Neuraminidase inhibitors for preventing and treating influenza in healthy adults: systematic review and meta-analysisBMJ 2009; 339 doi: https://doi.org/10.1136/bmj.b5106 (Published 08 December 2009) Cite this as: BMJ 2009;339:b5106
All rapid responses
I make no claims to be experienced in understanding clinical trials,
nor even to have a medical background: I am by training an engineer.
However, it was clear as long ago as June that the use of oseltamivir in
combating the current 'pandemic' A/H1N1 strain was neither
straightforward, nor without an element of risk.
Under the auspices of Godfrey Bloom MEP (Yorkshire & North
Lincolnshire) I undertook an analysis of existing publicly available
information relating to oseltamivir treatments and arrived at conclusions
which, to a layman such as myself, do not differ greatly from those in
Several questions arose from this research which deserved an answer
much earlier in the debate. These included:
whether the widespread use of oseltamivir would result in increased
resistance as appeared to be suggested by de Jong, Thanh and others (New
England Medical Journal, 12/2005) and Dharan, Gubereva, Meyer et al
(Journal of the American Medical Association)
Whether oseltamivir was more dangerous than the A/H1N1 it was
supposed to treat/prevent, as suggested by the US FDA (Pediatric ADRs to
Tamiflu, 2007), Maxwell's Tamiflu and neuropsychiatric problems in
adolescents (BMJ) and the work of Rokura Hama.
Whether the rush to use oseltamivir to treat A/H1N1 was related to
the imminent expiry of stockpiles purchased in 2005 in the previous 'bird
flu' scare which would have lead to the destruction of pharmaceuticals
worth £500m in the UK alone.
As someone involved in advising policy on these matters, I was
mystified as to why the scientific community could not address these
issues at the time and, worse, actively sought to deflect dissent to the
prevailing view which appeared to amount to 'unless we all take
oseltamivir we'll die of H1N1'. I am perfectly happy to accept that my
understanding of medicine may well be at fault in my interpretation of at
least some of the studies I quote, but there has always been a significant
body of opinion which has questioned both the seriousness of the supposed
A/H1N1 pandemic, and the efficacy of oseltamivir as either a treatment or
a prophylaxis. For any who are interested, my own analysis was published
at www.swineswindle.blogspot.com . My apologies for the title, but I am a
journalist and not a medical professional.
Head of Media
Europe of Freedom & Democracy Group,
Competing interests: No competing interests
Jefferson et al. amply document the conclusion that they found the
literature on oseltamivir and lower respiratory tract complications of
influenza "confusing" . With 8 unpublished trials known to exist, the
authors' characterization of the 3 published trials as "insufficient to
answer the question" is clear and unarguable.
The confusion is evident in statements concerning a meta-analysis the
authors conducted with the 3 published trials, in their understandable
frustration at being unable to obtain usable methods and results from the
other 8 trials . The analysis produced a summary hazard ratio of 0.55
(95% confidence interval 0.22 to 1.35).
Jefferson et al. conclude that this meta-analysis "showed no
benefit." But what does that classically ambiguous expression mean? Did
the analysis not show a benefit? Or did it show that there is no benefit?
The authors come perilously close to the latter meaning in the
abstract of their paper, where they state that the meta-analysis "suggests
that oseltamivir did not reduce influenza related lower respiratory tract
complications." They strengthen this suggestion by concluding, "Paucity
of good data has undermined previous findings of oseltamivir's prevention
of complications from influenza."
There should be no confusion on two elementary points:
First, it is indisputably, even dangerously, incorrect to say of an
estimated hazard ratio of 0.55 with a 95% confidence interval of 0.22 to
1.35 that it shows no benefit or that it suggests there is no effect.
Second, the 3 published trials available to the authors were not
merely insufficient to answer the question. They were insufficient for a
Meta-analysis is just a tool. It should be used only when indicated.
Refraining using it when it is contraindicated is not finding an "excuse"
to "avoid" its use . It is the exercise of wise and judicious
restraint. When usable methods and results are available from only 3 of
11 trials known to have been conducted on a topic, that paucity clearly
The only conclusion worth drawing about oseltamivir and lower
respiratory tract complications of influenza from the paper by Jefferson
et al.  is that the methods and results from all extant trials of
relevance should be made available, without excuse or delay, by their
proprietors. At the moment public use of the drug began, those methods
and results became public goods.
1. Jefferson T, Jones M, Doshi P, Del Mar C. Neuraminidase
inhibitors for preventing and treating influenza in healthy adults:
systematic review and meta-analysis. BMJ 2009;339:b5106. Available from:
2. Doshi P. Neuraminidase inhibitors: the story behind the
Cochrane review. BMJ 2009;339:1348-1351.
3. Ioannidis JPA, Patsopoulos NA, Rothstein HR. Reasons or excuses
for avoiding meta-analysis in forest plots. BMJ 2008;336:1413-1415.
Competing interests: No competing interests
Neuraminidase inhibitors for prophylaxis and outcome measure: How efficacy claimed? What can we learn from the Japanese study?
Tamiflu is claimed effective for prophylaxis of laboratory confirmed
influenza by the results of several RCTs. I wonder what the laboratory
confirmed influenza is and I have a serious question "Is the efficacy for
Tamiflu does not substantially enter the brain unless the users have
fever when liver carboxyesterase activity may decrease and blood-brain
barrier (BBB) may be impaired by hypercytokinemia or unless they are under
1-year old and have immature carboxyesterase activity and immature BBB [1,
Thus, it can not lower body temperature substantially when they are
not infected during taking it for prophylaxis of influenza for otherwise
However, when they have fever once with hypercytokinemia, unchanged
oseltamivir (OT) enter the brain and lower body temperature. Therefore, it
is easily expected that OT may shorten the duration of high fever, but the
incidence with high fiver could not be reduced.
Then, how efficacy of Tamiflu for prevention claimed?
It is due to a sort of magic of testing. Substantial proportion of those
who are symptomatic due to influenza virus infection may be negative,
because Tamiflu (neuraminidase inhibitor) protect virus to leave from the
surface of the epithelial cells of respiratory tract and are difficult to
be removed to enter in a swab.
Thus some substantial proportion of patients who have influenza
symptoms become and classified as non-"test-proven" influenza patients!
I analyzed the papers of RCTs for prpphylaxis [3-6] and found no RCTs
conducted in the Western countries [3-5] had data indicating the exact
proportion of asymptomatic test-proven influenza.
For example Hyden et al  reported the number and proportion of:
(1) culture positive influenza-like illness,
(2) laboratory confirmed influenza with fever
(3) laboratory confirmed infection (symptomatic or asymptomatic)
(4) influenza like illness without laboratory evidence of infection
Proportion of sum of group(1) , group(2), only symptomatic in
group(3) and group(4) is the most comparable between Tamiflu arm and
However, proportion of persons who had asymptomatic laboratory-
confirmed infection might have been concealed by Tamiflu because
neuraminidase inhibitor protected virus to leave from the surface of the
epithelial cells of respiratory tract and are difficult to be removed to
enter in a swab as described above.
