomised trials papers that change our conclusion on the efficacy of preventing hospital admission, but through the website www.eswi.org cited by Dr Smith and Dr Dutkowski, 4 we found the restrospective cohort study reported by Nordstorm et al. From this, we calculate a NNT of 297 to prevent one hospital admission. Given our doubts, we subjected the evidence to a second journal club. We wanted to consider other outcomes, such as the reduction of the duration of symptoms and its prophylactic use, by which to judge the efficacy of oseltamivir, which could be useful in reducing sickness absence in key workers. We analyzed the Hayden paper: 5 in terms of appraisal, it is unclear as to whether the randomization was concealed. The impressive 80% protection for flu contacts in the study, translates in all households to an absolute risk reduction of 7.4% [95% CI; 4.0-10.8%]: 40/392 (10.8%) in the expectant treatment arm and 11/400 (2.8%) in the post-exposure prophylaxis arm NNT =14 [95% CI; 9-25]. In such circumstances, the safety of oseltamivir is very important we were concerned that adverse events were greater in the Hayden study for children (20% for twice daily dosing, 10% for once daily dosing) as opposed to the reported 2-8% cited in the product summary 6 In conclusion, as Dr Rouse has commented, 7 we have been surprised by the confidence shown in oseltamivir in the national guidance on pandemic influenza. Policymakers should be more transparent in their use of evidence in their decisions. Competing interests: CH is funded by a DOH fellowship and runs a journal club and has an active interest in critical appraisal. RMW has given talks on influenza without payment. Carukshi Arambepola, David Mant Emma Meats, Megan Goldsmiths, Nia Roberts, Nourieh Hoveyda, Sue Smith, Rafael Perera, Peter Rose,, and Zhuang Wei helped with the critical appraisal, 1. Cole A. Experts question wisdom of stockpiling oseltamivir. BMJ 2005;331:1041.(5 November) 2. Kaiser L, Watts C, Mills T, Mahoney P, Ward P, Hayden F. Impact of oseltamivir treatment on influenza related lower respiratory tract complications and hospitalizations. Arch Intern Med 2003;163:1667-72. 3. Grove M. NNT for oseltamivir. BMJ rapid response bmj.com 25 November 4. Smith JR, Dutkowski. Stockpiling oseltamivir: Roche clarifies data for improved mortality with oseltamivir BMJ 2005;331:1203-b 5. Hayden FG, Belshe R, Villanueva C, Lanno R, Hughes C, Small I, et al. Management of influenza in households: a prospective, randomized comparison of oseltamivir treatment with or without postexposure prophylaxis. J Infect Dis 2004;189:440-9. 6. Tamiflu. Summary of Product Characteristics, UK/Ireland. Roche Ltd, June 2002 7. Rouse A. What is the number needed to treat with oseltamivir to prevent one flu death? BMJ 2005;331:1203 (19 November) Competing interests: None declared |
|||
|
|
|||
|
Martin Sprenger, Public Health coordinator Medical University of Graz / Austria, Christoph Pammer
Send response to journal:
|
The study: Nordstrom BL, Zhu S, Smith JR. Reduction of influenza complications following oseltamivir use [abstract]. Ingenix i3 Magnifi Epidemiology, Auburndale, MA and Hoffman-La Roche Ltd., Basel, Switzerland. ESWI 2005. Abstract 937. The study population was derived from the Ingenix Research Data Mart (RDM), which contains information on over 20 million individuals from 1995 to the present. The database included data both from patients insured by United Healthcare and from large national employer groups with administrative services provided by United Healthcare. RDM includes medical claims data and pharmacy claims data. Using the RDM, patients of all ages during the period of November 1999 to March 2004 clinically diagnosed with influenza were identified and were grouped into two cohorts: those with a pharmacy dispensing of Tamiflu on the day of influenza diagnosis (N=39,202), and those with influenza diagnosis but no antiviral treatment during the influenza season (N=136,799). The total number of patients in the study was 176,001. The outcomes measured included incidences of claims diagnoses of pneumonia or myocardial infarction, or in-hospital deaths identified in claims. All outcomes were considered potentially influenza-related only if they occurred within 30 days after the influenza diagnosis. Cox proportional hazards models stratified into ten-year age blocks provided hazard ratio estimates for the Tamiflu-exposed group compared to the non-exposed cohort. In the observational study of over 176,000 patients of all ages with influenza, the risk of death during the 4-week period after influenza diagnosis was 11-fold lower in the Tamiflu-exposed group (1 death in 39,202 patients [0.003%]) compared with the untreated group (56 deaths in 136,799 patients [0.041%]), with a hazard ratio of 0.09 (95% CI of 0.01, 0.65; p = 0.02). These statistically significant findings clearly show that Tamiflu use is associated with a reduced risk of death. Critical appraisal: As long as the full text of the study is not available our appraisal is limited to the following key questions: 1. What were the key selection (inclusion & exclusion) criteria? Inclusion criteria was the clinical diagnose „influenza“. We don’t know if the diagnostic test was standardised or if laboratory tests have been used. How many cases of influenza-like illness have been included in both groups? 2. Were exposure & comparison groups similar at start of study except for study exposures? Especially in relation to gender mix, age, stage of disease, social background, ethnic origin, co-morbidity, but also vaccination rates and time of death (first, second, third or fourth week after diagnose). 3. What outcome measures were used? The outcomes measured included incidences of claims diagnoses of pneumonia or myocardial infarction, or in-hospital deaths identified in claims. All outcomes were considered potentially influenza-related only if they occurred within 30 days after the influenza diagnosis. As there was just 1 death in the intervention group, it would be interesting to find out more about the cause and time of death of this single case. 4. Was follow up time sufficiently long? A 4-week follow up seems very long. It would be interesting to calculate relative, absolute risks and number needed to treat using the numbers of death in the first, second, third or fourth week after diagnose. 5. What methodological effect has a single case in a cohort study? What would have happened if nobody of the 39,202 persons would have died in the 4-week follow up? Then Tamiflu® would be a magic bullet with a relative and absolute risk reduction of 100%.. Especially in cohort studies it is a real methodological problem to have low numbers, as low numbers are often linked to great variation of the results 6. Absolute risk reduction and number needed to treat (NNT) to avoid one death The absolute risk reduction = 0,041 – 0,003 = 0,038%. The NNT to avoid one death in the 4-week period after influenza diagnosis is 2.632. 7. Eminence versus evidence (1) After the publication of the Cochrane Report in the Lancet on January 19th Roche launched a media release which is representative for the ongoing discussion: “Roche fundamentally disagrees with the conclusions reached by the authors that oseltamivir should not be used for the treatment or prevention of seasonal influenza. The conclusion is at odds with the opinion of experts and regulatory authorities around the world.” 8. Economic evidence: (2) There is good economic evidence that the business with Tamiflu® in the year 2005 was very successful. With Tamiflu® alone Roche earned one billion Euros in the year 2005. (1)Jefferson T, Demicheli V, Rivetti D, Jones M, Di Pietrantonj C,
Rivetti A. Antivirals for influenza in healthy adults: systematic review.
Lancet 2006; 367. Published online. 19.01.2006.
http://www.thelancet.com/journals/eop
(2)F. Hoffmann – La. Roche AG. Media release. Roche 2005: Record sales and operating profit. Basel, 01.02.2006. www.roche.com/mrar05e.pdf accessed 01.03.2006 Competing interests: None declared |
|||