According to news articles, the fatality rate of H5N1 is about 50%. I understand that this is with the benefits of modern medical care including ventilators and medication. Does anyone know the survival rate of patients who do not receive medical care (as would be the case for the majority in the occurence of a pandemic)? Also, is there any information on H5N1 vicitms who might develop a mild case and therefore do not seek medical treatment? And how does this factor into the fatality rate?
So far, I’d guess that’s a “known unknown”;
they are not testing (through lack of resources/and will?) all the populace in a given range around a bad case to see who has been exposed, they are not testing people who get sick but are not brought to the hospital whether they live or die, and you know they aren’t even testing some of the people who did come in and die. Cremated immediately after death; can’t test. There isn’t much health infrastructure in some of these places, and even places where we have health infrastructure, they aren’t testing much because they don’t think there is a risk, despite people coming off planes from all over every day.
Probably many people would like to know the info you seek; but how to get it??
Some governments won’t even let outsiders come in and test everyone, even if the international community wanted to spend money to do such a big study. Maybe they’ll get some info later as it goes into other places, but I don’t think anyone can figure the true fatality rate yet.
IMO the case fatality rate, the CFR (this is not really a rate in technical terms but a proportion) is overestimated. The numerator, the deaths, are relatively easy to ascertain (although not completely), but the denominator is not. It is the total number of cases infected, whether diagnosed or not. Thus it is likely underestimated, perhaps by a large margin, meaning that the CFR is overestimated. Since we don’t know how many mild or inapparent infections there are we don’t know the true CFR. If I had to make an educated guess, it would be 10% at this point. I would also expect this to drop as the virus adapts to humans but this is not an ironclad guarantee. Virulence could even increase.
The likelyhood of the virulence remaining the same or increasing while the virus adapts to humans (if it does) is very low.
revere: I know you and most epidemiologists believe that the true CFR is lower than the official CFR. And I understand that severe cases are easier to ascertain than mild ones. However, Thailand has a very extensive surveillance system which investigates all possible bird flu cases, including mild ones. In 2005, they examined 3,199 cases. They found 5 cases, 2 fatal. The WHO has reported that numerous investigations of families, HCWs and even whole villages have failed to identify large numbers of missing cases. In order to achieve, a 10% CFR one would have to assume that there have in fact been 760 cases, right? [Assuming that every fatality has been identified]. This would mean that 616 cases have not been detected. Thus, for every case detected approximately 4 are missed. Since most of the clusters have high fatality rates, can we assume each cluster should be multiplied by 4 to get the true cluster size? That might be a very interesting number when calculated for the Turkish clusters.
Although I think some mild cases probably are being missed, I would hypothesize that these will be among older people. For people under 25, I suspect this virus is as lethal as it appears.
Ceaser: What is the basis for your statement?
Monotreme, I am inclined to agree with you.
We can’t just consider unreported mild cases without considering unreported or uncounted deaths. There have been quite a few connected with the clusters that we DO know about, eg the young girl in Jakarta who died and was cremated before her father and younger sibling became infected. These two were tested and one was positive, but the chance of the index case and the sibling having died from anything other than bird flu is miniscule. So in this cluster we only counted one case out of three. Another cluster where a confirmed case (teenager in ?Indonesia) had 2 brothers who had died the previous week. These would not have come to our attention if they were not connected to known cases, so how many of these deaths there are is anybody’s guess.
I am not convinced that the current case fatality rate is anywhere near 10% unless we see results of epidemiological surveys showing a significant number of previously undetected positive cases.
anon_22, the last thing I want is someone to agree with me on this! I will have to go to plan E if the fatality rate is above 10%. Unfortunately, I have seen no convincing data that the rate is significantly less than 50%, especially in the under 25s, dammit. You are quite right about uncounted fatalities, again usually among children. I fear the real mortality rate in the under 25s currently really is 50% (or higher without Tamiflu and ventilators).
As for a drop in virulence as H5N1 increases its virulence, no sign of that. Also, no mechanism is proposed for this to occur. Still hoping, but hope is poor substitute for data or theory.
