It's flu season. Human flu, that is. Also, it seems, flu in poultry. So if someone comes down with high fever, aches and pains and a cough in an area where there is H5N1 in poultry, is it likely to be bird flu? The answer, so far, is "No." The reason is fairly straightforward, although this is counter-intuitive for many. First, the empirical evidence (from Thailand), then the explanation.
Nearly half of the patients on the bird-flu watch list have, in fact, caught human influenza, the Medical Sciences Department (MSD) disclosed yesterday.Since bird-flu infections among fowls were detected early this year, hospitals have put 424 patients with symptoms similar to avian flu under close medical surveillance pending laboratory tests.
Results for 373 of the patients are already available. MSD director-general Dr Paijit Warachit yesterday said up to 179 had tested positive to human influenza.
Most human-flu patients were young children and middle-aged adults. (The Nation, Thailand)
In this case, none of the 424 have so far tested positive for bird flu, but that's not the point of the exercise. Even if 42 out of 424 tested positive for bird flu, the probability of anyone with an influenza-like-illness (designated ILI) having bird flu would still be only 10%, i.e., nine out of ten of the suspect cases would not be bird flu.
Here's the explanation. During flu season there is a lot of seasonal ("human") flu around at any one time (i.e., influenza A). Not only seasonal flu, but respiratory syncytial virus, influenza B, adenovirus and parainfluenza, all causes of ILI. Without bird flu, all cases of ILI are one of these, amounting to hundreds, thousands or tens of thousands of people, depending upon the underlying population being considered. Now take H5N1. The experience at the moment is that transmission from whatever its reservoir is in nature (for the sake of argument we will adopt conventional wisdom and say it's sick poultry) is uncommon. Despite tens of thousands of people with encounters with sick birds, only a relative handful of infections have resulted.
True, that's just "so far" and this could change if the virus became more easily transmissible. But in the absence of that happening, during flu season there will be hundreds of thousands of ILIs from the usual causes (there always are) but only a few or none from H5N1. Thus the chance of any particular "suspect" ILI being H5N1 is very small (in the present circumstances).
This doesn't mean we shouldn't keep looking. Bird flu cases have features that set them slightly apart from other ILIs, principally that they run a rapid and severe downhill course. "Ordinary" human flu can do the same thing, but this is still a red flag. The other red flag is the presence of the virus in the environment, which for now we become aware of principally by its effect on poultry. The two together should raise the index of suspicion, and once a human case is detected, surveillance of other ILIs is indicated as an early warning for the virus having changed its transmissibility. When we do this, we should be ready to expect that the tens or hundreds of "suspect" cases we see referred to in news reports will mostly be non-bird flu cases, even in the instance where the virus has increased its transmissibility. Some warning signs of an incremental increase in transmissibility would be more small clusters (indicative of easier bird to human transmission) or larger clusters (indicative of human to human transmission).
Disease surveillance is both art and science. Sometimes the science seems counter intuitive and sometimes the art allows us to see things we wouldn't ordinarily be aware of. Experience shows that setting up routine surveillance systems (and watching their output) is the best strategy. Unfortunately investing in unglamorous surveillance systems is not what politicians do best, and public health officials, pressed by urgent needs and insufficient resources, have also allowed these essential infrastructure components slip in priority.
Meanwhile, we'll continue to do what Claude Rains advised in the movie, Casablanca. Round up the Usual Suspects.
- Log in to post comments
Revere! WELL SAID!
I am the first to admit, that we need to keep a close, watchful eye on this situation. But it must be noted, that human contact with carriers is probably quite high, however human infections appear rare (thankfully). It would appear, that it is becoming more easily transmissible from whatever host, to human....but not drastically nor definitively. Higher infections could simply mean more contact with a growing number of avian (or alternative) hosts.
Furthermore, literally WORLD attention is focused on each case and each outbreak. In the big picture, the human number is small, compared to other dangerous diseases. I realize the risk here is mutation toward more efficient transmission. But that remains only a possibility at this point. Although not likely, if this virus were to remain stable, we could live with H5N1 for a long time.
It's important too, to understand this virus could itself mutate AWAY from more efficient H2H transmission. Correct? And recombination stands a good chance of watering down the virulence.
These factors, make it easy for some folks to call us alarmists. It's important to keep things in perspective. This as been my point, all along. We gain nothing, by pointing toward the sky and wringing our hands. Keep things in perspective.