Moreover, if one has been infected with laboratory confirmed
influenza but has no symptoms, it is never be an adverse event for them.
So it should not be included in the outcome measures to assess the
Therefore, we have to exclude the number of asymptomatic laboratory
confirmed infected persons in order to do unbiased comparison of Tamiflu
arm to placebo arm.
Very fortunately, Japanese paper by Kashiwagi et al  reported the
number and proportion of asymptomatic laboratory confirmed infected
persons which enable unbiased comparison with placebo arm.
Attached figure shows total percent of persons with flu symptoms
(notes #1) who had positive testing and negative testing (notes #2) and
Persons withdrawn were three in the Tamiflu arm and none in the
Reasons of withdrawn were described in the paper: one adverse event,
one "not obedient to doctor" and one "voluntary withdrawn". It can not be
denied that true reasons of the latter two withdrawn might be adverse
reactions to Tamiflu. So if intention-to-treat analysis were adopted,
these adverse events should be included to compare the events as outcome
Thus events were 23.5 % for placebo arm and 23.9 % for the Tamiflu
75mg o.d. arm (for 42 days): no difference between both arms (Figure).
It may be the same as other neuraminidase inhibitors as zanamivir
except concerning with reducing body temperature.
One more important thing is that oseltamivir and zanamivir should be
separately analysed when adverse effects are systematically reviewed both
for treatment and prophylaxis, because oseltamivir and zanamivir are
completely different substances considering their toxicity. Zanamivir has
no direct central nervous system depressing actions, while unchanged
oseltamivir has, although oseltamivir carboxylate and zanamivir have both
properties as neuraminidase inhibitors and may impair delayed adverse
effects on nervous system and/or other organs .
(1) "Laboratory confirmed influenza" should be abandoned as the
outcome measure in not only RCTs but also observational studies assessing
the efficacy of neuraminidase inhibitors for prophylaxis of influenza.
(2) When proportions of symptomatic influenza are compared, number
and proportion of asymptomatic laboratory confirmed infected persons must
be excluded in order to enable unbiased comparison between experimental
arm with placebo arm.
(3) It may be the most appropriate to compare the proportion of
participants with any adverse events which could be considered as sever or
more as flu symptoms and/or flu-like events all together as the adverse
outcome in the systematic reviews on RCTs and/or observational studies for
prophylaxis of influenza.
(4) In order to conduct a systematic review concerning with adverse
effects of neuraminidase inhibitors, oseltamivir should be analysed
separately from other neuraminidase inhibitors.
1.Hama R.2008b Fatal neuropsychiatric adverse reactions to
oseltamivir: case series and overview of causal relationship. Int J Risk
Safety Med: 20: 5-36. available at:
2.Hama R. Oseltamivir is ineffective for prophylaxis of influenza.
The Informed Prescriber (in Japanese). 2005: 20 (2): 18-20.
3.Hayden FG, Atmar RL, Schilling M, Johnson C, Poretz D, Paar D, et
al. Use of the selective oral neuraminidase inhibitor oseltamivir to
prevent influenza. N Engl J Med 1999;341:1336-43.
4.Welliver R, Monto AS, Carewicz O, Schatteman E, Hassman M, Hedrick
J, et al. Effectiveness of oseltamivir in preventing influenza in
household contacts: a randomized controlled trial. JAMA 2001;285:748-54.
5.Peters PH Jr, Gravenstein S, Norwood P, De Bock V, Van Couter A,
Gibbens M, von Planta TA, Ward P. 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.
6.Kashiwagi S, Kudoh S, Watanabe A, Yoshimura I. [Efficacy and safety
of the selective oral neuraminidase inhibitor oseltamivir for prophylaxis
against influenza¡ªplacebo-controlled double-blind multicenter phase III
trial]. Kansenshogaku Zasshi 2000;74:1062-76.
Competing interests: No competing interests
Comments by Yorifuji et al  are very important in analysing
Japanese observational studies conducted by the task force of The Ministry
of Health Labour and Welfare (MHLW).
I would like to make some supplementary comments showing figures and
a table to make clearer and easy to read and with other references [2-4]
that also made critical appraisal against the MHLW's task force studies.
I would also like to make critical appraisal against the results of
observational studies including those reviewed by Freemantle and Calvert
 from other aspects than confounding with antipyretics .
Several authors argue that abnormal behaviors are not related to
Tamiflu use [7-12]. However, these are based on case reports or
observational studies with many methodological problems [2-4, 13-17] as
discussed below. They are weak evidence at most or may even be serious
obstacles for people to understand true figure of the toxicities of
The study by Yokota (one of the MHLW task forces  is not a
comparative cohort study with parallel follow-up groups but a sort of
before-and-after treatment study. It compared the frequency of abnormal
behaviour between children after treatment with Tamiflu ("treated" group)
and those not yet treated or never treated with it ("before/non-treated"
group) by dividing a day in three parts (morning: 6:00am to noon,
afternoon: noon to 18:00 and night 18:00 to 6:00 am next morning) for
seven days and calculated cumulative frequencies of both groups for the
whole study period.
Yokota et al  reported that overall cumulative incidences of
abnormal behaviours for 7-days were 11.9% vs 10.6%: hazard ratio 1.16,
Yokota's report  has many limitations. Among them one of the most
important ones is that it is not known which occurred earlier, the
ingestion of Tamiflu or the event within each period of first Tamiflu
intake for whole study period. However, exceptionally but luckily in the
afternoon on the first day of fever (the most important period), the
number of patients who have definitely been treated with Tamiflu (already
treated in the morning) and those who have definitely not been treated
with the drug (not yet treated with Tamiflu before 18:00 in the evening or
never treated) could be extracted [2-4] from the table of this study
report . Using these data, proportions of patients with abnormal
behaviour can be calculated particularly in this most important period.
Figure 1 shows the proportions of children with abnormal behaviour of
each group (modified from reference  and ).
According to the information provided by physicians proportions were
0.45% in "before/non-treated" cases and 1.82% in "treated" cases.
According to the information provided by families, the proportions of
children experiencing "terror or fear" were 0.38% vs. 2.00%. The figures
for "hallucination" were 0.055% vs. 0.66%, for "sudden screaming/delirious
speech" were 0.60% vs. 2.35% and for "anger" 0.55% vs. 2.03%. Thus
relative risks (and 95% confidence intervals) were 4.02 (1.52�E0.53), 5.22
(1.85�E4.68), 11.99 (1.57�E1.30), 3.89 (1.56�E.62) and 3.69 (1.40�E.67)
respectively (Figure 1).
