Health Care Renewal is an excellent blog with a special interest in medical conflict of interest issues. Last week one of the HCR bloggers, Roy Poses, posted about "The Threat of Pseudoevidence-Based Medicine." The occasion was a article by Smith in the UK journal Clinical Governance (2007; 12: 42-52), which I don't have access to, but there is enough in the excerpts and commentary at HCR to get the basic idea.
We hear a lot about Evidence Based Medicine (EBM), using the best available evidence to inform clinical practice. The gold standard for evidence is a randomized clinical trial, although other well done observational studies are also important in EBM evaluations. But the worthy goals of giving priority to certain kinds of evidence can be perverted. The example of pseudo-evidence is an example. Here's a pull quote from the Smith paper, as I found it at HCR:
Another, perhaps not new threat to the practice of EBM [evidence-based medicine] has been discovered -- pseudoevidence-based medicine (PBM). PBM can be defined as the practice of medicine based on falsehoods that are disseminated as truth. Falsehoods may result from corrupted evidence--evidence that has been suppressed, contrived from purposely biased science, or that has been manipulated and/or falsified, then published. Or falsehoods may result from corrupted dissemination of otherwise valid evidence. These falsehoods, when consumed as truth by unwitting and well-intentioned practitioners of EBM, then disseminated and adopted as routine practice, may well result not only in inappropriate quality standards and processes of care, but also in harms to patients.
EBM rests on the premises of professionalism in science and medicine. EBM presumes that evidence is produced by scientists who strive to be objective. EBM presumes that those producing evidence have no pre-conceived hopes or goals for what the evidence will show. EBM presumes that producers of evidence have no stakes in what the evidence will show. EBM presumes, or at least strives to assure, that the scientific evidence-production process is free manipulation by people with vested interests with goals other than improvement in patients? mortality, morbidity, or quality of life.
There are reasons to believe EBM's presumptions are in question, and that PBM is a "new" threat to EBM. Only two conditions are necessary for PBM to flourish. First, one link in the chain of evidence production, assembly, or dissemination must be purposely corrupted, resulting from a compromise of professionalism in science and medicine. Second, the belief must be promulgated that a given piece of evidence is true and of the highest quality possible, when in fact it is tainted. (Smith WR, Pseudoevidence-based meidicne: what it is, and what to do about it. Clinical Governance 2007; 12: 42-52 via Roy Poses at Health Care Renewal)
Poses likes the term pseudo-evidence. Whatever you want to call it, it's something to keep in mind. After ignoring the problem of influenza for decades because it wasn't profitable, a number of companies have now bet heavily on the profits that might be realizable with a pandemic threat. At the moment it isn't clear how, when or if the bet will pay off.
One way to improve your odds is to have evidence your product is effective. For that purpose pseudoevidence works just as well as evidence.
Something to keep in mind.
- Log in to post comments
More like this
Great post. EBM relies heavily on the practitioner's ability to assess the science; once the "evidence" is accepted practice, it'd require time and effort to change.
BTW: Second paragraph of the excerpt-"the scientific evidence-production process is free manipulation"? (Wouldn't surprise in some cases)
Hi - One of the things I emphasize with my students is that human being are intrinsically biased creatures. We can no more be unbiased than we can forgo breathing. There is a lot more to it than wheather the corporation is buying the results with funding or global warming is politicized with ideology. There are however, greater and lesser degrees of bias. What the scientific project should emphasize, I think, is conducting experimental research with as little bias in it as the investigator can manage. I think we can still cleave a distinction between experimental science and what amounts to natural philosophy. We are often, I believe, to loose with the term science. I teach political science but the science part of it is problematic at best. What separates, to a great degree, science and art or philosophy, is experiment and experimental evidence. That goes back at least to Francis Bacon and is one of the pillars of the enlightenment. I think it is experiment that produces evidence and pseudo-evidence might be characterized in part, by an inability to replicate the alleged evidence of the pseudoists (pseudoist? I think I just made up a word :-) ).
I wondered if this is in reaction to any particular new development in the bird flu front--like maybe the way the mainstream media is promulgating Webby's mouse experiment showing that around 21% of H1N1-immune mice have resistance to H5N1 to mean that 'some people may be immune to bird flu'.
In every pandemic, some people have been inexplicably immune, and maybe this observation begins to explain some of that phenomenon. But as you've pointed out, that is in *mice*, and as I'd like to point out, 21% isn't all that great a number.
That is unless you are a mouse! Kidding of course. The 21% even if it translated into human survivors Lisa would only be 63 million left in the US all things being equal. Even with the drain on the resources gone those 63 million would be struggling for years just to survive.
Someone always survives and everyone wants it to be themselves. My bet is that if and when it comes as it wades into us, the last thought by so many will be just one thing....astonishment.
One could well argue that artificially capping the worst case scenerio of a bird flu pandemic kill rate at 2% (as in 1918), is a prime example of PBM.
all: On the cross immunity vaccine paper, forget the 21% number. It is an estimate and pertains to mice and based on very small numbers. We don't know what the comparable number is for humans. It might be higher and it might be lower or zero. This paper opens up a line of inquiry of genuine interest, that's all. It also suggests there might be benefit from seasonal vaccines that have an N1 component.
It's a data point. Regard it as such.
