The Tamiflu Spin

I will start, for those of you who are new to the blog, with two disclaimers.

First, I am an infectious disease doctor. It is a simple job: Me find bug. Me kill bug. Me go home. I spend all day taking care of patients with infections. My income comes from treating and preventing infections. So I must have some sort of bias, the main one being I like to do everything I can to cure my patients.

Second, in 25 years I have, to my knowledge, accepted one thing from a drug company. The Unisin (that’s how I spell it) rep, upon transfer from my hospital, sent me a Fleet enema with a Unisin sticker on it. I show it proudly to all who enter my office. I do not even eat the drug company pizza at conference, and I cannot begin to tell you painful that is.

As we leave (I hope) the H1N1 season and enter seasonal flu season, there has been a flurry of articles, originating in the British Medical Journal , questioning whether oseltamivir is effective in treating influenza. The specific complaint at issue is whether or not oseltamivir prevents secondary complications of influenza like hospitalization and pneumonia. Although you wouldn’t guess that was at issue from the reporting.  As always, there is what the data says, what the abstract says, what the conclusion says, and what other people say it says.  Reading the medical literature is all about blind men and elephants.

There is, evidently, going to be an investigation by the European Union Council of Europe  into whether or not the H1N1 pandemic was faked to sell more oseltamivir. Sigh.

Is oseltamivir effective against flu? Lets start with background. What do I want to know about an antiviral to help determine its efficacy?

I would like a mechanism of action that would prevent replication of the virus. I can’t kill a virus, as I can a bacteria, but at least I can stop if from reproducing.

I would like the antiviral to be effective in the test tube at physiologically achievable concentrations.

I would like the antiviral to be effective in an animal model.

I would like the antiviral to be effective clinically: on challenge studies if ethical and in the real world.

There is the virulence of the organism, and there can be strain to stain variation in the virulence. Some influenza strains are better at spreading or killing than other strains. Everyone seems to have forgotten that when H1N1 started in Mexico it apparently had a horrific mortality rate in hospitalized patients.

” By 60 days, 24 patients had died (41.4%; 95% confidence interval, 28.9%-55.0%).”

The worry was that there was going to be a repeat of the 1919 influenza pandemic. Fortunately the subsequent mortality rate was much less as it swept the world. H1N1 turned out to be a highly infectious, low virulence strain of flu. This time. But I would still fret that one-day there will be another repeat of the 1919 pandemic. The current bird flu has a 66% mortality rate, but is not spread human to human. Maybe someday it will gain that ability.  But at the beginning no one knew what mortality of H1N1 or bird flu was going to be and they prepared accordingly. Of course public health officials are always in a no-win situation. They are either going to over-prepare or under-prepare, and as a result at best look a fool and at worst look evil. Next time they may err on the side of under-preparing so they are not accused of faking a pandemic.

There is the hosts ability to respond to the infection. Just as there is variability in the virulence of the pathogen, there is variability in the hosts ability to control the infection. It is often the case that whether you live or die from an infection may be due to the immune system with which you are born. Pubmed ‘toll like receptor’ ‘polymorphism’ and ‘infection’ for further information that is beyond the scope of this entry. There may be an inherited predisposition to dying from influenza.

In the H1N1 pandemic, mild as it was overall, caused disproportionately relatively high mortality in children

“Between May and July 2009, a total of 251 children were hospitalized with 2009 H1N1 influenza. Rates of hospitalization were double those for seasonal influenza in 2008. Of the children who were hospitalized, 47 (19%) were admitted to an intensive care unit, 42 (17%) required mechanical ventilation, and 13 (5%) died. The overall rate of death was 1.1 per 100,000 children, as compared with 0.1 per 100,000 children for seasonal influenza in 2007. (No pediatric deaths associated with seasonal influenza were reported in 2008.) “

and pregnant women

“Data were reported for 94 pregnant women, 8 postpartum women, and 137 nonpregnant women of reproductive age who were hospitalized with 2009 H1N1 influenza… In all, 18 pregnant women and 4 postpartum women (total, 22 of 102 [22%]) required intensive care, and 8 (8%) died.”