I analysed the odds ratio in every period including the afternoon of
the first day by calculating under the assumption that 4/5 of the abnormal
behaviors occurred after taking Tamiflu as the relative risks (or adds
ratios) are about four or more [2-4].
High odds ratio were only clearly observed in the afternoon on the
first day of fever and were not observed from the second day to the
seventh day (Figure 2: modified from the figure of reference  and ).
This tendency coincides with FDA's analysis  based on cases
reported from Japan and the high frequency of vomiting occurring only on
the first day of the treatment in the RCT targeted children [2, 4, 19, 20]
(Table: translated from reference ).
Retrospective cohort study analysing large scale database [10, 11]
compared incidence of neuropsychiatric events within 14 days or 30 days
after influenza diagnosis between oseltamivir and no antiviral group. Time
to any CNS-related or neuropsychiatric events (mean/median was not
described) was about 11 days in each group  and median time to first
occurrence of CNS outcomes was 13 days in each groups . Nonspecific
events during long period might have concealed the difference.
The report by Tanabe  is neither cohort study nor case-control
study. It is not even a cross-sectional study and cannot compare the
effects or adverse effect of oseltamivir. However, it reported that of
those thirteen patients who experienced neuropsychiatric event before
taking oseltamivir three (23 %) experienced neuropsychiatric event after
This percent is rather higher compared with another cohort study by
Fujiwara et al  which reported the incidence (10.7%) of
neuropsychiatric events in children in the same situation (incidence of
events after taking Tamiflu in children who had events before taking
Moreover reanalysis of data of MHLW's task force [13-17] yield
significant association between abnormal behaviors and Tamiflu use. For
example, intention to treat analysis of a prospective cohort study 
comparing proportion of abnormal behaviour within four days after onset of
influenza in patients who were prescribed Tamiflu and those who were not
prescribed it (but were prescribed other drugs) yields odds ratio of 1.56
(95% CI: 1.32, 1.84, p=0.00000) [13-16].
If you exclude events before taking Tamiflu from the Tamiflu arm and
estimated number of events before taking other drugs from the control
group, estimated odds ratio is 1.72 (95%CI: 1.44, 2.08)  or estimated
rate ratio is 1.57 (95%CI: 1.37-1.79) .
Fujiwara et al  reported that an odds ratio of abnormal behaviour
comparing patients treated and not treated with Tamiflu by multivariate
logistic regression analysis was 2.18 (1.23, 3.87, p=0.008).
Taking account of the evidence from these observational studies along
with those indicating significantly higher incidences of major psychotic
and/or major psychiatric disorders in the randomized controlled trials
 and with many findings from animal toxicity, toxicokinetic , and
clinical data [4, 23], showing central depressing action of unchanged
oseltamivir, a causal link between neuropsychiatric disorders and taking
oseltamivir seems beyond doubt.
1) Yorifuji T, Tsuda T, Kashima S, Suzuki E, Doi H. Rapid response:
Implications for future adverse effect studies of neuraminidase
2) Hama R. Tamiflu induces abnormal behaviours most in the afternoon
of the first day of fever (part 1). The Informed prescriber 2006: 21(11):
110 -116. (in Japanese)
3) Hama R. Tamiflu-induced encephalopathy is 'yakugai (drug
disaster)'. Web-Kusuri-no-Check International No10. available at:
http://npojip.org/english/no10.html (translation from Kusuri-no-Check
No25: January 2007)
4) Hama R. Fatal neuropsychiatric adverse reactions to oseltamivir:
case series and overview of causal relationship. Int J Risk Safety Med.
2008: 20: 5-36. available at http://npojip.org/english/no11.html
5) Freemantle N and Calvert M. What can we learn from observational
studies of oseltamivir to treat influenza in healthy adults? BMJ 2009 339:
6) Hama R. Rapid response (Dec 12 2009). What can we learn from
observational studies of oseltamivir without examining NSAIDs use?
7) Yokota S. et al.2006. The study on the signs and symptoms during
influenza: A report of the MHLW's task force (in Japanese): available at:
8) MHLW.2007b Documents for the fifth 2007 advisory Panel on Drug
Safety held on 25 December 2007 (in Japanese):
9) MHLW.2009a. Documents for the first 2009 advisory Panel on Drug
Safety held on 16 June 2009 (in Japanese):
10) Blumentals WA, Song X. The safety of oseltamivir in patients with
influenza: analysis of healthcare claims data from six influenza seasons.
Med Gen Med. 2007: 9(4):23.
11) 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.
12) Tanabe T, Hara K, Nakajima M, Shimakawa S, Tamai H. Oseltamivir
treatment for children showing abnormal behavior during influenza virus
infection. Brain Dev. (2009) Feb 4. [Epub ahead of print]
13) Hama R. Tamiflu and abnormal behavior-reanalysis of Hirota’s
report. The Informed Prescriber. 2008: 23(1): 1-7 (in Japanese) available
14) Hama R. Problems in The Japanese MHLW epidemiological study on
oseltamivir. The Informed Prescriber. 2008: 23(8/9): 69-74 (in Japanese)
available at: http://npojip.org/sokuho/TIP2008-78-1.pdf
15) Hama R. Serious misclassification in The Japanese MHLW
epidemiological study on oseltamivir. BMJ rapid response 8 Aug. 2008.
available at: http://www.bmj.com/cgi/eletters/335/7610/59#200295.
16) Hama R. Pharmacological toxicological clinical and
epidemiological evidences of neuropsychiatric adverse reactions to
oseltamivir and misleading analysis due to serious miscalculation in
Japanese MHLW's 2006/07 epidemiological study report. Jap J Clin Pharmacol
Therap. 2009: 40 (2): 13S-14S. (in Japanese)
17) Yorifuji, T., E. Suzuki, and T. Tsuda, Oseltamivir and abnormal
behaviors: true or not? Epidemiology, 2009: 20: 619-21.
18) Edwards E.T. and Truffa M.M. 2005. (Division of Drug Risk
Evaluation: DDRE) One-Year Post Pediatric Exclusivity Postmarketing
Adverse Events Review (Drug: Oseltamivir phosphate): Department of Health
and Human Services, Public Health Services, Food and Drug administration:
Center for Drug Evaluation and Research=FDA CDER
19) Chugai Pharm Co. 2002. New drug approval package (NAP) of
oseltmivir (in Japanese); Tamiflu dry syrup (in Japanese): available at:
20) Hama R.2005c. Discussion of the causal relationship between
oseltamivir phosphate (Tamiflu), and sudden death and death from abnormal
behaviour (presentation at a session of Japanese Society for Pediatric
Infectious Diseases (in Japanese ): available at:
21) Fujiwara F, Hibi N et al. 2008. 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 (in Japanese) The summary data are available
at: http://npojip.org/sokuho/no116- koen1.pdf or
http://npojip.org/sokuho/no116-symp.pdf (both in Japanese)
22) Hama R. Rapid response (Dec.10 2009): Oseltamivir: psychotic and
neurological adverse reactions in the randomized controlled trials.
