Watching TV in America is nothing but a continuous flow of commercials about symptoms and diseases that they really want to convince you that you probably have. We know advertising can make you want all kinds of things, so why not symptoms and diseases?
Funny fact, any commercial for a drug, here in Romania, is ending with a very fast auction-style warning that last less than 5 seconds. In US you have like 5 minutes of country music and elder people having a picknic while all the possible dangers in the world are softly planted in your mind. :)
Well that would explain why placebo is also rising against antidepressants, unless more American are diagnosed with being depressed than they truly are.
Why are placebos becoming increasingly effective only in the US ?
A couple of hypotheses were mentioned in the article. One is that only in USA and New Zealand is direct-to-consumer drug advertising allowed. Another, which seems to have some supporting evidence, is that size and length of the drug trial may boost the placebo effect. The latter point is presumably about the production value of the trial, mandated by increasingly stringent FDA regulations. The apparent cost sells the apparent effectiveness.
I wonder whether these results might be linked to USians sustained high level of religiosity, which is also unique in the developed world.
A friend of mine said that placebo is very effective in Romania (our country) too. I may mistake it, but she said that it can reach something like 70-80%. Maybe it's a developing countries problem. :)
I'm not sure if I'm missing something either about statistics or about how the trials work, so maybe someone can enlighten me:
> Based on patients’ ratings of their pain, the effect of trialled drugs in relieving symptoms stayed the same over the 23-year period — but placebo responses rose. In 1996, patients in clinical trials reported that drugs relieved their pain by 27% more than did a placebo. But by 2013, that gap had slipped to just 9%.
Wouldn't an easier way to look at it be to completely ignore the effects of the actual drugs in those trials, and just look at the data of how many people took placebos and to what extent those people reported changes?
I have no insight, but looking at the text you highlighted, I expect they must have done exactly the analysis you suggest.
It is likely that the researchers and the audience want to understand how to get more effective drugs onto the market more than why placebos per se are becoming relatively more effective. This may explain the unexpected reporting emphasis.
When you're doing a regular drug trial, you control using a placebo to account for the effect of just taking something.
In this reverse analysis you need the opposite: if the placebo effect is changing, that could either be because the placebo itself is more effective, or because people are reporting pain differently, or a number of other confounding factors. Analysing the people who took real painkillers too gives you a baseline to compare against.
It's impossible to test "absolute" effects of drugs, so they are always tested in relation to each-other, this makes analysis easy because the only variable that's different is between whether participants are taking the actual drug, or a placebo. What you're suggesting would require testing a placebo in relation to not taking anything, which would introduce a lot more variables, and would make it impossible to do double-blind testing.
OK so the answer is I don't understand drug testing :)
My understanding was: you need the placebo to compare to to counter out the placebo effect (i.e. if you only test "real drug vs. nothing", the real drug might show an improvement purely due to placebo and have no actual effect itself). But introducing more variables... surely there are lots of other variables there regardless? If you ran a test of "placebo vs. nothing" what variables come in that wouldn't be there for "placebo vs. drug"?
First off, the reason double-blind testing is so effective is that it reduces subconscious biases in behaviour. If the person administering drugs knows its a placebo, he/she will consciously, or subconsciously, behave differently. All sorts of unwanted effects come into play, most importantly Observer-expectancy (For a good example, https://en.wikipedia.org/wiki/Clever_Hans)
Double-blind testing allows one to test the effects of administering anything (the placebo effect), versus the effects of the actual drugs. By administering an inert substance, the baseline placebo effect can be tested, as no actual medicative effects are expected from this inert substance.
When testing placebo vs nothing, in the nothing case a lot changes. The participants will no longer be administered anything, the experimenters will be aware of not administering anything, etc. etc. Because so many variables change in such a set-up, you can no longer actually tell which of these variables changing caused the effect and which of them didn't. Thus making any results and conclusion derived from the suspect.