The last paragraph above should be:
As for a drop in virulence as H5N1 increases its transmissibility, no sign of that. Also, no mechanism is proposed for this to occur. Still hoping, but hope is poor substitute for data or theory.
There seems to be a theory going round that a pre-condition for a pandemic is that the case fatality rate has to come down. While this might be true to the extent that dead or severely ill people are less able to infect large numbers of people because they tend to stay put or not live long enough, the question is how low does the CFR have to get before significant transmission occurs that is sufficient to sustain H2H to result in a pandemic. While I’m not an epidemiologist, common sense tells me that in a disease as contagious as flu and where you get a second generation of cases in as little as 2–3 days, even a 50% CFR could still easily cause a sustained chain of infection as long as there is efficient h2h.
One hopes of cause that the CFR will come down, and I don’t personally believe that it will stay at 50%. But for the purpose of planning, even a CFR of 2% in a pandemic is potentially catastrophic. eg with a attack rate of 25% of population, a CFR of 2% for New York and suburbs would be 22M x 25% x2% = 110,000 deaths
(see http://www.fluwikie.com/index.php?n=Forum.WhyYouShouldTakeThisSeriously)
However way you look at it, if we have a pandemic due to H5N1 within the next 12–24 months, chances are we are looking at an extremely grim situation.
Thus for me it really doesn’t matter whether the current CFR is 50% or 35% or 65% or 20%, the bottom line is unless we see sure signs that it has gone below 2 digits WITHOUT CAUSING A PANDEMIC (i.e. before the virus achieves efficient h2h), my assumptions for pandemic preparation remain unchanged.
Monotreme, “the last thing I want is someone to agree with me on this!”
I know.
hat’s one of the reasons why I post here and talk to people. I am constantly trying to find someone to convince me that my numbers are wrong.
I hope to God you will continue to post. I need a touchstone, I guess, and I think you’re it.
keep watching the news in Turkey we have so far 3 deaths in 41 cases .. Those dead children were sick for 10 days before getting to hospital …
41 cases? Do you have a link to this information? I would like to read the full article. The thought of 41 cases in such a short time is more than a little disturbing.
I assume that is 41 suspected cases. Does any know how many are confirmed?
(Would sick for 10 days mean secondary infections, rather than cytokine storm/ARDS?)
You would think.
anon_22: I’m here for the same reason. I never get direct rebuttal to my points, just “it can’t possibly be that bad because its never been that bad before”. We all have defense mechanisms, but mine isn’t working.
We won’t know the true fatality rate for weeks if its spreading human to human.
On a relative scale, I’m less worried at the moment with the fatality rate than with how efficient is the h2h. As I noted above, even a low-ish fatality rate (eg single digit) in a PANDEMIC is catastrophic. Paradoxically, I’m somewhat concerned about complacency that will result if people think that the fatality rate is perceived to be dropping eg from 50% to 20%.
Maybe we should emphasize the real meaning of something like 25% attack rate with 2% fatality (as previously posted). Maybe we should start shouting these figures like: What would happen if 110,000 die in New York or 35,000 die in Hong Kong. When I have used these figures with people, they seem to get their attention better.
It occurs to me that estimating a fatality rate at this early point would be complicted by several reasons:
- Early cases being less likely to be recognized and treated promptly when the patients could be saved (as with the Turkish children) - Poverty, lack of access to health care stopping people from seeking help promptly
So you’d expect mortality to go down as awareness increases and barriers to prompt treatment are addressed, no?
On the other hand, if this hits in a human pandemic wave and no one can get access to treatment anyway, you’d expect the mortality rate to rise again. Not a reassuring thought.
(…and yes, I too am watching you all closely in the hope of finding evidence that we’re all just wingnuts!)
A study which supports a much lower mortality rate than advertised:
Avian Flu Transmission to Humans May Be Higher Than Thought
A new study suggests that there is an association between direct contact with dead or sick poultry and flu-like illness in humans and that the transmission is probably more common than expected, according to a new study in the Jan. 9, 2006 issue of Archives of Internal Medicine.