23) Hama R. Rapid response (Dec 8 2009): Early use of oseltamivir: a
case of respiratory depression after each of two doses, with sudden death.
Competing interests: No competing interests
In my previous rapid response (Dec. 17) �gEarly use of oseltamivir
and deaths in 09A-flu�h , I calculated proportion of attributable risk
using odds ratio instead of relative risk among oseltamivir users was
calculated as 97.9% (95%CI: 80.6-99.8%) and concluded that out of 36
deaths of sudden type, 35 (95%CI: 29-36) could have died due to harmful
effect of oseltamivir.
However, I am afraid that it may be overestimation, because the odds
ratio was derived from the comparison between Tamiflu users in sudden type
and in progressive type of death but not between control group consisting
people without serious outcomes.
There is no accurate estimation of proportion of Tamiflu users among
those who consulted medical facilities nor suitable case control study
conducted, the data which I used in my recent rapid response (Dec. 17)
�gAnother evidence on association of oseltamivir and death in 09A flu�h
) may provide approximate estimation: of 10.75 million who had been
infected with 09AH1N1 influenza since July to Nov 22 2009 in Japan ,
4.4 million oseltamivir users are estimated during July to November 22
2009, using the same proportion (about 70 percent) of patients infected
with 09AH1N1 influenza were prescribed neuraminidase inhibitor.
Compared with these data as proportion of oseltamivir users as
control group, 36 among 44 sudden-death after consulting medical
facilities yield odds ratio 6.55 (95 % confidence interval: 2.99-16.30).
Thus, estimated proportion of attributable risk is 84.7 % (95 %CI:
66.6-93.9%). Among 36 deaths of �gsudden type�h after taking Tamiflu, 30
deaths (95%CI: 24-34 death) were estimated attributable to oseltamivir
If the proportion of neuraminidase inhibitor user is 90 percent among
those who consulted, odds ratio (95%CI), proportion of attributable risk
(95%CI) and attributed deaths (95%CI) are estimated 4.09(1.87-10.19), 75.6
% (46.6-90.2%) and 27 deaths (16-32 death), respectively.
1) Hama R. (Rapid response Dec. 17 2009). Early use of oseltamivir and
deaths in 09A-flu:
2) Hama R. (Rapid response Dec. 17 2009). Another evidence on
association of oseltamivir and death in 09A flu.
3) The Japanese Infectious Surveillance Center of the National
Institute of Infectious Diseases. Infectious disease weekly report. No47.
2009 (in Japanese) http://idsc.nih.go.jp/idwr/kanja/idwr/idwr2009/idwr2009
Competing interests: No competing interests
A 36-year old Bulgarian musician was admitted with severe right-sided
abdominal pain and fever. A non-smoker and occasional drinker, he had had
an appendicectomy aged 19 but no other history of bowel disease. He had
visited Bulgaria for five days two weeks before the onset of his present
symptoms, and this visit coincided with feelings of lethargy. Two days
prior to admission, he had developed a fever with flu-like symptoms and a
diagnosis was made by his general practitioner of H1N1 influenza. He was
treated with oseltamivir, ibuprofen and paracetamol.
On admission, his temperature was 40oC, blood pressure 150/40 mm hg,
and he was tachycardic. He was pale with a petechial rash. Abdominal
examination revealed right iliac fossa tenderness with guarding and
rigidity. His venous blood showed pancytopenia with a white blood cell
count of 0.3 (normal value 4 – 10) x 109/L, haemoglobin 4.5 (normal value
12.5 – 17) g/dl and platelets 18 (normal value 120 – 400) x 109/L. Other
biochemical abnormalities included alanine transaminase 116 (normal value
0-37) IU/L, bilirubin 40 (normal value 0-17) μmol/L, APTT 30.8
(normal value 22-29) secs.
He was initially treated in a high dependency unit with broad
spectrum antibiotics, blood and platelet support and granulocyte colony
stimulating factor. Computed tomography showed gross wall thickening of
the proximal colon with a localised perforation arising from the proximal
ascending colon. Radiological appearances were in keeping with an
enterocolitis and subsequently a right-sided hemicolectomy was performed.
Histopathology confirmed neutropenic enterocolitis with ulceration,
transmural infarction and bacterial infiltration with minimal inflammatory
response in the gut wall (see figure).
Peripheral blood film showed toxic neutrophils with some reactive
lymphoplasmacytoid cells, but there were no blasts. The bone marrow biopsy
was markedly hypocelluar with a notable absence of fatty replacement. The
changes were more in keeping with acute toxic damage than conventional
aplastic anaemia. Parvovirus immunostains were negative and no granulomata
Serology for herpes simplex, varicella zoster, cytomegalovirus,
Epstein Barr virus, mycoplasma and rubella were negative. HIV 1 and 2
antibodies were not detected. Hepatitis A, B and C analysis did not show
current infection. There was serological evidence that toxoplasma
infection had occurred at some time; however, absence of toxoplasma IgM
excluded infection acquired recently. Serum zinc, selenium, arsenic,
lead, mercury and cadmium levels were all within normal limits. His H1N1
influenza screen from a naso-pharyngeal aspirate was negative.
He made a good clinical recovery with resolution of the pancytopenia.
He was discharged two weeks after presentation.
The appearance of the bone marrow biopsy and transient nature of the
aplastic episode would favour a toxic/drug aetiology. It is therefore
possible that this unusual clinical picture with life threatening changes
affecting the gut and blood, was a direct result of oseltamivir. The
absence of close patient monitoring during the current flu epidemic,
except in the most severe cases, may have led to adverse drug reactions
being overlooked. Oseltamivir has been associated with a haemorrhagic
colitis , , a condition which may have precipitated our patient’s
admission, but a resulting severe pancytopenia has never previously been
recorded, although there is a Japanese report of oseltamivir associated
with neutropenia . The transient nature of the pancytopenic episode may
favour a drug idiosyncratic reaction (or perhaps a reaction to an abnormal
metabolite) rather than causation by an underlying persistent viral
infection. An example is the occurrence of such a neutropenia associated
with a viral infection which occurs in H5N1 avian flu, where neutropenia
is accompanied by a poor prognosis . It is unlikely that the other
medications being taken by the patient, paracetamol or ibuprofen, were
responsible of the blood dyscrasia, as the patient had taken both these
drugs before without untoward effects. Although we have been unable to
exclude an unusual environmental aetiology, we concluded that the most
likely cause of our patient’s symptoms was a reaction to oseltamivir. In
view of the development of colitis, re-challenge with oseltamivir was not
advised. We suggest, therefore, that the prescribing of oseltamivir should
be with greater caution than is currently recommended . Certainly, a more
detailed scrutiny of oseltamivir administration is now indicated.
