One of the greatest concepts I have run across since I finished school is the Number Needed to Treat (abbreviated NNT). NNT was never taught back when I went to medical school (we had barely given up The Four Humors!). Instead, we were taught “the p-value.” Does anyone else remember the p-value? The p-value of a study, it turns out, is a relatively poor measure of study validity, partly because it implies an “all-or-nothing” kind of understanding of studies: either the study is “valid” (meaning a p-value of >95%) or it is not. Either the treatment being studied works or it does not. And if a treatment works, it must work for all people.
Of course, in real life, this is not the case. No drug is universally good or bad. All drugs help some people, harm some people (with adverse side effects) and make no difference one way or another in some people. These numbers can be derived from any study’s data. There is even a fabulous website devoted to this where you can look up the NNT and its corollary, the Number Needed to Harm (NNH) for all sorts of drugs and treatments (found here).
Since it is influenza season and time for us to get our flu shots (I got mine yesterday), I thought it would be a great time to see how beneficial the flu vaccine is. What is the NNT for the flu vaccine? And while we are at it, why don’t we also look at the data on oseltamivir (Tamiflu) while we’re at it?
First, some background information. Every year, in February, the World Health Organization has to pick which flu strains are going to go into that year’s vaccine. This involves more than a little educated guessing, since the WHO is predicting what will happen the next flu season, which is 8 to12 months in the future. Most of the time, the WHO predicts well and the 3 or 4 strains of influenza in that year’s vaccine match the influenza strains going around. We’ll call those “well-matched” years. Some years, however, the WHO misses its guess and the vaccine does not match the prevailing flu strains. We’ll call those years “not well-matched.
How Effective is the Flu Vaccine in Well-Matched Years? Answer: Very!
Statistically, in well-matched years in the US, the average likelihood of getting the flu is around 4%, or one in twenty people. If you get your flu shot in well-matched years, you reduce your risk of getting the flu to 1%, or one in a hundred. In other words, it is still possible to get the flu even though you were immunized, but you have reduced your risk by 75%. However, remember that 96 out of 100 immunized people would never have come down with the flu in the first place. The flu vaccine did not help these people. And, of course, the vaccine also did not help the one unlucky person who got the flu shot but still caught influenza. A total of three people out of 100 did not come down with the flu because they were vaccinated.
So, in well-matched years, if we divide 100 people who got the flu shot by the three people who benefitted, we get a Number Needed to Treat of 33. In well-matched years, for every 33 people who get immunized, one will benefit by not getting a bout of influenza she otherwise would have suffered.
A NNT of 33 is, by the way, a fantastically good therapy. This is far better than the NNT to benefit of many, many other common therapies, like, for example, anti-depressants, anti-hypertensive drugs and prostate cancer screening.
How Effective is the Flu Vaccine in Not Well-Matched Years? Answer: Still Pretty Good!
Even in not well-matched years, the flu vaccine is still effective, albeit not quite as well. In not well-matched years, the overall risk of getting the flu drops to 2 people out of 100. The flu vaccine cuts that number in half, for a 50% risk reduction. If you do the math, one person out of 100 benefits from the flu shot, making the NNT in not well-matched years 100.
I myself caught the flu a couple of years ago (I wrote about the miserable experience here), and I can tell you that I do not ever want to experience the flu again. Influenza means several days home, in bed and miserable. (In fact, the NNT with the flu vaccine to prevent one doctor visit is 38!) If the influenza vaccine drops my risk 75% most years and 50% even if the WHO misses their guess some years, I’m going to do it.
By the way, the influenza vaccine is even more effective in kids! The NNT in children is only 8. For every eight kids vaccinated, you will save one kid from influenza.
Remember also that people who get the flu are infectious for around 24-36 hours before they feel symptoms. Almost everybody who gets the flu is going to inevitably expose others. Getting a flu shot is not just for your own health, but for the health of your family, your co-workers and your patients, as well.
How Effective is Oseltamilvir (Tamiflu)? Answer: Not So Much.
As a justification for not getting immunized, I’ve had heard some say, “Well, if I get the flu, I’ll just take Tamiflu. That’ll beat it.” So is that true? If you get influenza, this drug Tamiflu will cure you? Just how effective is Tamiflu?
It turns out that there is an interesting story behind reports of Tamiflu efficacy. Perhaps the single most influential paper published about Tamiflu–the one that lauded the efficacy of Tamiflu; the one that helped spur the US government to stockpile huge amounts of Tamiflu in case of an influenza pandemic, that one–was a meta-analysis of 10 studies done by employees of Roche, the maker of Tamiflu. Besides the obvious bias of Roche employees themselves objectively evaluating the evidence about the effectiveness of their own drug, 8 of the 10 studies they cited had never been published and Roche refused to release the data from any of those eight trials! ( Here is a great TED talk about this controversy).
Finally, after many years, under pressure, Roche did release the data from the missing studies. The independent Cochrane Review analyzed the previously unpublished and withheld data (found here, if you want to peruse more than 500 pages of data!) and came to some interesting conclusions.
1. If you begin taking Tamiflu within 48 hours of the onset of influenza symptoms, you will reduce mayalgias and aches (flu-like symptoms) by a little less than a day
2. On the other hand, those taking Tamiflu had significantly more vomiting.
3. So it seems to be a straight swap: a little less achiness for a little more vomiting. Oh, and the not insignificant cost of the drug.
4. Tamiflu does not reduce serious complications of influenza, like death, pneumonia, hospitalizations, etc.
5. Tamiflu does not reduce the likelihood of spreading influenza to others.
1. The influenza vaccine reduces the risk of contracting the flu by between 50% and 75%. This translates into a number needed to treat of 33 to 100.
2. Tamiflu, if given in the first 48 hours of influenza, will reduce flu achiness by a little less than a day, on average.
3. On the other hand, it costs a lot and increases vomiting.
4. Tamilfu does not prevent the spread of influenza, nor does it reduce influenza complications.
Are you yourself getting the influenza vaccine this year? Please comment!
Note: Most of the data in today’s post came from a new and truly excellent Primary Care CME website: Primary Care RAP. Primary Care RAP requires a subscription and so cannot be added to the database of free educational websites on the CFOAM page of JailMedicine, but is well worth the price of admission!