Effect Measure

CBS News on swine flu testing: Fail!

A reader (h/t MVD) sent me this link to a “CBS News Exclusive,” Study Of State Results Finds H1N1 Not As Prevalent As Feared. As far as I can see the main aim was to raise CBS News’s profile and gain readership. That’s what news organizations do. We hope they do it by good journalism. I think this is an example where the reporters just didn’t have enough knowledge of what they were reporting and put the wrong spin on it.

The central claim is that CDC stopped testing for swine flu hastily and without advance notice to the states:

If you’ve been diagnosed “probable” or “presumed” 2009 H1N1 or “swine flu” in recent months, you may be surprised to know this: odds are you didn?t have H1N1 flu.

In fact, you probably didn?t have flu at all. That’s according to state-by-state test results obtained in a three-month-long CBS News investigation.

The ramifications of this finding are important. According to the Center for Disease Control, CDC, and Britain’s National Health Service, once you have H1N1 flu, you’re immune from future outbreaks of the same virus. Those who think they’ve had H1N1 flu — but haven’t — might mistakenly presume they’re immune. As a result, they might skip taking a vaccine that could help them, and expose themselves to others with H1N1 flu under the mistaken belief they won’t catch it. Parents might not keep sick children home from school, mistakenly believing they’ve already had H1N1 flu.

Why the uncertainty about who has and who hasn’t had H1N1 flu?

In late July, the CDC abruptly advised states to stop testing for H1N1 flu, and stopped counting individual cases. The rationale given for the CDC guidance to forego testing and tracking individual cases was: why waste resources testing for H1N1 flu when the government has already confirmed there’s an epidemic? (CBS News)

The implication here is that this was a precipitous and unilateral decision on the part of CDC that took the states by surprise. They support this by quoting unnamed public health “officials” who told CBS they disagreed with it and quoting from an email from the Council for State and Territorial Epidemiologists to its members (CSTE is a non-profit organization used by state epidemiologists to exchange information and advance their interests). The email was from CDC to CSTE members:

“Attached are the Q&As that will be posted on the CDC website tomorrow explaining why CDC is no longer reporting case counts for novel H1N1. CDC would have liked to have run these by you for input but unfortunately there was not enough time before these needed to be posted.”

It is clear that CDC is saying to the state epidemiologists they are sorry they did not run the Q&As about stopping testing by them , not the decision to stop testing. Contrary to this article, I heard throughout this period that the pressure on CDC to stop the testing was coming from the states, not the other way around. It’s no secret that state health departments are hard pressed to keep their heads above water financially and are short staffed all around. Expensive swine flu testing was something they couldn’t afford. The burden to do the testing was on the states, not CDC, but as long as CDC recommended it, states couldn’t easily stop on their own, especially if neighboring states were still testing. The reason for stopping was confusing (and CBS News shows themselves confused) and bound to be controversial. In effect, CDC decided to take the bullet for state health departments. And CBS News obligingly pulled the trigger.

The kind of testing asked for here — determining the subtype of an influenza isolate — is not a piece of cake. It requires specialized equipment and specialized expertise. Both the equipment and expertise exist in state health departments thanks to money and training CDC did in anticipation of a bird flu pandemic, but it wasn’t designed for high throughput. There was no way states could handle tens of thousands of specimens. If taxpayers don’t want to pay taxes, they shouldn’t complain when the infrastructure isn’t there, but in fact they might legitimately have complained about a laboratory capacity that goes unused except during an influenza pandemic. Since pandemics are unpredictable, that capacity would mostly lie fallow.

As an epidemiologist would I like to have more detailed information about the distribution and pattern of this disease? Of course. Influenza is an important public health problem. Yet influenza outbreaks happen seasonally and we never test and report to CDC individual cases with an eye to having a complete count. Why is that? First, a little digression on how diseases get their names.

A disease can get its name in two distinct ways, sometimes referred to as manifestationally and causally. Manifestationally named diseases are named by how they appear or manifest themselves. An example is cancer. We call diseases with malignant cells that form tumors and grow where they shouldn’t grow, cancers: in other words, by how they look or manifest themselves. Another example is a fracture, a break in the continuity of the bone. Another is congestive heart failure. What caused the disease isn’t in the picture. But the cause of a disease can be the basis for naming a disease, too. Some examples are asbestosis (scarring of the lungs from breathing asbestos), cholera (a disease caused by a specific micro-organism, Vibrio cholerae) or cyanide poisoning.