Written informed consent was obtained from the patient for
publication of this case report and accompanying images. A copy of the
written consent is available for review by the Editor-in-Chief of this
The authors declare that they have no competing interests
1. Matsushita M, Tanaka T, Omiya M, Okazaki K. Acute haemorrhagic
colitis associated with oral administration of Oseltamivir for treatment
of Influenza A. Journal of Infection and Chemotherapy 2007; 13:267 - 269
2. Yoneda S, Kobayashi Y, Nanoi T, Takeda K, Matsumori A, Andoh M et al.
Acute haemorrhagic colitis induced by neuraminidase inhibitor Oseltamivir.
Japanese Journal of Gastroenterology 2006; 103: 1270 - 1273
3. Cited in: Neutropenia side effect. www.patientsville.com. Tamiflu side
effect; 2009 report 5688780 – 9 page 6
4. Liem NT, Tung CV, Hien ND, Chau NQ, Long HT, Hien NT et al. Clinical
features of human influenza A (H5N1) infection in Vietnam 2004 – 2006.
Clinical Infectious Disease; 2009; 48; 1639 - 1646
5. US Department pf health and Human Services; Updated Interim
Recommendations for the Use of Antiviral Medications in the Treatment and
Prevention of Influenza for the 2009 – 2010 Season; Sept. 10th, 2009; pp 1
Competing interests: No competing interests
Jefferson and colleagues conducted a highly insightful study about
efficacy or effectiveness of neuraminidase inhibitors. In the study,
they also examined frequency of adverse events, but concluded that
evidence of rarer adverse events was of poor quality or possibly under-
reported. Thus, they emphasized the need for new studies to monitor the
safety of neuraminidase inhibitors by governments. We agree with that
opinion but, while we wait for new studies, a study conducted in Japan in
2005-2006 would be very informative.
The Japanese Ministry of Health, Labour, and Welfare funded two
epidemiological studies in the winter seasons of 2005-2006 and 2006-2007.
While the second, larger (>10,000) study is described elsewhere,[2, 3]
the first study is also informative because the study presented results
classified by disease duration. The first study mostly included 0-15 year-
old children (99.5%) (flu patients) across 12 prefectures in Japan, and
compared the incidence of adverse behaviors between patients who took
oseltamivir and those who did not. Adverse behaviors were collected by
questionnaires completed by doctors (2,846 patients) or families (2,545
patients). The former focused on abnormal behavior overall, while the
latter focused on more specific behaviors (frightening episodes,
hallucinations, abrupt screaming/delirious speech, and abrupt anger). The
questionnaire also asked when patients started administering oseltamivir
or manifested outcomes in three time-scales (morning, afternoon, or night)
for each day after the onset of fever.
The major limitation of the study was the lack of information on
temporality if the start of oseltamivir and behaviors were coincident in
the same time-scale. In the report, the study group considered subjects
who started oseltamivir and manifested abnormal behaviors in the same time
-scale as those who manifested abnormal behaviors after the intake of
oseltamivir. Although the study group concluded that there was no
significant difference between the two groups over 7 days, the relative
risk for abnormal behaviors in the afternoon on the first day was
consistently at least four times higher in the treated group including
starters (Table). However, similar findings were not observed after the
second day of the fever. We also showed results of alternative comparisons
to overcome the temporality problem. Although these adverse events are not
as severe as ones considered by Jefferson et al., the findings in this
Japanese epidemiological study (e.g., prevalence and elevated risks on the
first day) should provide further insight into the future studies.
Furthermore, it implies the existence of time-varying confounders or
effect modifications by disease duration, of which the latter seems more
1.Jefferson, T., et al., Neuraminidase inhibitors for preventing and
treating influenza in healthy adults: systematic review and meta-analysis.
BMJ, 2009. 339: p. b5106.
2.Hirota, Y., [Interim report of investigations about the frequency
of associated symptoms with influenza] (in Japanese), in Report of
Research on Health Science. 2008, Ministry of Health, Labour, and Welfare
of Japan: Tokyo.
3.Yorifuji, T., E. Suzuki, and T. Tsuda, Oseltamivir and abnormal
behaviors: true or not? Epidemiology, 2009. 20: p. 619-21.
4.Yokota, S., [Investigations about the frequency of associated
symptoms with influenza] (in Japanese), in Report of Research on Health
Science. 2006, Ministry of Health, Labour, and Welfare of Japan: Tokyo.
Table. Prevalence and risk ratios (RRs)* for abnormal behaviors in the afternoon (from noon to 6 p.m.) on the first day of fever in patients treated with oseltamivir (data in this table were retrieved from original report4)
Pre/untreated group Treated group, including starters† Already treated group only‡ Abnormal behaviors (%)§ 10 / 2185 (0.5) 12 / 636 (1.9) 6 / 329 (1.8) RR* (95% CI) 1 (Ref.) 4.1 (1.8 – 9.5) 4.0 (1.3 – 12.2) Frightening episodes (%)§ 7 / 1895 (0.4) 13 / 617 (2.1) 6 / 311 (1.9) RR* (95% CI) 1 (Ref.) 5.7 (2.3 – 14.2) 5.2 (1.8 – 15.4) Hallucinations (%)§ 1 / 1902 (0.1) 7 / 621 (1.1) 2 / 315 (0.6) RR* (95% CI) 1 (Ref.) 21.4 (2.6 – 173.9) 12.1 (1.1 – 132.8) Abrupt screaming/delirious speech (%)§ 11 / 1899 (0.6) 16 / 618 (2.6) 7 / 312 (2.2) RR* (95% CI) 1 (Ref.) 4.5 (2.1 – 9.6) 3.9 (1.5 – 9.9) Abrupt anger (%)§ 10 / 1896 (0.5) 12 / 615 (2.0) 6 / 309 (1.9) RR* (95% CI) 1 (Ref.) 3.7 (1.6 – 8.5) 3.7 (1.4 – 10.1)
* Risk ratios were estimated using the Pre/untreated group as a reference by Epi Info 3.3.2.
† Treated group includes subjects who started oseltamivir intake in the afternoon on the first day of the fever. In the analyses, we treated subjects who started oseltamivir and manifested abnormal behaviors in the afternoon as those who manifested abnormal behaviors after the intake of oseltamivir, as done by the original research group.
‡ Treated group includes only subject who started oseltamivir intake in the morning (between 6 am and noon) on the first day of the fever.
§ Number of cases with abnormal behaviors and the corresponding denominators (population at risk).