Who gets the disease may be, depending on the organism, more important than the strain of infection.

There is how strong or powerful (meaningless terms to an ID doc, but part of the popular language of medicine) the antibiotic is. Antivirals are not that effective because they do not kill viruses, they only halt their replication. Given the prodigious replicative capacity of viruses, it is going to be impossible to shut down every virus from replicating with an antiviral. With HIV it took the combination of three different anti-retrovirals to (almost?) completely shut down viral replication. As those without immune systems prove with depressing regularity, no antibiotic will work for long if the host’s immune system cannot help control the infection.

Finally, there is promptness of therapy. For serious infections even a day in the delay of appropriate antibiotics can dramatically increase mortality. So the later you begin therapy, the less effect one should expect. This is one of the issues with a disease with a very rapid onset like influenza and why the challenge studies, where you give the medication right after exposing the subject to influenza, show better effect. Viral replication will often rapidly outstrip the available antiviral.

What would I expect from an influenza medication?

Prevent disease after exposure?
Decrease the severity of the infection?
Prevent death in all cases? Prevent death in the more severe cases? Decrease the odds of dying?
Decrease complications of the disease?
Some combination of the above?

All good endpoints. As a rule, I expect antivirals for to lessen the severity of disease if given early in the disease.

So what can we say for oseltamivir?

Does it have a mechanism of action that would interfere with viral replication? Yes. Oseltamivir blocks the activity of the viral neuraminidase enzyme, preventing new viral particles from being released by infected cells. Not a thrilling mechanism of action. The virus can multiply all it wants if it infects a cell, and since no drug is 100% effective, one would expect it would slow down the disease, not stop it.
Sometimes, as we shall see, that may be enough to make the difference between life and death.

Does oseltamivir work in the test tube? Yes. In animal models? Yeah. In human clinical trials? Yes.

It is in human clinical trials and the choice of efficacy endpoints where the current brouhaha starts.

LA times headline reads

“British medical journal questions efficacy of Tamiflu for swine flu — or any flu.”

The Atlantic , which has now supplanted the Natural News for the worst medical coverage, has a headline that reads

“The Truth About Tamiflu”

followed by the opening paragraphs.

“Two months ago, we pointed out in our story on flu in The Atlantic that the antiviral drug Tamiflu might not be as effective or safe as many patients, doctors, and governments think. The drug has been widely prescribed since the first cases of H1N1 flu surfaced last spring, and the U.S. government has spent more than $1.5 billion stockpiling it since 2005 as part of the nation’s pandemic preparedness plan.

Now it looks as if our concerns were correct, and the nation may have put more than a billion dollars into the medical equivalent of a mirage. This week, the British medical journal BMJ published a multi-part investigation that confirms that the scientific evidence just isn’t there to show that Tamiflu prevents serious complications, hospitalization, or death in people that have the flu. The BMJ goes further to suggest that Roche, the Swiss company that manufactures and markets Tamiflu, may have misled governments and physicians. In its defense, Roche stated that the company “has never concealed (or had the intention to conceal) any pertinent data.”

The medical equivalent of a mirage. Hmmm. And death? The Cochrane review, upon which the whole controversy revolves, does not comment of prevention of death nor does the BMJ feature or editorial on the topic. Into the second paragraph and already they are apparently making things up. I remember when I trusted the Atlantic. Interestingly, the Atlantic never quotes the results of the Cochran review. They suggest that the review demonstrates that oseltamivir is worthless.

Does it? Start with the methods section.

They looked for randomized, placebo controlled trials. Cochran gold.

“We excluded experimental influenza challenge studies as their generalizability and comparability with field studies is uncertain.”