I'm not entirely sure about that. Some trial designs do indeed have an arm of "no intervention" or "usual care" or "waiting". Sometimes these groups may be given the active therapy after the trial ends, since withholding a potentially effective and efficacious therapy may be considered unethical. There are many trial designs, not all of them are lab-based - - "pragmatic" trials in particular don't tend to follow typical masked, randomised, placebo-controlled designs.
> In 1996, patients in clinical trials reported that drugs relieved their pain by 27% more than did a placebo. But by 2013, that gap had slipped to just 9%.
Wait, this can be read as "Drugs are less efficient" just as well as "Placebo are more efficients".
They do say "Based on patients’ ratings of their pain, the effect of trialled drugs in relieving symptoms stayed the same over the 23-year period", too
Yeah, but when they compare subjective measurements across studies like that, it's really an apples and oranges comparison. The entire reason we use placebos is because there is no objective measurement for things like pain or sadness or "how well are you feeling in general?".
If the gap between placebos and pain medications is narrowing, my first assumption is that the experimental designs have tightened over time, so that the blinding is better, and subjects and doctors are less able to distinguish between treatments. It's not that pain medication is "less" effective, it's that it was always only (say) 9% better than placebo, but in crappy earlier studies where it was less well blinded, some subjects (or researchers working with the subjects) could tell they were taking the placebo, and reported feeling worse, and some subjects could tell they were taking the treatment, and reported feeling better.
If you tell me the gap has narrowed while the subjective reports of the treatments' effectiveness has stayed the same, I still believe my first assumption above, but now I'm a little bit suspicious of actual researcher bias. There are a vast number of ways researchers could "tune" for a target effect size, from classic ways like choosing the stopping point of the experiment to more subtle things like interpreting patient responses up or down based on whether a running analysis is coming in hot or cold.
(The later could be less overtly malicious than you might think -- imagine that you're continually analyzing the results of your experiment, without breaking blinding. You "know", from past experiments, that treatment X is 70% effective. If you are seeing it as 80% effective in your running analysis, you might assume your interactions with your subjects are subtly biased to encourage over-reporting of the effect and consciously or unconsciously alter them to encourage under-reporting, and vice versa if you only see a 60% effect. If your blinding is still good, your gap in effectiveness will still narrow, but the "absolute effectiveness" of the treatment will still narrow.)
There are several problems with double-blind testing that have become apparent in the last few years, and that call for a new standard testing procedure.
Double-blind rests, among others, on the assumption that a) placebo effects are independent of and additive with actual drug-effects, b) placebo effects are nulled when participants know they're being given a placebo.
Although I'm currently without sources (hopefully I can rectify this soon) both a and b are actually false.
Certain drugs have been found to interact with placebo, which decreases or enhances their effects, thus again conflating the medicative effects of the drugs being administered and the effects of actually taking a pill (or rather going through the process of administering medication).
Second, placebo has become such a household term, that the original meaning as a "fake" effect, seems to have been lost. Logically this makes sense, because its the end result that matters (being cured), not the way in which the drug is effective (placebo or not). The average person being told they're receiving a placebo can actually increase the placebo effect, rather than nulling it.
A second variable should be introduced in double-blind studies, to account for both factors I mentioned above. Variable 1 should be whether or not they're receiving the actual drug (1a) or a placebo drug (1b) (as double-blind currently tests for) and variable 2 should be whether participants are told that they're receiving the actual drug (2a), or a placebo drug (2b).
Testing the effects of combination 1a/2b vs 1b/2a should test whether the drug actually interacts with a placebo effect. Providing a measure of validity for the 1a/2a vs 1b/2a test.
I'm not a medical expert, so I'm a bit inconsistent in terminology, but have ran my fair share of (cognitive) psychological experiments (and developed methods for them), which rely on similar logic as to that of double-blind testing.
It's a problem of testing the efficacy of going through treatment (with any substance, inert or active) vs testing the efficacy of a certain active drug. In double-blind experiments, where placebo effects are attempted to be disentangled from actual drug effects, such a baseline would introduce too many variables to be taken in account. There are psychological factors related to the participants response ("Hey, I joined this clinical trial, but I'm not being given anything, what's up with that?") that make it difficult to conclude anything from the results. The participant might even be adversely afflicted by this seeming "lack of care" from the experimenters side. Next to that, experiments might experience observer-expectancy effects, where the expectations of experimenters cause the participants to behave in certain ways.