Abbreviations: RR, risk ratio; CI, confidence interval
Competing interests: Pre/untreated group Treated group, including starters† Already treated group only‡ Abnormal behaviors (%)§ 10 / 2185 (0.5) 12 / 636 (1.9) 6 / 329 (1.8) RR* (95% CI) 1 (Ref.) 4.1 (1.8 – 9.5) 4.0 (1.3 – 12.2)Frightening episodes (%)§ 7 / 1895 (0.4) 13 / 617 (2.1) 6 / 311 (1.9) RR* (95% CI) 1 (Ref.) 5.7 (2.3 – 14.2) 5.2 (1.8 – 15.4)Hallucinations (%)§ 1 / 1902 (0.1) 7 / 621 (1.1) 2 / 315 (0.6) RR* (95% CI) 1 (Ref.) 21.4 (2.6 – 173.9) 12.1 (1.1 – 132.8)Abrupt screaming/delirious speech (%)§ 11 / 1899 (0.6) 16 / 618 (2.6) 7 / 312 (2.2) RR* (95% CI) 1 (Ref.) 4.5 (2.1 – 9.6) 3.9 (1.5 – 9.9)Abrupt anger (%)§ 10 / 1896 (0.5) 12 / 615 (2.0) 6 / 309 (1.9) RR* (95% CI) 1 (Ref.) 3.7 (1.6 – 8.5) 3.7 (1.4 – 10.1)
In December 2009, an 18-year-old girl, with a past medical history of
occasional migraines and a low body mass index (BMI), experienced new
onset symptoms of headache, sore throat, coryzal symptoms, myalgia and
fever. Within 24 hours of the onset, she had telephoned her GP practice.
practice advised she contact the National Pandemic Flu Service who made a
presumptive diagnosis of Swine-Origin Influenza A (H1N1) 2009 infection.
She was prescribed oral oseltamivir (Tamiflu®) 75mg twice daily for five
After 24 hours and three doses of oseltamivir she noticed a rash over her
abdomen. Her mother then contacted the GP who visited the patient and
noted a widespread maculopapular rash over the trunk and upper limbs,
associated tachycardia, pyrexia (temperature 39.4°C), significant cervical
lymphadenopathy, bilateral conjunctivitis and pharyngitis. He prescribed
penicillin V because of a concern about scarlet fever and advised the
stop oseltamivir. The next day the GP reviewed the patient noting blisters
erosions, particularly in the mouth and lips and referred the patient to
local dermatology team.
On examination at the department of dermatology she was alert and
orientated but noted to have an extensive macular rash on the trunk and
limbs with lesions forming a confluent sheet of erythema on the abdomen.
Scattered targetoid lesions were present over the periphery of the trunk
proximal limbs. There were initially small discrete areas of blistering on
face and lower abdomen (Figure 1). Crusted erosions were visible on the
and there was conjunctival suffusion. The rash was subjectively described
itchy but not painful. She was pyrexial (temperature 39.5 °C) and
but normotensive, normoxyaemic with a normal respiratory examination. Her
initial blood tests showed a thrombocytopenia, elevated serum C-reactive
protein, mildly deranged electrolytes and liver function and normal renal
function (Table 1). Blood film showed toxic granulation. She was admitted
hospital, a skin biopsy was taken from the thigh and nasal and pharyngeal
swabs were sent to HPA Southampton laboratory for H1N1 PCR. Initial
treatment of the skin lesions consisted of emollients and non-adherent,
silicone dressings. Topical chloramphenicol, prednisolone and lubricants
applied to the eyes. Regular paracetamol, cyclizine and intravenous fluids
were administered. She remained stable initially but deteriorated rapidly
overnight with significant breathing difficulties, hypotension and
due to worsening toxic epidermal necrolysis (TEN; SCORTEN score 1,
mortality 3.2%). She was transferred to ITU where she required
support and invasive ventilation. Intravenous immunoglobulin (IV Ig) was
administered for TEN and intravenous co-amoxiclav was started for a
suspected lower respiratory tract infection. She was noted to be
and treated with lenograstim (GCSF). Intravenous vitamins B and C
(Pabrinex®) were added because of concern over malnutrition (BMI 16 kg/m2
at presentation). By the third hospital day approximately half the body
surface area was denuded of epithelium. On the fourth day the patient was
transferred to the Burns Unit at the Chelsea and Westminster Hospital,
London, where her treatment is continuing. At bronchoscopy there was
evidence of mucosal erythema and necrosis extending into right and left
segments. Follow up bronchoscopy at day 6 revealed early reepithelisation
the mucosa. She was treated according to the local protocol: IV Ig, GCSF
cyclosporine. The initial biopsy of the skin showed evidence of
dermatitis with an erythema multiforme like picture in keeping with the
spectrum of TEN and a second biopsy taken in London was diagnostic of TEN.
Human immunodeficiency virus serology was negative. The swabs taken for
Swine-Origin Influenza A (H1N1) 2009 were negative.
Severe skin reactions (including Stevens-Johnson syndrome and toxic
epidermal necrolysis) following oseltamivir are listed as undesirable
the Summary of Product Characteristics. There has been much recent
concern about the safety profile of neuraminidase inhibitors in adults and
children with influenza.[4,5] Between 1st April and 24th November 2009,
adverse reactions to oseltamivir were reported to the Medicines and
Healthcare Products Regulatory Agency (MHRA) of which 329 (34%) involved
the skin or subcutaneous tissues. Of the 10 fatalities reported, none were
related to the skin. Two cases of TEN have been reported to the UK’s MHRA
but neither appear in the literature.
As of 3rd December 2009, the UK’s Health Protection Agency estimates
the cumulative total number of cases of Swine-Origin Influenza A (H1N1)
2009 is 790,000 since the pandemic began with 22,000 new cases in England
in the last week. The estimated case fatality rate is 26 (range 11–66)
100,000. Between 23rd July and 17th November 2009, 941,890 courses of
oseltamivir have been collected via the National Pandemic Flu Service in
England. It is unclear whether the trial data for oseltamivir can be
generalised to a pandemic scenario; however, Jefferson and colleagues
evidence of benefit for oseltamivir if taken within 48 hours of the onset
symptoms in shortening the duration of influenza-like illness. Data on the
effectiveness of oseltamivir against complications of influenza remains
controversial and a paucity of good data has undermined previous findings
for oseltamivir’s effectiveness.
Toxic epidermal necrolysis can present with a prodromal influenza-
illness similar to that experienced by our patient prior to the first dose
oseltamivir. Although infection or other agents have been reported to
TEN, there is a consensus amongst dermatologists that idiosyncratic drug
reactions account for the majority of cases. Our patient and her mother
us that she had not taken any medications other than oseltamivir before
onset of her rash. Therefore we have significant concern that this
the first case of TEN following oseltamivir, administered empirically for
influenza-like illness by the UK’s National Pandemic Flu Service. Defining
aetiology of serious adverse events may be difficult; information about
patient's setting, symptoms, and timing of the event is often
precise, missing or absent. Nevertheless, the risks and benefits of
oseltamivir remain uncertain. Therefore life-threatening adverse
however rare, must be carefully considered when the distribution of
oseltamivir is planned for future pandemics.