Which is a shame, as the challenge studies always show efficacy. But they are not representative of the real world as they have the best case for treating with an antiviral: you can start the medication right when the disease may start.

They screened 1416 articles and ended up with 29 studies, 10 for effectiveness, the cream of the crop. 1416 articles is a lot of articles. I know they are not the crème de la crème, and is why I am not a fan on just relying on meta-analysis alone. That is a lot of ignored information. And their results?

Here is the conclusion in the abstract:

“Neuraminidase inhibitors have modest effectiveness against the symptoms of influenza in otherwise healthy adults. The drugs are effective postexposure against laboratory confirmed influenza, but this is a small component of influenza-like illness, so for this outcome neuraminidase inhibitors are not effective. Neuraminidase inhibitors might be regarded as optional for reducing the symptoms of seasonal influenza. Paucity of good data has undermined previous findings for oseltamivir’s prevention of complications from influenza. Independent randomized trials to resolve these uncertainties are needed.”

I am already confused.

“The drugs are effective postexposure against laboratory confirmed influenza, but this is a small component of influenza-like illness, so for this outcome neuraminidase inhibitors are not effective.”

They are effective but they are not. But the abstract suggests in normal people, oseltamivir has modest efficacy.

Later, there is the Take Home message box, which says


Neuraminidase inhibitors (especially oseltamivir) have become global public health drugs for influenza
They prevent symptoms and shorten the duration of illness by about one day if taken within 48 hours of the onset of symptoms
Toxicity and the effects on complications have been debated


Neuraminidase inhibitors reduce the symptoms of influenza modestly
Neuraminidase inhibitors reduce the chance of people exposed to influenza developing laboratory confirmed influenza but not influenza-like illness
Evidence for or against their benefit for preventing complications of influenza is insufficient
Evidence for or against serious adverse events is lacking, although oseltamivir causes nausea”

To me the take home message suggests some efficacy. Lets move on to the results and the discussion. Perhaps it will clarify the situation.

“The data suggest that neuraminidase inhibitors are effective at reducing the symptoms of influenza. The evidence is of modest benefit—reduction of illness by about one day. “

Which is what I would expect in a population of healthy, mostly young people with seasonal flu of low virulence.
Note the caveats. Young, healtyh people and influenza of low virulence.
Every season we have new strains of influenza and the ability of influenza to kill people varies from year to year.

“This benefit has been generalized to assume benefits for very ill people in hospital. This seems reasonable, although it is worth remembering that we have no data to support this, and it is unlikely that ethics committees would allow a trial of no treatment for people with influenza who have life threatening disease.”

Yeah. We will probably not have randomized controlled trials in the seriously ill. While we do not have randomized, controlled, clinical trials, we do have data to support the use of oseltamivir in ill patients. Not no data.

In the Mexican experience with H1N1 mentioned above,

“After adjusting for a reduced opportunity of patients dying early to receive neuraminidase inhibitors, neuraminidase inhibitor treatment (vs no treatment) was associated with improved survival (odds ratio, 8.5; 95% confidence interval, 1.2-62.8).”

And in pregnant women with H1N1 mentioned above

“As compared with early antiviral treatment (administered 2 days after symptom onset) in pregnant women, later treatment was associated with admission to an intensive care unit (ICU) or death (relative risk, 4.3).”

Or this experience with H1N1

“Of the 268 patients for whom data were available regarding the use of antiviral drugs, such therapy was initiated in 200 patients (75%) at a median of 3 days after the onset of illness. Data suggest that the use of antiviral drugs was beneficial in hospitalized patients, especially when such therapy was initiated early.”

And what about seasonal flu?