Also, if a positive result was found in the "not given any drug" vs "given the active drug" trial, one would still not be sure if this was due to the participant going through the treatment, or due to the treatment + the effects of the drug.
Whole lot of speculation in this article. I would guess more people know about the placebo effect in the U.S., than they did 20 years ago? Twenty years ago, we didn't question the efficacy of these drugs? Everyone I know now is very leery of any claims by drug companies, and their Wonder Drug.
"For companies trying to develop treatments, one remedy might be to compare new drugs against their best competitors instead of against placebo — or to go back to conducting smaller, shorter trials."
Lets Not do this. Lets continue to look for drugs that actually work on pain. The last thing we need is another drug that works slightly better than an ineffective pain drug that squeaked past the FDA? In the mean time there's a whole bunch of patients who are in extreme pain, and doctors are afraid to prescribe opioids. Doctors know opioid type medications work, but are afraid to prescribe them for a lot of reason, but mainly they don't want the
FDA breathing down their backs.
My best friend died a horrid natural death a few years ago. He was in bad shape, but the doctors couldn't find anything really wrong with him besides being very old, and had a very bad case COPD. He was in constant daily pain.
I bought him to "the best pain clinic in San Francisco". They wouldn't alleviate his pain? They offed to cut some nerve, but wouldn't medicate him. (To that clinic, if you are going to stereotype patients, and offer no help--Close down? You are of no service to anyone.)
I brought him back to his apartment, and offered my support every night. All I could do is cook for him, and offer hope. He died in his sleep a few months later. Those few months he were hell. All he had was ibuprofen, and some other safe/no potential to abuse drug. We looked into Hospice, but his doctors said he wasn't sick enough. Sometimes I think they have just seen too much, and common sense is lost?
My point is I have seen too many people die in extreme pain, including my father who was under medicated by Hospice.
I got off track. This is about painkiller trials, and placebo. I just don't want ineffective pain medication on the market.
I am shocked this group of researchers are trying to blame patients. It's your ineffective new medications? Lets not run off to some third world country and get "positive" results, or rely on a bunch of smaller studies that can be buried--if not satisfactory?. (By the way, this is going on as we speak, and has been for years. They offshore these double blind studies to the lowest bidder, and surprise, "Our drug worked better than we could have predicted!". Present their wonder drug to the FDA, and start the Marketing Party.
Right now, a drug only has to be slightly better than placebo in order to get past the FDA. That statistical difference is so small, in so many instances; I think if some of you knew just how slightly your drug works better than placebo, you might just go off the medication.
MY father died 11 years ago, and I still have weekly nightmares of his absolute misery. His last words to me were "Son when will this end?". If I could do it over, I would have gone to the bad part of my county, and bought him Heroin.
> Whole lot of speculation in this article. I would guess more people know about the placebo effect in the U.S., than they did 20 years ago? Twenty years ago, we didn't question the efficacy of these drugs? Everyone I know now is very leery of any claims by drug companies, and their Wonder Drug.
I would argue this is now the case in most of the developed world, not just the US.
I don't know much about the American medical system, but in Canada, once you get into palliative care, you basically get put on narcotics, so your father probably would have been okay. For most people, this works well (there are some kinds of pain originating from the CNS that as far as I know have no good treatments).
The situation for out-patients is not that great though. All the strong analgesics make you loopy. This is fine if you are in a hospital, and under supervision, but it's dangerous if you're not in a hospital.
Unfortunately, the science just isn't there yet. Treating pain is like treating cancer: there are so many different forms of pain, different locations, different "causes", etc.
There isn't really a good alternative. "Americans" doesn't work; Simón Bolívar went to war to liberate "the Americans" from the Spanish. I'm surrounded by Americans here in Buenos Aires. But almost none of them are USians. And no, I don't know how to pronounce "USian" either.