What about influenza? It’s a weird hybrid. On the one hand it’s part of a group of influenza-like illnesses (ILI), which is a manifestational name. When a doctor tells you that you “have the flu” these days, most often he or she is using the manifestational name. But there is also a causal name for influenza, referring specifically to the ILI caused by infection with influenza virus. That’s the name that can be placed on your disease after some very specific and sophisticated testing to see what is causing your ILI. There are a lot of viruses that can cause ILI (we’ve talked about some of them here a couple of times, for example, here and here), and undoubtedly they are responsible for some ILIs that are now being diagnosed as swine flu. We know, however, that during flu outbreaks, the proportion of ILI that is influenza goes way up, so that the kind of clinical (manifestation only) diagnoses are much more often correct (there is a theorem in probability theory that explains why that is). So it makes a difference that CBS News was looking at data from May, June and July rather than now. At that time there were other viruses around besides swine flu, and the tests indicate that. Why not do better testing?

There are “rapid flu tests” that are done in doctor’s office that can show if you have influenza, but they miss somewhere around half the cases of real flu, so a negative test isn’t helpful, nor does the test tell what kind of flu it is (seasonal or swine flu). For that you need the expensive lab test. And even if we could do all those lab tests, it would entail many days’ wait, during which you aren’t being treated with antivirals, or if you are, then getting the test didn’t make a difference. Most people who have the flu probably never see anyone about it. They suffer and get over it or if it is not so bad they keep going about their business. So they would never get counted anywhere. And many people have ILIs that aren’t flu. And they aren’t counted either as other respiratory diseases.

CBS’s own investigation found that in all 50 states, prior to stopping testing in July, lab-confirmed cases showed that most specimens were not influenza. As I noted, since other viruses cause ILI during non-flu season this isn’t too surprising. What was surprising was the amount of influenza there was at a time when we expect to see very little. Now that flu season is here, the chances that an ILI is truly influenza (causally defined) is much greater. And frankly, there is no possibility of testing all ILI cases for swine flu. There will be millions of cases of ILI and they can’t all be tested and most won’t even be seen by anyone. Moreover, during the period of the CBS “study” many people were having specimens taken that would not ordinarily have seen a doctor. The data they looked at were from all 50 states (and we don’t know what data it was or what they counted or whether they even calculated things correctly) and were mostly cases that were not epidemiologically linked.

This last is important. CBS News cites an outbreak of 250 cases at Georgetown University as an example where no testing was done so it isn’t certain this was a flu outbreak. Admittedly, it could have been adenovirus or respiratory syncytial virus or a bunch of other things. But when something like this happens in the setting of a pandemic, the odds are that it was influenza. More importantly, data on the number of positive specimens for all the people of Georgia over a several month period three months prior is not the same as 250 cases among students who were in contact with each other and that occurred over a few weeks. And think about the alternative? Nothing prevented Georgetown University from testing all those cases. Nothing except that it would have been infeasible and the resources weren’t available, that’s all. And it wouldn’t have changed how they were treated, except perhaps to delay treatment.

Still, as an epidemiologist, we need to know what virus subtypes are out there and roughly where they are. CDC has a multipart surveillance system we’ve written about quite a lot here. It supplies a great deal of useful information for epidemiological purposes and special studies fill in some of the gaps. So it isn’t true that lack of routine testing makes it impossible to know where flu is occurring, what is happening over time and what kind of influenza it is. We have a jury-rigged but fairly efficient influenza surveillance system that has been operating for years and continues to operate.

But compared to countries that have universal health care with efficient electronic medical records we have a much harder time. Our health care system, which is mainly private, isn’t set up to make it easy to get the kind of information people think we should have.

That’s not CDC’s fault. And I haven’t seen that CBS News has done such a bang up job of reporting on health insurance reform. Unless being a stenographer for talking points from the insurance and health care industries counts.


  1. #1 PalMD
    October 22, 2009

    well done, again, EM. Now I’m back to the clinic to see more ILI.

  2. #2 Karen
    October 22, 2009

    I got this in my inbox this morning from a “caring” friend.


    I’m getting far too many chances to discuss this sort of thing with my kids. I hadn’t planned on this particular aspect of biology and critical thinking in this year’s curriculum planning.

    I’ve found that trying to discuss the details of what these people are passing on is very much unwelcome. They “know” what they know and don’t want to think about it.

  3. #3 pd
    October 22, 2009

    Thank you for addressing this!

  4. #4 Phila
    October 22, 2009

    Our health care system, which is mainly private, isn’t set up to make it easy to get the kind of information people think we should have.

    It’s not set up to deliver timely vaccinations, either. We just went to the H1N1 clinic, which was supposed to be open from 10 – 2. About 1,500 people showed up; only the first 500 got shots. Everyone else was sent home, after having waited in the cold for a couple of hours.

    Apparently, they were “surprised by the turnout.”

