Slashdot Mirror
AIDS Vaccine Is Partially Successful
ifchairscouldtalk writes "A Phase III 'RV 144' study in Thailand succeeded in reducing HIV infection rate in trial with 31.2% effectiveness. The study was conducted by the Thailand Ministry of Public Health and used strains of HIV common in Thailand. It is not clear whether the vaccine, which combines AIDSVAX with Aventis Pasteur ALVAC-HIV canarypox vector, known as 'vCP1521,' would work against other strains in the United States, Africa or elsewhere. Strangely, the vaccine had no effect on levels of HIV in the blood of those who did become infected, providing 'one of the most important and intriguing findings' of the trial, according to Dr Fauci, director of the National Institute of Allergy and Infectious Diseases, which is one of the trial's sponsors."
I'm not normally a stickler for these, but AIDS is a syndrome, HIV is the virus that causes it. The vaccine can prevent you from acquiring HIV and thence from developing AIDS. It's not a cure, it's a preventative measure.
I was promised flying cars...Why are there no flying cars?
Funny, but you forgot about blood transfusions. While it's extremely rare for contaminated blood to be used in the U.S. and many other western nations, it's a very real possibility for a lot of the world.
512 MB RAM, 20 GB disk, 200 GB transfer, five datacenters. $19.95/month.
The sample sizes were not 74 and 51. The sample size of people vaccinated was "more than 16,000 volunteers". 74 and 51 were just the number of people infected, which is still statistically significant. [to what confidence level, I do not know].
someone do some analysis on the statistics and tell us all something and get +5
Sure. It's Poisson statistics, so the standard deviation is the square root of the count.
placebo: 74 plus or minus 8.6
vaccine: 51 plus or minus 7.1
The statistical significance of the difference (23) is equal to the standard deviation of the sum (not the difference!) of the counts, so:
difference between placebo and vaccine:
23 (=31%) plus or minus 11
= (2.06 standard deviations)
Assuming they set their criteria for statistical significance at two standard deviations, then they are significant.
http://www.geoffreylandis.com
No, they just went to a population where HIV is already relatively common and a large number of people don't usually take adequate precautions against it (i.e. use condoms) and then studied the effects of the vaccine on that population's total infection rate over time. It's not the greatest way to test this (since you have no way to tell if it's just down to random variations in the two population's levels of exposure) but doing it properly (i.e. deliverately exposing people) is pretty unethical to say the least.
Of course, you can also test the vaccine on animal models which are deliberately exposed to HIV so we know there's a good chance it will be effective if the population study then shows these kinds of results to corroborate it.
Ultimately one of the purposes of drug trials is also to look at side-effects. Assuming the side-effects of this vaccine weren't too bad, with that kind of effectiveness rate it would seem this stands a reasonable chance of widespread deployment, in which case it'll be possible to gather more data.
At least the article has some numbers so we can to the (approximated) maths ourselves. The numbers of infected for medicine/placebo are 51 and 74, respectively. The error of the difference is then sqrt(54+71) = 11 (poissonian statistics). The difference itself is 74-51 = 23, so we can conclude that 23 +/- 11 persons were saved from infection. That means that we're just two standard deviations (23/11) away from the null result. This will happen by coincidence 5% of the time. So if they'd done the study 20 times, you'd expect this outcome once. Now this study has only been done once, but I wouldn't be surprised to hear that 20 different AIDS medications have been tested over time, and so it's quite likely to see this outcome once. Conclusion: This warrants further study, but they really haven't proven much yet. In fact, if they were physicists and not physicians they would have proven nothing. Especially since they, by their own admission, cannot explain the result.
Er... they *all* underwent the same sex and infection-control education courses, according to the BBC article. Medical researchers don't throw people to the wolves just for the sake of science... at least, not any reputable ones whose research they expect to be followed up.
And if they had never run the test, the results would have been the same. Please, don't push blame onto others, we have enough of that into modern society. There is plenty of education forced down your throat at every turn about HIV/AIDS, if you don't know about it, it is your fault.
-- Lattyware (www.lattyware.co.uk)
Didn't the article say that one group got a vaccine, and the other got a placebo?
