Hacker News Reader

Comment by LamdbaMamba | Hacker News Reader

Well... maybe. Whenever I see folklore like this I always suspect there are real effects but they may be attributed incorrectly. For instance, an ancient Greek feels sick, and worships at the temple of Apollo to ask the god to heal him. The Greek starts feeling a bit better when they leave. Apollo or the placebo effect?

My first thought with telling the bees is that just speaking our thoughts to an impartial third party can help ease our burdens and make us feel better. Maybe it's a therapist, a stranger at the bar, or a garden of bees. Now this doesn't explain the supposed effect of bees dying/leaving if the practice is not followed, but it could explain why the practice continues.

View on HN

Maybe it's as simple as a tradition that helps make sure that nobody forgets to take care of the beehives (an important part of a community's agricultural infrastructure) when somebody dies or leaves the home (due to marriage, etc.)

That's an excellent observation.

> Apollo or the placebo effect?

Note that the placebo effect is not about real improvement perceived by the sufferer except in a few very specific symptoms (pain, high blood pressure, and some psychiatric illnesses, mostly). In most cases, the placebo effect is simply optimistic interpretation/collection of data by people wanting to see the medicine work.

I'm curious why you think this. Is there some research to support this view? From what I've read, the placebo effect seems quite real, with quite physical effects. And the surgical placebo effect is stronger than the medicinal placebo effect.

It's not so absurd if you fully accept that the connection between mind and body is bidirectional.

Placebos aren't curative. It's well known that they can affect a subject's perception of how their condition is progressing, and this may have physical consequences: e.g. a reduction in psychological stress leading to reduced blood pressure, etc. But the placebo won't affect the underlying condition. They don't make the brain somehow cure the body.

Quote:

"Placebos won't lower your cholesterol or shrink a tumor. Instead, placebos work on symptoms modulated by the brain, like the perception of pain. "Placebos may make you feel better, but they will not cure you,"... "They have been shown to be most effective for conditions like pain management, stress-related insomnia, and cancer treatment side effects like fatigue and nausea."

https://www.health.harvard.edu/mental-health/the-power-of-th...

There are multiple factors that contribute to the placebo effect. The psychosomatic response is one of them. "Optimistic interpretation/collection of data by people wanting to see the medicine work" is another factor, but one that's typically removed by double-blinding. With some conditions, just plain reversion-to-the-mean is also part of the placebo effect--if you take a pill to cure your headache and your headache goes away, who's to say the headache wouldn't have gone away of its own accord anyway?

Hmmm nit picking but I think regression to the mean is different from placebo.

Regression to the mean: most headaches go away within a few hours.

Placebo: your favourite brand of paracetamol seems to cure headaches quicker than the unbranded version.

They can interact, of course.

Placebo refers to any effect where a patient's outcomes seem to improve for reasons not related to the specific intervention attempted. Regression to the mean, incorrect data collection, psycho-somatic effects etc are all elements of placebo, in various degrees in various instances.

Well, fair enough. To split hairs ever finer, though...

Placebo is present in all circumstances, as is nocebo.

Regression to the mean is a statistical phenomenon. It can help aid understanding of the size and role of placebo effects, but it isn't a part of the placebo effect. Ditto data collection and analysis.

But yeah, I'm probably splitting hairs :)

I am curious why you think that it is different. The base assumption should always be that non-medicine have non-effects - apart from perceived effects and the desire to "co-operate" with the study (which is why we double-blind tests are so important).

[0] is a Cochrane study that looked at this. Quoting from their conclusions:

> We did not find that placebo interventions have important clinical effects in general. However, in certain settings placebo interventions can influence patient‐reported outcomes, especially pain and nausea, though it is difficult to distinguish patient‐reported effects of placebo from biased reporting. The effect on pain varied, even among trials with low risk of bias, from negligible to clinically important. Variations in the effect of placebo were partly explained by variations in how trials were conducted and how patients were informed.

[0] https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD...

The Cochrane review only looked at randomised placebo-controlled trials. That's reasonable, and a good and necessary thing to do. But by definition it hasn't looked at non-randomised trials, or pragmatic trials (i.e. under "real world" conditions) or circumstances where a "pure placebo" is impossible or unethical (e.g. suicide prevention programs, pregnancy termination procedures, etc).

It also does point out that "pure placebo" can have a small but real effect.

Clinically meaningful effects are important: if patients feel that something doesn't work well or they don't like it, then it doesn't matter how convincing the statistical analysis is, the intervention will have poor uptake.

Outside of pharmacology trials, "pure placebo" is rare. Every intervention contains a degree of placebo and nocebo. This is normal and we shouldn't seek to deny it, or we risk wasting resources on meticulously designed interventions that work well in tightly controlled lab conditions but fail horribly in the real world.

Caveat: I fully believe that most modern medicine generally works (although often not for the reasons we think) and wouldn't want my comment to be misconstrued as bashing medicine.

> But by definition it hasn't looked at non-randomised trials, or pragmatic trials (i.e. under "real world" conditions) or circumstances where a "pure placebo" is impossible or unethical (e.g. suicide prevention programs, pregnancy termination procedures, etc).

They looked at quality studies, not bad studies that fail to control for all the confounding factors, yes.

> This is normal and we shouldn't seek to deny it, or we risk wasting resources on meticulously designed interventions that work well in tightly controlled lab conditions but fail horribly in the real world.

Randomized controlled clinical trials do happen in the real world, not in the lab, at least in general. They are the best possible way to study these effects, since real sick people in real hospitals are administered real/placebo medicine by real healthcare professionals. The data is collected by these professionals, who do not know whether they are administering placebo or real medicine.