“No randomized trials of neuraminidase-inhibitor treatment of hospitalized influenza patients have been conducted. However, three observational studies suggest that oseltamivir treatment of hospitalized patients with seasonal influenza may reduce mortality. In one prospective Canadian study among hospitalized patients with seasonal influenza, (N=327; mean age, 77 years), in which 71% began oseltamivir treatment >48 hours after illness onset, oseltamivir treatment was significantly associated with a reduced risk of death (OR, 0.21; P=0.03) within 15 days after hospitalization as compared with untreated patients.7 In a subanalysis, in a Hong Kong study of hospitalized seasonal influenza patients (N=356; mean age, 70.2 years), oseltamivir treatment initiated within <96 hours after illness onset was independently associated with decreased mortality as compared with untreated patients (OR, 0.26; P=0.001).8 A retrospective chart review of hospitalized seasonal influenza patients in Thailand (N=445; mean age, 22 years), including 35% with radiographically confirmed pneumonia, reported that any oseltamivir treatment was significantly associated with survival (OR, 0.11; 95% CI, 0.04 – 0.30) as compared with untreated patients.”

In the seriously ill, people who are going to die from influenza, taking oseltamivir appears to decrease the odds of dying. Remember that the authors of the Atlantic article said olsetamivir didn’t prevent death. Maybe their firewall blocks access to Pubmed and Google. Or maybe they are no good at evaluating elephants.

So what would you do if you were in charge of public health policy and confronted with a new strain of influenza, be it H1N1 or bird flu, with a potentially catastrophic mortality rate? Ignore it? Or get as much oseltamivir and vaccine as you could lay your hands on?

Maybe not the highest quality studies, but we have a biologic mechanism, we have test tube and animal studies, we have challenge studies, we retrospective studies that show efficacy. We have a lot of data to support efficacy of antivirals in some patient populations.

The question is not whether oseltamivir is effective, but in what population the medication will be effective and for what strains of influenza. Certainly I am not going to withhold oseltamivir from a hospitalized pregnant female with presumptive H1N1 given an 8% mortality rate. Does a 45 yo with H1N1 or seasonal flu need oseltamivir? No. An 85 year old with multiple medical problems? Probably. Nuance. Subtlety. Understanding the breadth and depth of a topic no longer seems to be part of the Atlantic medical reporting. I originally typed  the last sentence as the Atlantis medical reporting. Pity I have to change it; it seemed much more accurate the first time.

The Cochran review states, “Because of the moderate effectiveness of neuraminidase inhibitors, we believe they should not be used in routine control of seasonal influenza.” The first half of statement is a fact, the second half is opinion.  The ‘we’ evidently being “Tom Jefferson, researcher, Mark Jones, statistician, Peter Doshi, doctoral student, Chris Del Mar, dean; coordinating editor of Cochrane Acute Respiratory Infections Group.” It is not a ‘we’ that I would use for deciding medical treatment. It is one thing to say that in healthy people from age 14 to 65 treatment is modestly effective, quite another to extrapolate that information to everyone, young and old, healthy and ill, pregnant or not.
If large populations are ill one less day during widespread disease, the positive effects of people returning to school and work can be enormous. Whether olsetamivir is worth the cost and the breeding of resistance requires a complicated cost-effective analysis that I will never understand.

In the Atlantic article they confuse the treatment of seasonal flu in studies in young(er) patients when there is some vaccine immunity and relatively low virulence with preparing for a new pandemic strain with no vaccine, and what appeared at first to be an fearsome mortality rate. In hindsight, now that we know the virulence of H1N1, we did not need to stockpile the oseltamivir for H1N1. If it had continued with the high mortality rate and, based on the studies mentioned above, we would be glad they stockpliled. And if avian flu becomes infectious while maintaining virulence, well, lets just say I am practicing not inhaling for a year.

Does oseltamivir prevent complications? Maybe. But that is not a primary endpoint in treating infections. You want to cure the patient, shorten the illness and prevent death. Oseltamivir does this. If it prevents complications, so much the better, but if it doesn’t I, and my patients, can live with that.

The Atlantic article partly concludes:

“There are a couple of take-home messages here. One is pretty obvious: Tamiflu may not be doing much good for patients with the flu who take it, and it might be causing harm.”