"Col. Jerome H. Kim, a physician who is manager of the armyâ(TM)s H.I.V. vaccine program, said half the 16,402 volunteers were given six doses of two vaccines in 2006 and half were given placebos."
Oh yea, that's what it said.
I don't see anything wrong with the basic kind of study. As I said, they may have fucked it up somehow, such as fucking up the selection of the participants and grouping them.
And why would they want to control against additional groups? They're measuring one thing. How effective is the vaccine. Your proposal to control against other groups are actually separate studies. They can and should be run independently at first. I can totally understand them not wanting to add complexity to a study that already has more than 16,000 participants.
So, I still don't see any valid objection as to why this kind of study won't work or is flawed somehow. In fact, this basic type of study is done all the time.
Fascism trolls keeping me up every night. When I starts a preachin', he HITS ME WITH HIS REICH!
The total working group for this test was around 16,000 people. Only 125 actually became infected with HIV during those 3 years. The infected portion shows about 1/3 more in the placebo group. So yes, the sample is statistically significant, and someone wasted a mod point.
Who is John Cabal?
This page and this page indicate that the study was double-blind. If it was, then I do not see how your worry is reasonable. If both groups were unaware of whether they received the treatment or not, then I do not see how one group that happened to be the control group would reliably act differently than the experimental group. Am I missing something? Or are you claiming that once people believe they have the vaccine, that they will have more unprotected sex and thus increase their risk of contracting HIV?
Every post I make begins with the assumption P=~P.
No, they just went to a population where HIV is already relatively common and a large number of people don't usually take adequate precautions against it (i.e. use condoms) and then studied the effects of the vaccine on that population's total infection rate over time.
How do you know? I RTFA, and I don't see it mention their procedures anywhere.
It's been reported fairly widely elsewhere. See, e.g., this article from the BBC:
Unfortunately, as the article notes, the sample groups may themselves be problematic. A previous study of another vaccine that was found to *increase* infection rates may not have been dangerous; they just neglected to control for circumcision and intravenous drug use. Given this study dates back to before the end of the previous study, their samples may be skewed the opposite direction. Circumcision has been found to reduce the risk of infection by 40-70%. IV drug use is insanely dangerous (I don't know the exact multiplier, but it is rather high). If the test group had a couple hundred extra circumcised individuals, or a dozen or so IV drug users, it could easily skew the results.
Of course, this is still overlooking another problem with the vaccine. It's not one injection, or even one plus a booster or two. It's a two vaccine regimen, with six injections of each component, for twelve total shots (they may eventually be combined, but that all depends on whether the components react with each other outside the body). And the duration of the protective effect is unknown, and likely short (since the vaccine doesn't seem to trigger the production of antibodies). Even if it was incredibly cheap, it's hard to get people to follow up for a second MMR shot, or keep up to date on their tetanus, both of which protect against diseases which are easier to catch without engaging in risky behavior. Can you imagine asking people to pay a few hundred dollars (a guess based on the cost of Gardasil), and visit the doctor half a dozen times to get such a relatively small benefit (reducing risk by about a third, with only two years of testing)?
Even if we assume the samples are good, this is only a first step, and a very short one at that.
In response to the PP: I suspect the confidence level is 95%. Most published studies require that level of precision, and no one likes to hamstring themselves by demanding greater confidence; after all, they spent a lot of money and rejecting the drug would waste it. Of course, if you've ever played D&D, you know how often you get fumbles or critical successes. 95% means the odds of it being insignificant could be as high as the odds of fumbling a roll.
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
Uh no, heterosexual anal sex carries the EXACT SAME RISK with a known infected partner. But hey, way to be a bigot. Anal sex in general is far more risky as the vagina is an acidic environment that is hostile to the virus, plus microtears can occur int he rectum walls during anal sex that creates a better blood pathway.
Specifically, you could say you were unlucky at the 5% chance level -- that's the (approximate) odds of getting results more extreme than this, given the number of people in each group that actually got infected, purely by dumb luck, if the vaccine did exactly nothing. (74 vs 51, out of a total of 16402, broken into two groups; that's just using a poisson approximation, since I don't have the precise group sizes, which gives 2.06 standard deviations, or significant at the 5% level.)