I enjoyed reading your post.
What's frustrating on the surveillance side is that the testing for H5N1 seems to be unreliable. In Indonesia in particular, the doctors seem to be testing and re-testing the patients that they believe to be infected, and still coming up with negatives. These "boots on the ground" appear to have a good grasp of the clinical presentation, but the test results don't seem to be bearing their suspicion out. It still seems to be the case that most of those who are testing negative are those who have died.
Mara, if you mean most of those testing positive are those who have died, I'd agree with you.
I don't know if that's because they can be more invasive in taking samples from a dead body, or because they are just that much more suspicious when an otherwise healthy young person dies of an ILI that presents like H5N1. But it can be hard to credit reports of close family members of confirmed H5N1 cases being direly ill, hospitalized for days, but testing negative. Some of them doubtless are negative, but considering it took literally dozens of tests for some cases in Turkey to finally be confirmed last January, it's hard to believe all of them are.
And then there are all the cases for which we don't have a lot of info, don't know why they were suspect, or how long it was before they were sent home. And the way some officials seem to reflexively deny bird flu, only to have it later found (China, Indonesia, and Nigeria all spring to mind).
I accept and fully believe Revere's statement that most suspect cases are not H5N1. A list of hundreds of suspects, particularly in countries facing active bird outbreaks, is not necessarily worrisome in itself. But for all the reasons I listed, the picture is murky enough that nonspecific suspicion remains.
Caia - Yes, you put it much better than I did. I agree with Revere on both points: (a) that most of the cases are not H5N1 and (b) surveillance is crucial. However, I don't have a lot of confidence that the surveillance is rendering reports that are accurate.
This would be good reading for the folks tracking Indonesia on the fluwiki discussion board.
Quickest way to get shredded over there is to suggest that the high number of negative tests, might, in fact, be true negatives.
Lisa GP - AS good as Fluwiki is, I can't go there anymore, for that very reason. Too bad too, cause it't a great site for BF info. I fear the hard-cores are scaring many reasonable, information seeking people off.
Lisa and Patch,
Maybe you should check FluTrackers thread on Indonesia reassortment. Here's a link
http://www.flutrackers.com/forum/showthread.php?t=15934
After assessing all the recent genetic changes in the H5N1 virus in indonesia (I don't pretend to follow all the genetics with any depth of understanding, but do get it generally) the analysts there are wondering out loud whether all these mutations might have caused the old PCR tests to no longer be effective because they no longer match the new viral sequences.
When you write as if those who question the validity of those negative tests are a bunch of hard core fear mongers, you purposely ignore the quite reasonable arguments they give for questioning the certainty of the tests. As a matter of fact, most are only saying be open to the possibility that some of these "negatives" may be in error, and don't throw out the cases yet until more data has accrued. This is actually a highly reasonable investigative stance. There is ample genetic evidence that the strain infecting humans in Indonesia is significantly different from the strain infecting poultry, and now these most recent changes in the human strain give even more reason to question how accurate the PCR tests might be. IMO, it is actually less reasonable to decree all negatives are indeed negative, while ignoring clinical data, statistical data and this genetic data.
False Negatives can be caused by Tamiflu usage. It could be caused by testing too early. It could be caused by the markers moving off the radar of the PCR tests. It could be due to sloppy lab work. There are many reasons we are seeing so many negs. We've seen it before... over and over again. There is alot of pressure to find Negatives... very little pressure to find positives.
BUT it is very good that these cases are being tracked and tabulated and mapped. It is giving us insight into what's going on in indonesia - on the ground. If we wait for WHO to confirm cases, we will most likely be up to our eyeballs in H5N1 and too late to do anything to reduce the Pandemic's effects. The trackets at Flu Wiki are doing an amazing job. Cheers to them.
Goju: You miss my point. Yes, FN are not uncommon. But to be a FN it has to be a positive first. My argument is a simple application of mathematics, not molecular biology. If there are a lot of ILIs from human flu and non-flu viruses (and there are, depending on the base population, thousands), then even if there is a marked increase in transmissibility, most cases will not be H5N1. Think about it. If there are a thousand ILIs in an area at any one time without the presence of H5N1, they will still be there. If there are ten true H5N1s, then only 1% of the ILIs will be H5N1. If there were even 100 H5N1s among them, then only one in ten would be H5N1. By this reckoning, forty or fifty suspect cases would mostly be non H5N1. It's arithmetic. This is the same as any screening procedure. The positive predictive value depends on the prevalence, not just the sensitivity of the test (which is what you are raising).
