This page displays the following correspondence:
- BMJ editor in chief Fiona Godlee's October 2012 letter to Roche board member John Bell, plus a follow-up email sent on December 6, reminding him he had yet to respond
- Cochrane Collaboration researcher Tom Jefferson's emails to Roche about the missing data, dating back to September 2009
- A reactive statement issued by Roche to the media about the BMJ's Tamiflu open data campaign. On 8 November Peter Doshi, Tom Jefferson and colleagues wrote to Roche saying it contained misleading statements
- Roche's letter to Professor Chris Del Mar, in his capacity as a Cochrane Collaboration editor. The letter announces the establishment of an advisory board to look at the Tamiflu data. Also included is the Cochrane Collaboration's response, sent on 26 November
In September 2009 Jefferson first asked Roche for the unpublished dataset used in a Roche supported analysis, published in 2003.
Jefferson needed the data by the following month to update the Cochrane Collaboration’s review on neuraminidase inhibitors in healthy adults.
Jefferson’s October deadline passed. Two months later the Cochrane review, published in the BMJ, said that because eight of the 10 randomised controlled trials on which effectiveness claims were based were never published, the evidence could not be relied on.
The review concluded: "Paucity of good data has undermined previous findings for oseltamivir’s prevention of complications from influenza. Independent randomised trials to resolve these uncertainties are needed.
In December 2009 Roche promised to make full study reports on the 10 trials available to doctors and scientists.
In October 2012 BMJ editor in chief Fiona Godlee reminded the company, in a letter to board member John Bell, that Roche had still not made the full clinical study reports available.
All rapid responses
We thank the Cochrane ARI Group for their response to our correspondence and their request for more detail on our Reporting Odds Ratio (ROR) calculations.
RORs and Confidence Intervals (CI) were calculated by using standard formulas outlined in Bate and Evans (1) and Liu et al (2). However, we constrained our variables to the time period beginning with oseltamivir’s FDA approval date and ending with the most recent FAERS data available to us (October 27, 1999 – August 27, 2012).
The ROR is a measure of reporting disproportionality within FAERS. True incidence is difficult to ascertain since accurate prescription volumes are not available globally and most AEs are underreported.
The formulas we used can be found in the attached document.
(1) Bate A, Evans SJ. Quantitative signal detection using spontaneous ADR reporting. Pharmacoepidemiol Drug Saf. 2009; 18(6):427-36
(2) Liu M, et al. Comparative analysis of pharmacovigilance methods in the detection of adverse drug reactions using electronic medical records. J Am Med Inform Assoc. 2013; 20(3):420-6.
Competing interests: KBH is an employee of AdverseEvents, Inc.
EDITOR, we read the report by Hoffman et al with interest.
Earlier this year a member of our group identified a case of confusion in oseltamivir trial WV 15825 which had been published in 2001 (1).
The case had been originally misclassified in the placebo arm, and it was only extensive examination of the relevant clinical study report which brought the error to light and led to the publication of an Erratum and the changing of the statistical significance linking oseltamivir exposure in the elderly with confusion (2).
The episode shows both the importance of accurate classification, the relative instability of the harms evidence base and the absolute need for independent scrutiny based on all the relevant evidence.
On this basis we urge Hoffman et al to share with BMJ readers a detailed account of their methods to construct their Reporting Odds Ratios (RORs). The findings may have major implications for the use of the drug, especially in the presence of continued uncertainty of its benefits.
(1) Peters PH, Gravenstein S, Norwood P, De Bock V, Van Couter A, Gibbens M, et al. Long-term use of oseltamivir for the prophylaxis of influenza in a vaccinated frail older population. J Am Geriatr Soc. 2001 Aug;49(8):1025–31.
(2) Gravenstein S, Peters P. Erratum. JAGS 2013; 61:478.