Tom Parks (1,2) academic foundation year 1 doctor
Iaisha Ali (1) specialist registrar in dermatology
Kamal R. Mahtani (1,3) specialist trainee in general practice and
clinical fellow (firstname.lastname@example.org)
Hanif Esmail (4) specialty registrar in infectious diseases
Arani Chandrakumar (5) specialist registrar in dermatology
Suveer Singh (6) consultant intensivist (email@example.com)
Jorge Leon-Villapalos (7) consultant plastic surgeon (jorge.leon-
Ruth Asher (8) consultant dermatopathologist (firstname.lastname@example.org)
Chris Bunker (5) consultant dermatologist
Carl Heneghan (3) clinical lecturer (email@example.com)
Vanessa Venning (1)* consultant dermatologist
(1) Department of Dermatology, Churchill Hospital, Old Road,
Oxford OX3 7JL
(2) Oxford University Clinical Academic Graduate School, Medical
Division, Level 3, John Radcliffe Hospital, Oxford OX3 9DU
(3) Department of Primary Health Care, University of Oxford, Rosemary
Building, Old Road Campus, Headington, Oxford OX3 7LF
(4) Department of Microbiology and Infectious Diseases, Level 7, John
Radcliffe Hospital, Oxford OX3 9DU
(5) Department of Dermatology, Chelsea and Westminister Hospital, 369
Fulham Road, London SW10 9NH
(6) Department of Anaesthesia and Intensive Care, Chelsea and
Hospital, 369 Fulham Road, London SW10 9NH
(7) Department of Plastic Surgery, Chelsea and Westminster Hospital,
Fulham Road, London SW10 9NH
(8) Department of Cellular Pathology, Level 1, John Radcliffe
*Correspondence to: Dr Vanessa Venning (firstname.lastname@example.org)
Author contribution: TP and KRM conceived the idea of rapid
submission as a
case report. TP wrote the first draft which was edited by KRM, VV and CH.
IA, HE, AC, SS, JLV, RA, CB and VV were responsible for the clinical care
patient. Funding: No external funding was sought prior to submission of
Competing interests: All authors have completed the Unified Competing
Interest form at www.icmje.org/coi_disclosure.pdf (available on request
the corresponding author) (URL) and declare that all authors had: (1) No
financial support for the submitted work from anyone other than their
employer; (2) No financial relationships with commercial entities that
have an interest in the submitted work; (3) No spouses, partners, or
with relationships with commercial entities that might have an interest in
submitted work; (4) No Non-financial interests that may be relevant to the
submitted workPatient consent: A signed consent form has been received
from the patient and submitted to The BMJ.
Licensing agreement: The Corresponding Author has the right to grant
behalf of all authors and does grant on behalf of all authors, an
licence (or non exclusive for government employees) on a worldwide basis
the BMJ Publishing Group Ltd and its licensees, to permit this article (if
accepted) to be published in BMJ editions and any other BMJPG products and
to exploit all subsidiary rights, as set out in our licence
1. Bastuji-Garin S FN, Bertocchi M, Roujeau JC, Revuz J, Wolkenstein
SCORTEN: a severity-of-illness score for toxic epidermal necrolysis. J
2. de Sica-Chapman A, Williams G, Soni N, Bunker CB. Granulocyte
stimulating factor in toxic epidermal necrolysis (TEN) and Chelsea &
Westminster TEN protocol. Br J Dermatol. 2009 Nov 11. [Epub ahead of
3. Roche. Tamiflu 75mg hard capsule – Summary of Product
Available from: http://emc.medicines.org.uk/ Accessed on: 9th December
4. Godlee F CM. Why don’t we have all the evidence on oseltamivir?
5. Jefferson T JM, Doshi P, Del Mar C. Neuraminidase inhibitors for
preventing and treating influenza in healthy adults: systematic review and
meta- analysis. BMJ. 2009;339:b5106.
6. Medicines and Healthcare Products Regulatory Agency. UK Suspected
Adverse Drug Reaction (ADR) Analysis: Influenza antivirals - oseltamivir
(Tamiflu) and zanamivir (Relenza): 3rd December 2009. London: Medicines
and Healthcare Products Regulatory Agency; 2009
7. Health Protection Agency. Weekly pandemic flu media update 3
December 2009. London: Health Protection Agency; 2009
8. Donaldson LJ, Rutter PD, Ellis BM, Greaves FEC, Mytton OT, Pebody
al. Mortality from pandemic A/H1N1 2009 influenza in England: public
surveillance study. BMJ 2009;339:b5213
9. Jefferson T JM, Doshi P, Del Mar C. Possible harms of
for urgent action. Lancet. 2009;374:1312-1313.
The authors would like to thank Dr George Moncrieff, the patient’s
the Department of Plastic Surgery, Department of Anaesthesia and Intensive
Care and the Burns Unit at the Chelsea and Westminster Hospital, London,
their help in reporting the case and their care of the patient. Thanks
to Dr Chris Conlon, Infectious Diseases, Churchill Hospital, Oxford, and
Duncan Young, Intensive Care Unit, Churchill Hospital, Oxford who cared
the patient while she was at the Churchill Hospital.
Figure 1. The patient’s face soon after admission
See: Oseltamivir Figure 1.jpg
Table 1. Blood results on admission Sodium 135 mmol/L Potassium 3.4 mmol/L Urea 5.2 mmol/L CRP 135 mg/L Creatinine 54 mmol/L Bilirubin 25 mmol/L ALT 58 IU/L ALP 79 IU/L Albumin 30 g/L Adj. Calcium 2.25 mmol/L Phosphate 0.48 mmol/L Magnesium 0.71 mmol/L Haemoglobin 13.5 g/dL WCC 6.55 x 109/L Platelets 74 x 109/L Neutrophils 1.77 x 109/L Lymphocytes 0.46 x 109/L Band forms 3.93 x 109/L Metamyelocytes 0.39 x 109/L
Editorial note: Patient consent obtained.