  5. #5 daedalus2u
    October 22, 2009

    Would there be any value in testing these samples “off-line”? That is freeze them, put them in the queue, and test them eventually? A special high-throughput system with a longer turn-around might substantially reduce the cost per test.

  6. #6 revere
    October 22, 2009

    daedalus2u: It still wouldn’t pick up all the people who had flu and never had a specimen taken or didn’t see a doctor or clinic. If the purpose is to help the patient, then there would be no point. If the purpose was epidemiology, then a well designed study with rasonalbe spatial and temporal sampling would probably be cheaper and more informative.

  7. #7 downeast
    October 22, 2009

    Revere: Thanks again. Got my teens scheduled for pneumonia vaccine; pediatrician agreed it is a sensible move. Wouldn’t have considered it if had not read these threads.

    Being the tail end of the country, we little influenza vaccine to speak of, seasonal (zippo) or H1N1 (just making its way into the state), so common sense dictates other precautions. School nurses, local doctors write columns, we print them — over and over advice about handwashing, rest, cough into sleeve. Lately, some others have been added: gargle twice a day with warm salt water, or Listerine; drink hot liquids, such as tea and chicken soup, to ward off or diminish virus proliferation in throat. Question for you: is that true? Can viruses be muted by these simple acts? Had thought those were palliative, but looking through archives of 1918, 1919 newspapers, several advertisements delivering similar message. Precursors of Listerine.

    Thanks for clarifying.

  8. #8 John
    October 23, 2009

    Sorry to digress from the main point of your post but this caught my eye:
    “There are “rapid flu tests” that are done in doctor’s office that can show if you have influenza, but they miss somewhere around half the cases of real flu, so a negative test isn’t helpful, nor does the test tell what kind of flu it is (seasonal or swine flu). For that you need the expensive lab test.”
    I don’t know about the US but here in Japan, where I temporarily live, the doctor sticks a swab far up the nose of my 2-year-old to test for influenza (the result appears in 10-15 minutes). If the test is negative, sometimes the doctor asks us to return the next day for the same nose test, on the chance the first test was done too early (on the first day of sniffles and fever).
    Any comment on the efficacy of this testing process for very young kids, especially when a 2nd test is done?
    Are you saying that even a 2nd (and 3rd, 4th, etc.?) influenza test can produce a false negative?
    It appears you’re also saying that now that we’re into the normal flu season the doctor’s quick 10-min. test will be unable to reveal whether it’s seasonal or swine flu, yes?

  9. #9 Lisa the GP
    October 23, 2009

    As things are now, if it was easy to make medical record systems talk to each other, you can be sure that insurance companies would exploit that to discriminate against policy holders even more than they already do. (off-topic snark)

  10. #10 revere
    October 23, 2009

    downeast: The recommended “remedies” (gargling, listerine, etc.) as far as I know have no effect other than symptomatic relief, but that’s not a bad thing. It’s not clear that things like nasal lavage with salt water do anything at all except wash more virus down your oropharynx. But I doubt any of these remedies will hurt you either, and if they make you feel better, fine.

    John: The rapid flu tests miss about half of actual infections. The reasons could be the test is not well-matched to the virus (there are several vendors with different variations of the test) or that the person wasn’t shedding virus enough on the day of the test or that the swab didn’t hit the right part where the virus was replicating. Doing a repeat will lessen the cances of the latter two reasons, but not the first, so, I guess like a lot of things, the answer to your question is, “It depends.”

  11. #11 Tim McCormack
    October 24, 2009

    Thanks for this rebuttal. It’s great to get a different view on this issue.

    I think I’ll still get the combined vaccine (assuming that exists), just to help herd immunity. (I’ve only twice gotten symptomatic flu, but I may be more of a carrier…)

  12. #12 karen
    October 24, 2009

    While you are on the subject of testing can I ask a sciencey question and have an as idiot proof answer as possible ?
    Here in the UK the health protection agency do random testing via nasal swab of everyone who is prescribed tamiflu,how accurate is that sort of test likely to be bearing in mind the people I know who were tested were five days into flu when prescribed tamiflu and the nasal swabs didn’t turn up until a week after that,it seemed obvious to me a lay person that the chances of a positive swab for virus was extremely unlikely.
    It is just that I have a sneaking feeling that our numbers are way lower than than could be expected.

  13. #13 revere
    October 24, 2009

    karen: It’s hard to say from your description. There may be certain triggers for who is eligible to be randomly tested, and I don’t know what kind of test they are doing. But the test numbers don’t usually do a good job of estimating the true number of infected, but can be useful for following trends over time.