I do see your point - and it's worth making, because there's lots of good evidence that even doctors neglect prevalence information in their diagnoses, causing serious errors - but it's unfair to talk in terms of ILIs as though the FW trackers were tracking every case of an ILI. They are focusing on cases where the doctors on the ground suspect H5N1, and generally on people who are very seriously ill. I'm not a tracker but I have the impression that there are a lot more seriously ill people with ILIs in hospital in Indonesia than there are in a typical year. That's epidemiologically interesting, surely. Bad seasonal flu year, combined with bad dengue year? One hypothesis. Lots of unidentified H5N1 cases? Another hypothesis. We could wait for the authorities to work it out and tell us. Or we can track the news. Some people prefer to "do something", and about all we can do is to track the news.
Pulling the rug out from under the trackers feet is bad business as far as i am concerned. I'd rather be nervous watching a map of growing suspected cases and having them prove negative than not seeing it at all.... which is what will happen if the trackers feel they are wasting their time.
I for one am very thankful for their very hard work... at some point "they" will show us "IT" has begun. If no one is manning the lookout posts, the enemy will be upon us while we are sleeping. The enemy is at the gate trying to get in... lets not forget that.
Lets also not forget that the CDC has pulled 2 weeks out of the testing time for any reverse engineered or other vax trail... fast tracking it gives us a clue as to the expectations of TPTB. Whats the usual time for Vax tests? Whatever it is, it seems that we may not have enough time. (hat tip to Pixie)
Math, Goju: If you thought my point was that we shouldn't track these cases then you really did miss my point. I explicitly said that surveillance is a key activity. I was only explaining why when we do look at suspect cases we usually don't find much H5N1. We shouldn't expect to. That's not a reason not to look. It is a reason not to come up with all sorts of explanations for why the results must be wrong. Some of them may be, but overall, this is what we would expect to find, even if H5N1 was being transmitted to others and we are missing some of it. As a co-founder of Flu Wiki and someone who has repeatedly noted the importance here of doing what is done there and the other flu boards I don't think I should have to explain this. My purpose here is to explain what is being seen as well as I can.
have there been false negatives to date?
goju: Yes. Quite a number. It is common in clinical testing, especially in settings like this. And as some have pointed out, there is an incentive not to find things. Regardless, most ILIs at this moment will be non H5N1. If there is a pandemic, then that no longer holds true, although even then the other causes of ILI will co-circulate so many ILIs won't be pandemic flu.
As I understand it, different flu strains enter a system and "take over" the turf, so to say.
Does H5N1 have to "push out" H1N1 in order for it to go Pandemic?
goju: No. flu strains can co-circulate, especially if they are different subtypes because there is less cross reactivity.
Before a pandemic begins, there will be a few false negatives.
In the early onset days of a pandemic, there will be a few more false negatives.
Once the pandemic starts in earnest, there will be lots of false negatives.
These PCR tests are like the early days of radar- a great, inaccurate, unreliable, slow idea. Kool, but better to have people on the coast lines with binoculars. I am sure that in 5 or 10 years, we will have quick reliable ways to identify flu bugs.
In 1918, a flu pandemic swept the globe. At that time, nobody had even seen a virus- or had any idea that they existed.- despite that, the diagnosis were made- often the color of the skin.
The FW newshounds are going to be way ahead of everybody else when this thing finally comes out of its den. They are the main reason I visit FW.
A flu virus has values, they are just not human values. We will not be able to talk our way out of this. H5N1 views us with exactly the same respect and concern that we have for sheep, oats, carrots and plums- just some random thing to eat. I just pray we are not the main course.
A simple question: for other flus, and ILIs, are family clusters of 50-70% mortality common? "Oh, the whole fam dammily died of norovirus. Pneunomia. The vapors. The heebie-jeebies."
If false negatives, or undiagnosed cases, are rampant there should be these everybody dies including the nurses, doctors, and the lunch truck guy clusters for whatever else is circulating in Indonesia.
Would this not be the smoking gun?