Lisa: If you mean a reaction by me, the answer is no. It is a general comment that occurred to me on seeing the HCR post.
Moeb-Yep, someone isnt reading the numbers. 5% would be devastating, the 8% catastrophic. The current 83% one year rolling rate is unbelievable in what it would do. Wouldnt have to worry about oil anymore, no one around to pump it.
I noticed that news now has this blog among the top stories on avian flu today. Just for reference in case any 'flu newbies' are reading this, the current case fatality rate among those who are confirmed by laboratory to have caught 'bird flu' is 61%.
I mention this because a poll on a non-flu board I frequent recently showed that the vast majority of non-flu-watchers think that the death rate is under 10%, probably because most of the pandemic planning documents have used the 1918 death rate as the worst case scenario.
Now, the current virus is not (yet?) a pandemic virus, and we know that the precursor to the 1918 virus had a 60% mortality rate in 1916 in one limited outbreak of about 150 people, before it became easily transmissible. But the point is, we don't have any scientific reason or explanation that would *require* a drop in mortality rate for a flu virus to become transmissible. Since we don't have that level of understanding, for planning purposes, we cannot entirely rule out the possibility that the currently observed death rate could persist, even though past pandemic viruses have dropped back from their early-mortality highs.
Possible does not mean probable. But it should remain an element in our thinking.
Revere said:
"On the cross immunity vaccine paper, forget the 21% number. It is an estimate and pertains to mice and based on very small numbers. We don't know what the comparable number is for humans. It might be higher and it might be lower or zero. "
The mainstream media is disseminating this as if it were immediately generalizable to humanity. Maybe a few 'letters to the editor' are in order.
Kruger is probably right about the death rate on a more recent basis--it has been higher recently. The overall numbers include Turkey's high survival rate of 67% last January; more recent figures include Indonesia and North Africa's rather dismal outcomes. It is hard to tell to what degree the difference in survival is due to differences in the virus itself or due to differences in healthcare.
Lisa, where was that 1916 outbreak? I have been tracking H1N1 for a long time now along with its path and merging it up with the suspected flyways of the birds back then. Schenzen had an outbreak of something in 1900 during the Boxer Rebellion. Took many people and the mandarins recorded it as a fever and pneumonia episode but with the backdrop of the Europeans carving up the country it was kind of lost. The Catholics recorded something else in and around Hong Kong in 1910 but wasnt considered to be a big thing. About 2000. Next thing you see something happening in France in the Ardenne during 1914-1916. It was indeed a contagious pneumonia of some kind but attributed mostly to trench warfare. My grandfather was in those trenches and wrote letters to my grandmother about how bad it was. He would volunteer for sentry duty just to stay away from the sick. Bad enough that someone is shooting at you he said, but going to sleep with old Bill at your back, coughing and hacking meant you likely would be leaving for the French hospitals.
John Oxford, in his review in Nature of Michael Greger's book "Bird Flu, a Virus of our Own Hatching", states:
"However, the book fails to confront the question I am asked daily: "Why are you so worried about 151 deaths from H5N1?" Well, go back to 1916, to Etaples in northern France, where a form of flu causing heliotrope cyanosis (a characteristic lavender coloration of the face) with a case fatality of 60% was beginning to spread. There were 145 cases. At some point in the next two years it mutated to become more infectious and 30 times less virulent. Then it killed 50 million people. Doesn't this ring a nasty bell?"
That is where I got the information, though I'm sure John Oxford got it from a more scholarly source than a book review.
This new blog has the objective is stimulating discussion of a universal means to destroy any virus, newly-evolved or existing, so that viral infection may no longer afflict the human species
http://virology.wordpress.com/about/
The media and interested kibbitzers have latched onto 21% as being an 'official' number and intuitively 'more accurate' than CDC's (perceived) official worst-case "Cat 5, the experience of the 1918 spanish Flu, whose CFR has been roughly estimated to be around 2 - 3%".
http://scienceblogs.com/effectmeasure/2007/02/the_new_cdc_pandemic_resp…
Remember, the numbers are totally without meaning to them, except that bigger is scarier.
I would argue that "purposely corrupted" applies here, although the purpose is media sales rather than medicine sales.
Lisa thanks. Those I knew about and the source was the French government. There was a lot of concern about it because the Grippe was aparently mingling at the time with another form of it. The purpling you described made doctors suit up in their PPE (muslin cloth) and sprayed the areas where they were with phenolic. Early decontamination... The army was a lot less pragmatic about it. They thought it was a German bioweapon or speculated it as such. Their bent though was that the medical high command said that if it were it would be quantum leaps ahead of anything they were working on. Bioweapons....Sheesh.
Given that viruses weren't even discovered until 1926 and DNA as genetic material until the 1943, it is a fair bet that viral genetic manipulation in 1916 was beyond the Germans' capabilities.
It's very hard to pinpoint military disease reports of the distant past.
Until WW2 more soldiers died of disease than combat injuries. Epidemics of a zillion diseases sweep through armies on the march or "stuck in traffic". This doesn't mean they're all pandemic flu.
Reporting on the ground is poor, partly because they didn't know what to look for. Anyone with a compromised immune system probably got nailed with two or three other diseases immediately. It's also possible that other common diseases have changed over the years and present with slightly different symptoms than they did a hundred years ago.