  14. #14 karen
    October 24, 2009

    Thanks for that.These people were really very sick for ten days or so,sadly the only information they have is a letter which states”The laboratory tests were negative.We were not able to find the ‘swine flu’ or the usual winter influenza virus in your samples”
    I recommended taking the vaccine if and when offered as a precaution,much as I have with a couple of my grown kids who got pretty sick during H1N1s first big run through the country.Without a positive test result goodness knows what any of us are sneezing round our neighbours.

  15. #15 Josh Fulton
    October 25, 2009

    You completely ignore the fact that the overwhelming majority of cases when tested did not show swine flu, and that the states, despite the cost to them, continued with the testing, yet the CDC simply chose not to release those numbers. So, it’s not a matter of cost at that point. It’s a matter of interest.

    The swine flu has been overhyped. This report uses the same exact info the CDC would have used, if they had chosen to release it, to show just how much.

  16. #16 revere
    October 25, 2009

    Josh: You ignore the fact that these data are from the first three months, outside of flu season. The current figures are running around 30%, which is very high . I don’t know what your basis for saying this problem has been exaggerated or by whom, but if you were in the trenches you wouldn’t say it. You sound as if you don’t know how serious flu can be or why the changed epidemiology of this is the key factor. Feel free to look at other posts, going back months (or years, for that matter). I don’t know when you got into this business but we’ve been blogging on it for 5 years and in public health and medicine for over 40 years. Our assessment differs from yours.

  17. #17 purlogik
    October 30, 2009

    Under the Public Readiness and Emergency Preparedness Act of 2006, manufacturers of swine flu vaccines are granted TOTAL LEGAL IMMUNITY. Do you think THAT is important enough to mention? Hmmm, I wonder why drug companies would need to lobby for legislation that granted legal immunity for their vaccines. With that knowledge, only a complete fool or a government/industry shill would defend these vaccines. Which one are you?

  18. #18 revere
    October 30, 2009

    purlogik: We argued strongly against that provision, which I believe was put in by the Senator who represents the drug industry, Joe Lieberman. But they wanted it for the same reasons doctors want immunity from malpractice and corporations want immunity from personal injury suits. They don’t want to be accountable for their negligence. Meanwhile don’t see a doctor since he/she wants immunity from lawsuits, too. And you know what that means. And don’t go to the hospital because they want it, too.If it bothers you for vaccines, then don’t get it. Your choice. I hope you don’t infect someone else, but as for you, you take your chances. There’s a shortage and I am hoping we will donate yours to people who need it and want it.

  19. #19 FarSide
    November 2, 2009

    revere –

    Thank you for posting something that contains facts and reasoned explanation in your rebuttal, instead of the typical ‘these people are idiots’ type of arrogance I’ve been seeing while looking into this (including other authors on scienceblogs). It’s truly refreshing.

  20. #20 trm
    November 3, 2009

    Not sure you will be able to answer this question, but if my son had a non-confirmed case of H1N1 a few days ago, should he get the vaccine just in case?

  21. #21 revere
    November 3, 2009

    trm: CDC says, yes, vaccinate. The reasoning is that there are a lot of viruses that aren’t flu that cause flu like illnesses, so you are better off making sure your child has immunity.

  22. #22 Webs
    November 13, 2009

    Did my last comment get lost?

  23. #23 Grim
    November 22, 2009

    “Admittedly, it could have been adenovirus or respiratory syncytial virus or a bunch of other things. But when something like this happens in the setting of a pandemic, the odds are that it was influenza.”

    So, if CBS info was correct, and there weren’t very many cases of actual influenza, the government hastily declaring a ‘pandemic’ would be incorrect, would it not? It seems to me that the pandemic wasn’t really occuring, but was more of a safeguard against the possibility of one occurring, or just trying to scare people (which I don’t know if they’d just do). If the government was given the same information that CBS was able to acquire, would they have been able to determine that 1-2% of normal tests done were actually swine flu and in good conscience declared an epidemic?

    if you want, reply to my e-mail address.

  24. #24 revere
    November 22, 2009

    Grim: No. It was a pandemic by every measure: a novel strain of influenza passing from person to person easily on several continents.

  25. #25 Todd
    November 25, 2009

    I enjoyed reading your perspective on this issue. My concern lies with the acceptance of the CDC to quote estimates of H1N1 infections and deaths without any real confirmation. With the ability to test, but choosing not to, this is unacceptable in my opinion. Cost as a factor is real, but the cost of vaccinating 40-100 million people pales in comparison. Why wouldn’t we want to KNOW with at least some certainty we are seeing deaths related to H1N1, and not some other cause of ILI.

  26. #26 revere
    November 25, 2009

    Todd: It’s not a cost issue. It’s a time/equipment/staffing/value issue. None of the four are sufficient to test tens of millions of people.

New comments have been disabled.