And what we see in these conditions is what we expect a priori: for some conditions (those that are known to be semi-consciously controlled), there are real improvements from the placebo effects, for others only the real medicine has any effect at all.

I really don't understand what your point is. You claim that you believe that modern medicine works, but you also claim it "often" works for reasons that we don't understand - are you implying most modern medicine is placebo? If so, why is it improving overall?

> > by definition it hasn't looked at non-randomised trials, or pragmatic trials > > They looked at quality studies, not bad studies that fail to control for all the confounding factors, yes.

That's quite the straw-man argument. You've equated RCTs with being "(good) quality" and anything else with being "bad". So, to address those assumptions:

Not all trials are good, even if they are double-masked randomised placebo-controlled. (I'm not saying they're all bad, either.) To be fair, the Cochrane review process and other similar processes do assess quality of the papers they review, and the authors in the cited paper do this (see their Figures 1 and 2). They point out several weaknesses in the studies they reviewed: "All included trials were randomised, but in only 28 trials (12%) was it clear that patient allocation had been adequately concealed. [...] We regarded the risk of bias as low in 16 trials (8%) [...] In 61 trials the comparison between placebo and an experimental active treatment was described as ‘double blind’, whereas in the remaining 141 such trials comparisons were not double blind (or not reported). Observer‐reported outcomes were clearly assessed by a blinded observer in 22 trials, but this was unclear in 41 trials."

In other words, the raw materials they have reviewed were of low-to-moderate quality in a lot of cases. This is not to say the conclusions of the Cochrane review are wrong, it is just that the quality of their raw data (and their assessment thereof) must be born in mind.

The authors also make a somewhat sweeping judgement, which is tucked away here: "In 29 out of the 234 trials (12%), outcome data had not been reported in a way that was suited for meta‐analysis [...] Based on a qualitative assessment, there was no clear tendency for the findings in the 29 trials without outcome data to be different from the findings in the 202 trials we meta‐analysed." This seems to mean that they decided that 88% of the trials that could not be statistically assessed should be considered similar to the 12% they could statistically assess, based on some "qualitative assessment" that is not explained in the review. It would be very useful to have the details of this qualitative assessment made public - how was it done? Again, this isn't to denigrate their findings, but it does demonstrate that even in a strongly quantitative Cochrane review, there is a component of subjectivity.

The above brings up another point: not all reviews are good quality. This one probably is (the Cochrane process is rigorous) although as I've suggested there can still be subjective judgement in even the most quantitative approach. That's perfectly normal and can be fine, iff subjective processes are described in full.

Next point: not all non-RCTs are bad. There is nothing wrong with, for instance, a well-conducted pragmatic trial since it will expose what actually happens in the real world. Patients forget to take their pills, or refuse to do so. Doctors are 'just human', they rely on their instincts (and that's neither "good" nor "bad", it just is). There is absolutely no point in wasting years and millions of $CURRENCY in developing an intervention in a lab if it fails in the clinic for reasons that could have been uncovered beforehand - and research waste is a major problem. Non-RCTs are a part of this process. I'm not saying that RCTs are not part of this process; they have their place alongside other forms of investigation. RCTs are simply not a complete scientific healthcare methodology in themselves.

Next point: RCTs are not always appropriate. It's a real shame that the myth of the RCT as the One True Way persists. Yes, the RCT is the gold standard in some disciplines, and rightly so. But it isn't The Only Way, and other approaches can be appropriate. A huge amount has been written on this in recent years. For example in [0]: "[Evidence-based medicine] can and should shift from evidence-based individual decisions (in which the evidence is generally simple, with a linear chain of causation and derived from randomised controlled trials) to evidence-based public health (in which evidence is complex, with non-linear chains of causation and derived from a wider range of research designs including natural experiments and community-based participatory research)." Frankly, this seems obvious to me. People live and exist in highly non-linear, stochastic situations.

> You claim that you believe that modern medicine works, but you also claim it "often" works for reasons that we don't understand - are you implying most modern medicine is placebo?

Nope. I'd point out that healthcare is broader than medicine, and medicine is broader than pharmacology. But let's start with pharmacology since RCTs are well-suited here. We certainly understand pharmacokinetics very well, and the procedures that take drugs to market are robust, necessary, and rightly include placebo-controlled RCTs. But most guidelines for drug prescription are based on single drugs for single diseases [1] that have been trialled in non-representative patient groups [2]. This falls apart when the patient is in their 80s and taking multiple medications [3]. And that's just a pharmacology example where RCT is very well suited.

Finally, I reassert that the placebo effect is omnipresent. It makes no sense to compare "real medicine" with placebo since "real medicine" includes placebo whether we like it or not. Removing confounding factors under controled conditions is necessary to gain a particular understanding e.g. risk ratios, number needed to treat, etc etc. These are all good and necessary, although difficult to do well and prone to misunderstanding e.g. [4], [5]. But healthcare doesn't stop there. Even the most robust, rigorous, large-scale placebo-controlled RCT of drug efficacy won't stop the real-world reality of a red pill selling better than a yellow one, and won't reduce the value of humane nursing during chemotherapy, for example.

[0] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4556220/

[1] https://www.bmj.com/content/350/bmj.h1059.full

[2] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4632358/

[3] https://bmjopen.bmj.com/content/9/8/e031601

[4] https://journals.plos.org/plosone/article?id=10.1371/journal...

[5] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5455127/