Competing interests: Drs Jefferson, Doshi and Heheghan are co-recipients of a UK National Institute for Health Research grant to carry out a Cochrane review of neuraminidase inhibitors (http: //www.hta.ac.uk/2352)
Oseltamivir and Neuropsychiatric Adverse Effects in the FDA Adverse Event Reporting System through 2012
Keith B. Hoffman1*, Andrea Demakas1, Colin B. Erdman1, Mo Dimbil1, P. Murali Doraiswamy2
1AdverseEvents, Inc., 2Departments of Psychiatry and Medicine, Duke University Medical Center
*Corresponding Author: Keith B. Hoffman, 230 Center Street, Healdsburg, CA 95448, Email: firstname.lastname@example.org
We provide an update to the on-going controversy regarding neuropsychiatric adverse events (NPAE) linked to Tamiflu (oseltamivir). These were initially noted in Japanese reports (Yokota et al., 2007; Fujiwara et al., 2008) as well as in cases from around the world. Publications from Roche (Blumentals and Song, 2007; Smith and Sacks, 2009; and Toovey et al., 2008 and 2012), as well as analyses of data from a 2007-2010 cohort of matched pairs in the Vaccine Safety Datalink Project (Green, 2013), suggest that there is no evidence, nor plausible mechanism of action, to link oseltamivir with NPAEs. They surmise that the side effects associated with oseltamivir are symptoms of influenza itself.
Members of the Cochrane Collaboration, the Editor of BMJ, and others, however, contend that Roche’s assertions can neither be confirmed nor denied because much of the data from oseltamivir clinical trials remains unpublished and unavailable (http://www.bmj.com/tamiflu/roche and Doshi, 2009). Roche recently agreed to make their data available in a “staggered” fashion but updated side effect data through 2013 are not yet available.
To shed further light, we examined NPAE signals associated with oseltamivir in the US Food and Drug Administration’s Adverse Event Reporting System (FAERS) from October 27, 1999 through August 27, 2012, the most current data available. We used a data mining platform (RxFilterTM) to generate case report counts and reporting ratios for multiple NPAE MedDRA®* categories, including those queried in the Roche papers. The platform is a combination of computer algorithms and in-house data analysis designed to make FAERS data more accessible to healthcare providers, consumers and health management companies (www.AdverseEvents.com Hoffman et al., 2013). The proportionality measure known as the Reporting Odds Ratio (ROR) was calculated using methods as outlined in Bate and Evans, 2009, however, data for numerator and denominator variables were constrained to the period of time beginning with oseltamivir’s FDA approval date and ending with the most recent FAERS data (August 2012). An ROR provides a reporting estimate for a given agent-side effect combination relative to other side effects with that agent and how this compares to other side effect-drug combinations in the database for the chosen time period. High RORs suggest disproportional reporting of a given drug-side effect combination.
Table 1 shows the number of primary suspect cases found for a selection of NPAE-related MedDRA terms for the time period of 1999-2012, with corresponding ROR scores and 95% Confidence Intervals (CI) (for CI method see Liu et al., 2013). MedDRA categories included in the Roche papers are marked with an asterisk. ROR results of note included: “abnormal behaviour” with an ROR of 29.35; “psychiatric and behavioural symptoms not otherwise classified (NEC)” with an ROR of 15.36; “delirium” with an ROR of 13.50; “hallucination” with an ROR of 12.00; “perception disturbances,” with an ROR of 8.41; and “depressed level of consciousness” with an ROR of 4.69. Roughly half of the reports were from Japan (consistent with large usage in Japan) but reports were also observed in US and other countries. Temporal patterns suggest that NPAEs linked to oseltamivir peaked in the 1999-2007 period, but new reports continued through the present. Finally, “life-threatening” and “death” were listed 188 and 81 times, respectively, within case reports in the largest NPAE category searched, “psychiatric disorders (SOC)” (2,527 total primary suspect cases).