So disingenuous since this is not the conclusion of the Cochrane review. It depends on who you are treating and what strain. It seems that understanding nuance is not part of the Atlantic’s oeuvre. Or even reading the primary sources.

The other part of the Tamiflu kerfuffle is more difficult to discuss because it concerns information we do not have.

As the Cochran review alluded to the fact that they wanted to get original data about the ability of olsetamavir to prevent secondary complications of influenza but the company would not release the information.  Note. Secondary complications. Pneumonia increases the risk of heart attack.  Treating the pneumonia with penicillin does not prevent the heart attack, but that does not make the treatment of pneumonia less beneficial.  Only living people can get a heart attack anyway.

“Attempts to deal with these shortcomings were unsuccessful: although three of five first authors of studies on oseltamivir treatment responded to our contact, none had original data and referred us to the manufacturer (Roche), which was not able to unconditionally provide the information as quickly as we needed it to update this review.”

But the Atlantic was more, well, descriptive

“The dog ate my homework
But when the Cochrane team, led by Chris Del Mar, from Bond University in Australia, re-examined the studies they had previously used in 2006, they found some discrepancies. It turned out that only two of the ten studies had ever been published in medical journals, and those two showed the drug had very little effect on complications compared to a dummy pill, or placebo. So the Cochrane reviewers decided to look at the data for themselves.

First they went to the lead authors of the published studies—the researchers who were supposed to have access to all of the data. One author said he had lost track of the data when he moved offices and the files appeared to have been discarded. The other said he’d never actually seen the data himself, and directed the Cochrane team to go directly to the company.

Four months and multiple requests later, the Cochrane researchers had a hodgepodge of data from the company, including two studies that showed the drug was ineffective, but which the company had never published. Roche also provided data from a third study, which involved 1,447 adults and adolescents aged 13-80, the largest study of the drug ever conducted. Yet the company never published that one either. (A summary of this and other studies is available at But with only partial data, the Cochrane team couldn’t even figure out what the study had been intended to measure.”

That is a problem.

One of things I have learned in blogging and podcasting is how limited and unimpressive meta-analysis and structured reviews are. They do pool the best studies. But often it seems that the process of choosing the studies, much important and relevant information is not considered. Most of infectious diseases is not based on randomized, placebo controlled trials and does not need to be.  After reading the influenza Cochrane reviews I am starting to  understand the point behind another BMJ meta-analysis.  One would have to be a wackaloon at the most bizarre fringes of medicine to demand that I treat endocarditis or meningitis on the basis of such a trial. Other diseases? Not so much.

There is a bit of irony in the whole process. The first Cochrane review of oseltamivir that suggested benefit from oseltamivir was published in 2005.
One would have thought they knew what they were doing since the lead author is touted by the Atlantic as the master of the influenza literature.
Subsequently, a Japanese physician wanted to know more about the details of effect of oseltamivir on complications. Part of the data to demonstrate that oseltamivir is effective is from an article entitled “Impact of oseltamivir treatment on influenza-related lower respiratory tract complications and hospitalizations.” which was a summary of 10 unpublished trials done by the makers of olsetamivir. The review suggested that oseltamivir halved hospitalization.

The Cochrane review, which had used the published data in an earlier meta analysis, decided that they needed to be more rigorous this time. Why they didn’t bother, with their alleged mastery of the literature, to be this rigorous the first time I am uncertain, and it casts a sliver of doubt in my mind as to the rigorousness of the influenza vaccine meta-analysis, so touted in another Atlantic article discussed on this blog.

Turns out Roche tried their best to not release all the data. First they asked for the Cochrane reviewers to sign a non-disclosure contract, then said they were giving the information to another group instead. Why the majority of the studies were unpublished were explained by  lame excuses.