Well, I just gave you an example of people from outside the US being referred to as Americans. It's a political issue, though, like the use of "men" to mean "people", so you can't solve it by appealing to common usage. People who don't like your usage are still offended by it even when they know it's common. And they probably won't bother to talk to you, or, in many cases, even learn English.
Me, I'd prefer to be able to listen to them, especially since I live in Buenos Aires. So I'm going to continue to not use "American" in that sense, which means I still don't have a better alternative to the abominable "USian".
Why does "Americans" not work? Isn't the USA the only country with "America" in its name? (And I thought Mexico's full name is the "United States of Mexico" so "US" is ambiguous, as well).
I've seen that usage a fair number of times, usually in discussions relating to some definition of "american" and why that does not include everyone in both North and South America. For instance is a Guatemalan not an American?
I am a European and I used that term because the article declared that USA demonstrated such results. For me, USians more clearly distinguishes different types of American. Since I have not read the scientific paper, I was not sure whether, say, Canadians had also displayed similar results. If not, I suspect there is possibly a political / legal / societal cause implied.
Amongst other ways, USA is unique in that it is, perhaps more than any other country, populated by people whose ancestors had an ideological reason for living there. It is highly speculative, but it would be really interesting if it could be shown that such an historical filter were somehow behind these results.
Were they using the same placebo? Perhaps 20 years ago they used a inert placebo, like (a sugar pill) and now they use an active placebo that mimic the side effects of the drug (for example, heartburn and indigestion). Both are placebos, but may have changed the supposed inert drug.
Placebo is highly sensitive to trial design, which is what I suspect is happening here. This is an intriguing effect, meaning that trials are increasingly reporting that all sorts of therapeutic effects are apparently less effective than previously thought. Or are they?
55 comments
[ 0.22 ms ] story [ 108 ms ] threadSo now a higher ratio of trial patients aren't I'll in the first place.
Pain is sometimes one of these. The pain the person reports is real pain - you can do scans and see it - but it doesn't have a physical cause.
It's likely that these people experience strong place o responses.
But the. We know that cognitive behaviour therapy is used for cancer patients, and we wouldn't normally describe their pain as psychological.
So, pain is complex and fascinating and difficult.
Here's a recent paper talking about placebo and anti-depressants.
http://onlinelibrary.wiley.com/doi/10.1002/wps.20241/abstrac...
A couple of hypotheses were mentioned in the article. One is that only in USA and New Zealand is direct-to-consumer drug advertising allowed. Another, which seems to have some supporting evidence, is that size and length of the drug trial may boost the placebo effect. The latter point is presumably about the production value of the trial, mandated by increasingly stringent FDA regulations. The apparent cost sells the apparent effectiveness.
I wonder whether these results might be linked to USians sustained high level of religiosity, which is also unique in the developed world.
Wait, I guess that's not really bad.
And then when the first suicide by placebo pills comes, we'll know we nailed it.
> Based on patients’ ratings of their pain, the effect of trialled drugs in relieving symptoms stayed the same over the 23-year period — but placebo responses rose. In 1996, patients in clinical trials reported that drugs relieved their pain by 27% more than did a placebo. But by 2013, that gap had slipped to just 9%.
Wouldn't an easier way to look at it be to completely ignore the effects of the actual drugs in those trials, and just look at the data of how many people took placebos and to what extent those people reported changes?
It is likely that the researchers and the audience want to understand how to get more effective drugs onto the market more than why placebos per se are becoming relatively more effective. This may explain the unexpected reporting emphasis.
In this reverse analysis you need the opposite: if the placebo effect is changing, that could either be because the placebo itself is more effective, or because people are reporting pain differently, or a number of other confounding factors. Analysing the people who took real painkillers too gives you a baseline to compare against.
My understanding was: you need the placebo to compare to to counter out the placebo effect (i.e. if you only test "real drug vs. nothing", the real drug might show an improvement purely due to placebo and have no actual effect itself). But introducing more variables... surely there are lots of other variables there regardless? If you ran a test of "placebo vs. nothing" what variables come in that wouldn't be there for "placebo vs. drug"?