FAERS post-marketing data over the past decade show disproportionally elevated reporting of certain NPAEs linked to oseltamivir. While these data support the cautionary warnings added to the drug’s label in several countries, the absolute number of FAERS reports remains small in relation to the millions of people treated with oseltamivir. Thus, potential safety signals are likely to be small and difficult to detect. Post-marketing data are subject to many biases, such as inability to ascertain causality, marked underreporting, masking, amplification, and confounding by comorbidities (Szarfman et al., 2004; FDA, 2013). On the other hand, the marked under-reporting that is common in FAERS may have resulted in ROR being underestimated. Further, it can be difficult to causally link NAPEs to oseltamivir, since a dose response or withdrawal and rechallenge are not usually tested in patients. Despite such limitations, we hope these findings offer an update for clinicians, while awaiting the release of all prospective trial data.
1. Yokota S, Fujita T, Mori M, et al. Epidemiologic survey of influenza-associated complications (I): clinical assessment of symptoms and signs, and medication. J Japan Pediatric Society 2007; 111: 1545-58.
2. Fujiwara F, Ikushima S, Hibi N, et al. An analysis of risk factors of abnormal behavior in two seasons (07, 08) of influenza infection. Presentation at the 40th Annual Meeting of the Japanese Society for Pediatric Infectious Diseases; 2008 Nov 15-16; Nagoya.
3. Blumentals WA, Song X. The safety of oseltamivir in patients with influenza: analysis of healthcare claims data from six influenza seasons. MedGenMed 2007; 9(4):23.
4. Toovey S, et al. Assessment of neuropsychiatric adverse events in influenza patients treated with oseltamivir: a comprehensive review. Drug Saf. 2008; 31(12):1097-114.
5. Doshi, P. Neuraminidase inhibitors - the story behind the Cochrane review. BMJ 2009; 339:b5164.
6. Smith JR, Sacks S. Incidence of neuropsychiatric adverse events in influenza patients treated with oseltamivir or no antiviral treatment. Int J Clin Pract 2009; 63(4):596-605.
7. Toovey S, et al. Post-marketing assessment of neuropsychiatric adverse events in influenza patients treated with oseltamivir: an updated review. Adv Ther. 2012; 29(10):826-48.
8. Greene SK, Li L, Shay DK, et al. Risk of adverse events following oseltamivir treatment in influenza outpatients, Vaccine Safety Datalink Project, 2007-2010. Pharmacoepidemiol Drug Saf. 2013 22(4):335-44.
9. Hoffman KB, Overstreet BM, and Doraiswamy PM. Development of a Drug Safety ePlatform for Physicians, Pharmacists, and Consumers based on Post-Marketing Adverse Events. Drugs and Therapy Studies 2013; (in press)
10. Bate A, Evans SJ. Quantitative signal detection using spontaneous ADR reporting. Pharmacoepidemiol Drug Saf. 2009; 18(6):427-36
11. Liu M, McPeek Hinz ER, Matheny ME, Denny JC, Schildcrout JS, Miller RA, Xu H. Comparative analysis of pharmacovigilance methods in the detection of adverse drug reactions using electronic medical records. J Am Med Inform Assoc. 2013; 20(3):420-6.
12. FDA(http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Surveil...). Accessed June 2013.
13. Szarfman A, Tonning J, Doraiswamy PM. Pharmacovigilance in the 21st century: new systematic tools for an old problem. Pharmatherapy, 2004; 24, 9: 1099-1144.
Competing interests: PMD has received research grants and advisory or speaking fees from several pharmaceutical companies and is a shareholder in AdverseEvents Inc. (AEI). KBH, AD, CBE, and MD are employees of AEI.
On 23 April 2013 Don MacLean, Life Cycle Leader - Tamiflu, contacted Peter Doshi and colleagues about their request to receive the 74 CSRS. Attached is the correspondence and accompanying spreadsheet in pdf format.
Competing interests: No competing interests