Anna Thorson, MD, PhD, from the Karolinska Institutet in Stockholm, Sweden, and colleagues analyzed data from household interviews conducted in FilaBavi, a Vietnamese demographic surveillance site in Bavi district, northwest Vietnam, with confirmed outbreaks of highly pathogenic avian influenza (HPAI) in poultry, subtype H5N1. The researchers included 45,478 randomly selected people in the district to answer screening questions about exposure to poultry and flu-like illnesses (defined as a combination of cough and fever). The study was performed from April 1 to June 30, 2004.
As background information, the authors explain that in Vietnam, an epidemic of HPAI in poultry, subtype H5N1, has been ongoing since 2003, despite efforts by the government to eliminate the diseased birds from flocks. “In addition to being an important source of income, poultry is kept by many rural households for subsistence farming. The current epidemic in poultry is thus not only a public health problem but also an economic drawback for the many Vietnamese who live in rural areas.” The researchers note the Vietnam is the country hardest hit by the ongoing H5N1 epidemic, with 87 confirmed human cases of HPAI (with 38 deaths) and 1,838 verified outbreaks in poultry in July 2005.
In this study, the researchers found “a total of 8,149 individuals (17.9 percent) reported flu-like illness, 38,373 persons (84.4 percent) lived in households keeping poultry, and 11,755 (25.9 percent) resided in households reporting sick or dead poultry.” The researchers report that having poultry in the household was not a risk factor for developing a flu-like illness, but having direct contact with sick or dead poultry produced the highest risk for flu-like illness. “Low socioeconomic status, female sex, and young or old age were also risk factors for the disease,” they note. The researchers estimate that between 650 and 750 cases could be attributed to direct contact with sick or dead poultry.
“Our results from a large epidemiological population-based study in an area with an ongoing epidemic of HPAI in poultry are consistent with a higher incidence of HPAI among humans than has been recognized previously. The results suggest that the symptoms most often are relatively mild and that close contact is needed for transmission to humans,” the authors conclude.
Reference: Arch Intern Med. 2006;166: 119–123.
Gang,
We’re not nuts. While the amount we don’t know about possible pandemic flu caused by H5N1 is huge, at this point we have enough evidence to have some genuine concern and to begin making plans based on what our estimate of the risks are. In good risk communication, you examine all of the options, including the worst case scenario, and make plans based on what you believe your own exposure to the risk is. That’s what we are going here.
Sorry, no offence intended to anyone, just wishful thinking.
Ceasear: Interesting study. But this is pretty indirect. No H5N1 antibodies much less isolation of virus. I have seen several comments questioning why poultry workers are NOT overrepresented in the confirmed cases. One explanation is that they have been exposed to mild strains of H5N1 prior to being exposed to lethal strains. Since they may have been exposed on a yearly basis, they developed antibodies that cross-reacted sufficiently to prevent death, though perhaps not mild illness, which would be consistent with the study you cite. The high fatality rate in clusters and the animal studies still lead me to conclude that the currently circulating version of H5N1 is highly lethal to immunologically naive, and young, people.
Melanie: I don’t think anyone is crazy. This is a stressful time and I think the idea that a virus that kills 50% of its victims could be going pandemic soon is very disturbing. There is a natural tendency to wish this away. Scientists are human beings too and are not immune to this desire. I would very much like to see hard data indicating that the true CFR is very low. But no-one seems to have it. None of us knows what the risk is. Debating the implications of various pieces of data is healthy and potentially productive. We all know that no-one is going to say with any kind of certainty what will happen, but its good to be debating the issues as more data becomes available.
Monotreme,
As Revere and others have remarked earlier, it is impossible to know what the current CFR is. Remember that the CFR rate in 1918 was estimated to be 2–5%, which was catastrophic, though the odds of *you* actually dying from it were pretty low.