Double-blind testing allows one to test the effects of administering anything (the placebo effect), versus the effects of the actual drugs. By administering an inert substance, the baseline placebo effect can be tested, as no actual medicative effects are expected from this inert substance.
When testing placebo vs nothing, in the nothing case a lot changes. The participants will no longer be administered anything, the experimenters will be aware of not administering anything, etc. etc. Because so many variables change in such a set-up, you can no longer actually tell which of these variables changing caused the effect and which of them didn't. Thus making any results and conclusion derived from the suspect.
Wait, this can be read as "Drugs are less efficient" just as well as "Placebo are more efficients".
If the gap between placebos and pain medications is narrowing, my first assumption is that the experimental designs have tightened over time, so that the blinding is better, and subjects and doctors are less able to distinguish between treatments. It's not that pain medication is "less" effective, it's that it was always only (say) 9% better than placebo, but in crappy earlier studies where it was less well blinded, some subjects (or researchers working with the subjects) could tell they were taking the placebo, and reported feeling worse, and some subjects could tell they were taking the treatment, and reported feeling better.
If you tell me the gap has narrowed while the subjective reports of the treatments' effectiveness has stayed the same, I still believe my first assumption above, but now I'm a little bit suspicious of actual researcher bias. There are a vast number of ways researchers could "tune" for a target effect size, from classic ways like choosing the stopping point of the experiment to more subtle things like interpreting patient responses up or down based on whether a running analysis is coming in hot or cold.
(The later could be less overtly malicious than you might think -- imagine that you're continually analyzing the results of your experiment, without breaking blinding. You "know", from past experiments, that treatment X is 70% effective. If you are seeing it as 80% effective in your running analysis, you might assume your interactions with your subjects are subtly biased to encourage over-reporting of the effect and consciously or unconsciously alter them to encourage under-reporting, and vice versa if you only see a 60% effect. If your blinding is still good, your gap in effectiveness will still narrow, but the "absolute effectiveness" of the treatment will still narrow.)
Double-blind rests, among others, on the assumption that a) placebo effects are independent of and additive with actual drug-effects, b) placebo effects are nulled when participants know they're being given a placebo.
Although I'm currently without sources (hopefully I can rectify this soon) both a and b are actually false. Certain drugs have been found to interact with placebo, which decreases or enhances their effects, thus again conflating the medicative effects of the drugs being administered and the effects of actually taking a pill (or rather going through the process of administering medication).
Second, placebo has become such a household term, that the original meaning as a "fake" effect, seems to have been lost. Logically this makes sense, because its the end result that matters (being cured), not the way in which the drug is effective (placebo or not). The average person being told they're receiving a placebo can actually increase the placebo effect, rather than nulling it.
A second variable should be introduced in double-blind studies, to account for both factors I mentioned above. Variable 1 should be whether or not they're receiving the actual drug (1a) or a placebo drug (1b) (as double-blind currently tests for) and variable 2 should be whether participants are told that they're receiving the actual drug (2a), or a placebo drug (2b).
Testing the effects of combination 1a/2b vs 1b/2a should test whether the drug actually interacts with a placebo effect. Providing a measure of validity for the 1a/2a vs 1b/2a test.
EDIT: Fixed some spelling mistakes
EDIT2: Found some sources: Guardian article on how placebo can be enhanced when participants know they're being given a placebo: http://www.theguardian.com/science/2010/dec/22/placebo-effec...
Published paper that used the set-up I suggested to find caffeine-placebo interaction effects - http://www.ncbi.nlm.nih.gov/pubmed/21092089
EDIT3:
This is the paper I remember reading on the subject - http://www.ncbi.nlm.nih.gov/pubmed/24416197
I'm not a medical expert, so I'm a bit inconsistent in terminology, but have ran my fair share of (cognitive) psychological experiments (and developed methods for them), which rely on similar logic as to that of double-blind testing.