“It begged the question: why were so many of the trials still unpublished and not easily accessible?
When the BMJ expressed concern to Roche that eight of 10 treatment trials were unpublished and therefore unverifiable by the general medical community, Roche said that the additional studies “provided little new information and would therefore be unlikely to be accepted for publication by most reputable journals.”
They also added that now it is standard practice for Roche to publish all its clinical trial data, but this was not standard policy within Roche or elsewhere within the industry seven to 10 years ago. “At the time, it was considered that the studies that were published (2 abstracts and 2 full manuscripts) reflected accurately the benefits of the drug,” they said.”

Good questions. Given the history of pharmaceutical companies hiding and obfuscating important data about their drugs, it would be nice if they released the data. Most drug studies are funded by pharmaceutical companies. Does that invalidate the study? No. A drug company study can be just as well done as one funded by an NIH grant. That outcomes will be slightly tilted in favor of the drug if it is funded by the company as compared to non drug company funded studies is recognized. Some studies are well done, some studies are little better than infomercials. In the end you have to read the literature, not the reviews. So I can’t comment on the validity of olsetamivir to prevent complications.  Neither can the Cochrane review.  Just the Atlantic, who concludes that it is all a mirage.  So much sound and fury, signifying nothing.

Then, for each study, there is what the results of the study are, there is the spin put in the discussion, and finally the spin that the drug company reps put on the study when they show it to a doctor.

Kind of like a Cochrane review, huh?

The review says Tamiflu is modestly effective.
The conclusion says “Because of the moderate effectiveness of neuraminidase inhibitors, we believe they should not be used in routine control of seasonal influenza.”
Fact, then Spin.
And the drug rep, er, I mean Atlantic says it is nearly worthless.
Lots of unjustified spin.
Remember the conclusion of the review, “Evidence for or against their benefit for preventing complications of influenza is insufficient.”
Pretty mild. The BMJ has an nice discussion on the information, as such as can be determined, from the studies on prevention. By themselves they showed no efficacy but did (may be) when combined. But with no access to the data, there remains uncertainty.

Do drug companies hide and spin important information? Do drug companies inflict bias into studies? Do drug companies influence doctors prescribing habits? Is homeopathy nothing but water?
Yes. It is why you need to consider the entire medical literature.

But to quote the ever helpful Dr. Hill

“All scientific work is incomplete, whether it be observational or experimental. All scientific work is liable to be upset or modified by advancing knowledge. That does not confer upon us a freedom to ignore the knowledge we already have, or to postpone the action that it appears to demand at a given time.”

Like evolution, there are multiple lines of evidence to demonstrate oseltamivir efficacy against influenza. When it should be used to get the most bang for the buck depends on the strain of influenza and the patient being treated. And when (not if) we get a pandemic influenza that is both highly infections and highly virulent, I hope we will have a drug like olsetamivir. It will not be a panacea, like penicillin for syphilis. But it will keep some people alive who would otherwise die. The Atlantic concludes

“The more important issue, however, involves the need for trust in science and medicine. Governments, public health agencies, and international bodies such as the World Health Organization, have all based their decisions to recommend and stockpile Tamiflu on studies that had seemed independent, but had in fact been funded by the company and were authored almost entirely by Roche employees or paid academic consultants. So did the Cochrane Collaboration, at least in its earlier assessments of Tamiflu. Millions of flu patients have taken the drug as a result.”

We also need the 4th estate to write and publish quality articles on science and medicine and not go for the easy story of good and bad and obscure importance detail in inflammatory prose. It would be nice if the 4th estate took the time to, oh, maybe actually understand the nuances involved in influenza prevention and treatment. The Atlantic is 0 for 2 with their reporting on influenza. At least I get a topic to write about, and, like their spiritual colleagues Dr. Mercola and The Natural News, the worse the reporting, the better the blog.

Posted in: Pharmaceuticals, Politics and Regulation, Public Health, Science and the Media

Leave a Comment (36) ↓