Monotreme, If forced to believe a CFR based on “your observations” or basing them on the study I’ve referenced, I’ll go with the scientific study. I mean no disrespect, however, everyone has their own “observation” that is rarely scientific in nature. In fact it’s usually biased in some why to support their predisposed notions. My point is exemplified by the actual number of people in Turkey that have tested positive for H5N1. Can you tell me what the number is based on “your observations”? Not knowing has not stopped people from throwing around various estimates and using those estimates to support the “pending” pandemic.
Melanie: I agree we probably don’t know the true CFR. However, we do have an official CFR. Revere and other epidemiologists have been asserting that the true CFR must be much lower, without any data. You just cannot do this in science, no matter how eminent you are. Assertions without data don’t count. Please believe me when I say I want the CFR to be extremely low. However, I have to go with the data we have, which is considerable. It all points to a high CFR in people under 25. This may change with additional data, but we cannot just assume this until the data is available.
There is nothing magical about the 1918 CFR and no reason why that has to be the worst case scenario. I don’t expect to die from H5N1, I will isolate. I am, however, very concerned that the necessary steps to blunt a true worse case scenario are not being taken because of a psychological barrier to accepting the possibility of a high CFR. I think this unwillingness to think the unthinkable is present in many decision-makers, both health care experts and politicians.
Ceasar: I’m not in the least offended by your post. I do not expect anyone to accept “my observations” over data. The WHO has reported that the official CFR is about 50%. The high fatality rate among the clusters of cases is a fact, not my observation. These clusters have been reportedly investigated intensively by the WHO. The extensive investigations of the Thai public health service, and their failure to find a large number of mild cases, is a fact, not my observation. The severity of H5N1 symptoms in an animal model is a fact, not my observation. I think we can all agree more data is needed before any firm conclusions can be drawn. My point is that the available evidence does not lead me to conclude that the “true” CFR must be much lower than the official one, especially in young people and children. I am willing to change my mind when strong data is presented, but, IMO, this has yet to be presented.
Note: I have provided refs for all of the above previously, in other threads, but will do so again here if you wish.
Monotreme, the “clusters” are those cases where H5N1 has been officially determined (ie strong enough symptoms to warrent testing). For the (3) Turkish deaths, the official cluster is either 3, or 4. This would suggest 75% to 100% mortality. The total number of those not showing strong enough symptoms will likely never be tested. How often do you go to the doctor with mild symptoms? Do you think those in a third world county would be more or less likely to go to a doctor with mild symptoms? The study I’ve referenced has included those cases, in a quantitative way. The WHO has never attempted to include milder cases in their CFR.
Monotreme,
I spent four days in a flu meeting in San Francisco in November with all of the flu heavies trying to work out the data. There isn’t any. You can’t postulate a CFR when you don’t have a clue about the number of cases. China is a black hole for data, Henry Niman’s unreliable Boxun reports not withstanding. Viet Nam has a surveillance system which is more porous than swiss cheese. The actual infection rate could be several orders of magnitude greater than what has been reported so far, and there are undoubtably mild or asymptomatic cases which are never reported. There are also, undoubtably, deaths which are simply uninvestigated and never tested. PCR tests aren’t all that common in SE Asia. That’s not cheap technology.
Making any hard conclusions about CFR in the face of the available data is a fools errand and I can’t think of one epidemiologist who’d be willing to take a stab at it.
Melanie:”Making any hard conclusions about CFR in the face of the available data is a fools errand and I can’t think of one epidemiologist who’d be willing to take a stab at it.”
Ah, but they are. Not mentioning any names.
I agree we have insufficient data to draw firm conclusions, but don’t agree that there is no data. We have cluster information which is informative (these are reported to be very intensely investigated by the WHO. A high CFR is observed). We have animal challenge experiments, which are informative (H5N1 is highly lethal to cats, a species which is immune to all other forms of influenza. Animal challenge experiments provide a controlled experiment). I agree that many, perhaps most, H5N1 cases have not been reported. But, the results of the cluster studies and the animal studies are not dependent on the completeness of surveillance. A fact which has not been appreciated sufficiently, IMO.
Cesear: “The WHO has never attempted to include milder cases in their CFR.” Um, I don’t think this is true. The WHO reports all the verified H5N1 cases they have, mild or not. You don’t think don’t have a rule against reporting mild cases, do you?