Also, if a positive result was found in the "not given any drug" vs "given the active drug" trial, one would still not be sure if this was due to the participant going through the treatment, or due to the treatment + the effects of the drug.
"For companies trying to develop treatments, one remedy might be to compare new drugs against their best competitors instead of against placebo — or to go back to conducting smaller, shorter trials."
Lets Not do this. Lets continue to look for drugs that actually work on pain. The last thing we need is another drug that works slightly better than an ineffective pain drug that squeaked past the FDA? In the mean time there's a whole bunch of patients who are in extreme pain, and doctors are afraid to prescribe opioids. Doctors know opioid type medications work, but are afraid to prescribe them for a lot of reason, but mainly they don't want the FDA breathing down their backs.
My best friend died a horrid natural death a few years ago. He was in bad shape, but the doctors couldn't find anything really wrong with him besides being very old, and had a very bad case COPD. He was in constant daily pain.
I bought him to "the best pain clinic in San Francisco". They wouldn't alleviate his pain? They offed to cut some nerve, but wouldn't medicate him. (To that clinic, if you are going to stereotype patients, and offer no help--Close down? You are of no service to anyone.)
I brought him back to his apartment, and offered my support every night. All I could do is cook for him, and offer hope. He died in his sleep a few months later. Those few months he were hell. All he had was ibuprofen, and some other safe/no potential to abuse drug. We looked into Hospice, but his doctors said he wasn't sick enough. Sometimes I think they have just seen too much, and common sense is lost?
My point is I have seen too many people die in extreme pain, including my father who was under medicated by Hospice.
I got off track. This is about painkiller trials, and placebo. I just don't want ineffective pain medication on the market.
I am shocked this group of researchers are trying to blame patients. It's your ineffective new medications? Lets not run off to some third world country and get "positive" results, or rely on a bunch of smaller studies that can be buried--if not satisfactory?. (By the way, this is going on as we speak, and has been for years. They offshore these double blind studies to the lowest bidder, and surprise, "Our drug worked better than we could have predicted!". Present their wonder drug to the FDA, and start the Marketing Party.
Right now, a drug only has to be slightly better than placebo in order to get past the FDA. That statistical difference is so small, in so many instances; I think if some of you knew just how slightly your drug works better than placebo, you might just go off the medication.
MY father died 11 years ago, and I still have weekly nightmares of his absolute misery. His last words to me were "Son when will this end?". If I could do it over, I would have gone to the bad part of my county, and bought him Heroin.
I would argue this is now the case in most of the developed world, not just the US.
The situation for out-patients is not that great though. All the strong analgesics make you loopy. This is fine if you are in a hospital, and under supervision, but it's dangerous if you're not in a hospital.
Unfortunately, the science just isn't there yet. Treating pain is like treating cancer: there are so many different forms of pain, different locations, different "causes", etc.
Out of curiosity, where are you from?
https://news.ycombinator.com/item?id=10288270
I feel like the ambiguity exists only in some pedantic formal sense, that in colloquial usage it's clear enough what is meant.
I don't think I ever heard anyone from outside the US being referred to as "an American"
Me, I'd prefer to be able to listen to them, especially since I live in Buenos Aires. So I'm going to continue to not use "American" in that sense, which means I still don't have a better alternative to the abominable "USian".
I am a European and I used that term because the article declared that USA demonstrated such results. For me, USians more clearly distinguishes different types of American. Since I have not read the scientific paper, I was not sure whether, say, Canadians had also displayed similar results. If not, I suspect there is possibly a political / legal / societal cause implied.
Amongst other ways, USA is unique in that it is, perhaps more than any other country, populated by people whose ancestors had an ideological reason for living there. It is highly speculative, but it would be really interesting if it could be shown that such an historical filter were somehow behind these results.
A placebo placebo trial if you will. But what would you use as a control? ;-)
B/c people will rather eat pills to loose weight then to eat more vegetables and do some sports ...
More on this at the excellent Mind Hacks blog:
http://mindhacks.com/2015/07/08/cbt-is-becoming-less-effecti...