Competing interests: Table 1. Blood results on admissionSodium 135 mmol/LPotassium 3.4 mmol/LUrea 5.2 mmol/LCRP 135 mg/LCreatinine 54 mmol/LBilirubin 25 mmol/LALT 58 IU/LALP 79 IU/LAlbumin 30 g/LAdj. Calcium 2.25 mmol/LPhosphate 0.48 mmol/LMagnesium 0.71 mmol/LHaemoglobin 13.5 g/dLWCC 6.55 x 109/LPlatelets 74 x 109/LNeutrophils 1.77 x 109/LLymphocytes 0.46 x 109/LBand forms 3.93 x 109/LMetamyelocytes 0.39 x 109/L
The Japanese Infectious Surveillance Center of the National Institute
of Infectious Diseases reported that it was estimated that 10.75 million
Japanese had been infected with 09AH1N1 influenza since July to Nov 22
According to the press releases posted by The Japanese Ministry of
Health, Labour and Welfare (MHLW), 74 died with 09AH1N1 influenza
infection since July to Nov 27 2009. I analysed the data from press
releases and estimated the median and average time from onset of symptoms
to presenting medical facilities about one half day and less than a day
respectively. Median and average time from onset of symptoms to death were
estimated 3 days and 5.8 days respectively.
There may be some patients who survive but are potentially fatal.
Therefore, case fatality rate is roughly estimated less than one in 100
thousand infected. The case fatality rate is approximately one tenth to
one fifth of those observed in seasonal influenza of recent years in
Of 74 died, 39 were given oseltamivir and 6 were given zanamivir
before deterioration leading to death.
On the other hand, according to the Chugai's data submitted to MHLW
on Nov. 30 2009 , in the 2008/2009 season, about 68 percent of
influenza-infected patients were given neuraminidase inhibitors. Of these,
about 53 % were given oseltamivir and about 47 % were given zanamivir.
According to Glaxo Smithklein�fs data in the same document , estimated
market share of zanamivir during May to October 2009 was 41.8 % and the
remaining (58.2 %) was that of oseltamivir.
Number of oseltamivir users are estimated 4.4 million during July to
November 22 2009 and 3.1 million for zanamivir in the same period.
These data yield relative risk of mortality for oseltamivir to
zanamivir as 4.69 (95 % confidence interval: 1.96-13.49, p=0.0001).
In conclusion, these data indicate that oseltamivir seems
substantially harmful on survival in 09A influenza compared with
1) The Japanese Infectious Surveillance Center of the National
Institute of Infectious Diseases. Infectious disease weekly report. No47.
2009 (in Japanese)
2) MHLW : Press release on novel influenza (in Japanese)
http://www.mhlw.go.jp/kinkyu/kenkou/influenza/houdou.html �gOn a fatal
case during infection with novel influenza�h (in Japanese)
accessed on Dec 16 2009.
3) Documents submitted to The Ministry of Health, Labour and Welfare
by Glaxo Smithkleine and by Chigai (in Japanese) available at
Competing interests: No competing interests
The Japanese Ministry of Health, Labour and Welfare has since July
2009 been posting press releases on its website on all fatal cases of
"New type influenza A/H1N1" . Most of the documents reported
information on the use of neuraminidase inhibitors, but scarcely of
antipyretics. I found two types of cause of death.
One is "sudden type" with sudden deterioration showing such as
respiratory arrest or death during sleep and the other is "progressive
type" with progressive course of deterioration typically sepsis and/or
acute respiratory failure with other multiorgan failure. If a case with
progressive course of deterioration was followed by apparent unexpectedly
sudden respiratory arrest or death it was classified as sudden type.
The sudden type is typically seen in the fatal case which I reported
previously  and recently . Among them the definite one is the case
in which two episodes of sudden respiratory disturbance and hypoxemia
followed the first and the second dose of oseltamivir is noted . Most
of the autopsied cases have non-cardiac hypoxic lung edema with mild right
cardiac enlargement suggesting transient pulmonary hypertension due to
lung edema .
Both types are seen not only after but also before presenting at
medical facilities. I have analysed 74 deaths reported up to November 27
The number of sudden type was 54 (10: "before consultation", 44: "after
consultation") and that of the progressive type was 20 (7: "before
consultation", 13: "after consultation"). All the prescribed medicines
were considered as taken by the patients unless there are any comments
suggesting they were not taken.
Data including age, sex, risk factor (underlying disease), time to
presenting, results of rapid antigen testing official cause of death,
prescription of medicines including neuraminidase inhibitors and
antipyretics and time from use of neuraminidase inhibitors to
deterioration (mechanically ventilated or death) were extracted and
Of 44 patients who were given neuraminidase inhibitor (38 oseltamivir
and 5 zanamivir and one both) before deterioration (mechanically
ventilated or death), 40 (91 %) was commenced within 2 days after the
onset of symptom (mainly fever). Of 39 sudden type patients in 36 patients
(92.3%) they were commenced within two days.
36 of sudden type (67% in 54 all sudden type or 82% in 44 "after"
consultation) and 3 of progressive type (15% in 20 all progressive type
and 23 % in 13 "after" consultation) were given Tamiflu before
Odds ratio was 11.33 (95% confidence interval (CI): 2.68-85.91) for
both "before" and "after" consultation, and 15.0 (95%CI: 2.82-97.8)
only for "after" consultation.
Odds ratios of proportion of Tamiflu use within 2 days and 1 day before
deterioration among those "after" were 32.0 (95%CI: 3.74-1409) and 42.4
4 of sudden type and 2 of progressive type were given Zanamivir (no
significant difference detected).
Of 36 patients who were given Tamiflu 15 took it within several hours
before deterioration and 27 took it within a day, while none of
progressive type took it within a day before deterioration.
Multivariate logistic regression was used to estimate odds ratios and
95%CI for proportion of neuraminidase use in relation to type of death
using StatsDirect version 2.7.7. Adjustment in a multivariate analysis for
age, sex, presence of underlying disease and zanamivir yield adjusted odds
ratio for oseltamivir use was 47.2 (95%CI: 5.16-432.4). Adjusted odds
ratio for zanamivir was 1.46 (95%CI: 0.13-16.23).
Inclusion of other factors in the multivariate model, such as time
from onset of symptoms to consultation and results of rapid testing for
influenza did not substantially change the results. Data of antipyretic
use was not included because of too little information was available.
Proportion of attributable risk using odds ratio instead of relative
risk among oseltamivir users was calculated as 97.9% (95%CI: 80.6-99.8%).
I conclude that out of 36 deaths of sudden type, 35 (95%CI: 29-36)
could have died due to harmful effect of oseltamivir.
1) MHLW : Press release on novel influenza (in Japanese)
http://www.mhlw.go.jp/kinkyu/kenkou/influenza/houdou.html "On a fatal
case during infection with novel influenza" (in Japanese)
accessed on Dec 16 2009.
2. Hama R, Fatal neuropsychiatric adverse reactions to oseltamivir:
case series and overview of causal relationships. Int J Risk Safety Med.
2008: 20: 5-36. available at : http://npojip.org/sokuho/published-
paperJRS431.pdf accessed on Dec 16 2009.
3. Hama R. Rapid response: Early use of oseltamivir: a case of
respiratory depression after each of two doses, with sudden death.
Competing interests: No competing interests