We don’t have information from the Turkish cases to draw any firm conclusions. In a couple of weeks, it will be clearer. However, statistically, how likely is that 4 children in one family would be infected with H5N1 and 3 die if H5N1 has a low CFR? This happens again and again in the many family clusters observed. BTW, I want to emphasize that I think the high CFR applies primarily to people under 25. Mild cases in older people would not surprise me.
See my last response to Melanie for my take on the sampling bias question.
I’m clearly out of my depth on the finer points of CFR, yet a point that Monotreme keeps making seems valid to me.
No one wants to contemplate that we could be dealing with a virus that spreads easily yet kills 50% or more of its victims. It sounds paranoid and extreme even to contemplate that, and scientists especially don’t want to sound like paranoid extremists, I’d imagine. So there is a tendency to say, well, 1918 was the worst thing on record, so it’s more reasonable to stick with that as a more likely upper limit for discussion purposes. Further, if you start exploring those real scary scenarios, people will throw their hands in the air and say, forget it, let’s just go get drunk while we can. And don’t even get started on the backlash from govts & business leaders re the harm you’d do the economy.
So there is a natural psychological bias re looking harder for and being more willing to accept reasons to downplay (vs. reasons to exaggerate)& to try to keep the discussion to a range of possibilities that we are more comfortable discussing and that are likelier to result in productive action.
That’s what I was alluding to in my previous wingnut comments.
“We don’t have information from the Turkish cases to draw any firm conclusions. In a couple of weeks, it will be clearer. However, statistically, how likely is that 4 children in one family would be infected with H5N1 and 3 die if H5N1 has a low CFR?”
First, I never said “low CFR”.
Secondly, there were entire families killed during the 1918 pandemic and others that were unaffected…this with a CFR of less than 2%.
The vast majority of the H5N1 family clusters, and there have been many, have a high CFR. This was not true in 1918, as near as I can tell.
And if you agree that the CFR is not likely to be low, then I don’t think we are disagreeing. Although one person’s low may be someone else’s high.
Monotreme, all I’ve done is post a scientific, quantitative study that shows H5N1 likely present as a mild form in many, many cases (making the CFR much lower than advertised). If you want to disregard the study and cling to the belief that the CFR is 50%, then so be it. The choice is yours.
Caesar, the problem with the study that you quoted is that they only surveyed retrospectively for ‘flu-like symptoms’ which for the current purpose is not sufficient. Such studies are notoriously prone to subject bias. There could be myriads of reasons why certain groups of people have more or less of certain symptoms, including possibly more anxiety or awareness of such problems. I don’t think anybody is dis-regarding such data. It’s just not sufficient for the purpose of declaring whether the CFR is lower than the current documented 50%.
Well said, anon_22.
I wonder to what extent the situation(s) in Turkey is being put under a microscope to see how widespread the virus is in humans beyond those who have been hospitalized and tested. If many more people are tested, and if fair numbers with antibodies found, would that survey clarify presently uncertain morbidity and mortality stats?
Also, would Turkish stats be more or less reliable than Chinese stats?
Dubina: The parents of the 3 children who died have not been tested for H5N1 infection, unfortunately. But this may be due to limited facilities which are likely overburdened. One hopes that the WHO will be helpful in this regard. In general, Turkey seems to be much more transparent than China so one hopes that the proper epidemiological studies, which have been done long ago, will finally be completed.
Turkey is trying very hard to get into the EU; or trying very hard not to give excuses for excluding them. Let’s hope that gives them a lot of incentive to co-operate.
Yes a suiter will not offend, will bend over backwards. Its a good thing they didn’t get married quite yet.
My understanding from what I have read is that there are many situations where someone is tested (by nasal and throat swabs) and the results show as a false negative because the virus has moved to the respiratory and is no longer in the nasal and throat. This has happened so often that there is now a name for it, I believe it is “goldilocks” Imagine how many undocumented cases there must be if they are even giving a name to the inaccurate testing going on…
closed for volume issues