When both the trial drug and placebo are effective

I was reading a study in JAMA evaluating the ise of krill oil (which I’d never heard of before) for knee pain, and was unsurprised to read that its effect was no different from a placebo. Looking at the data tables and summary, however, I was struck that both the krill oil and placebo were followed by reductions in pain averaging 20 points on their 0-100 scale (about a 40% reduction in pain!).

That effect is pretty large, and it got me thinking (assuming that result would hold up) that maybe I was okay with snake oil companies selling something that everyone seems to think actually helps, including the patient. (but then that got me thinking about homeopathy, which I have mocked for forever)

Anyway. Not sure I have a question here, but I’m interested if anyone has thoughts on all this.

Which means krill oil acts as a placebo, and has no additional medical benefit.

I am paywalled. What was the placebo? Some other oil?

If 40 percent reduction in number of patients with any pain, that is impressive.

But I’m not sure how existing pain intensity can be reduced to a number. You can ask someone to rate their pain from zero to ten, but I question whether a four is in some real sense twice as painful as a two. When asked this question in my own medical care, I feel unable to give a good answer. Pain is not a number.

JAMA is free, but it’s a pain to register with. Here’s their infographic. The placebo was a soft gel.


As for pain measurement, it’s not measuring the actual pain, rather your perception of that pain. I dunno, 10% reduction would’ve felt squishy to me, but 40% is big.

That’s what Extra Strength Placebo is for, dummy! Sheesh.

So should I advise my 97 year old mother to massage a bit of vegetable oil into her knees daily (if she can do herself, or she could get help)? She has terrible knee pain and is obviously no longer a candidate for knee replacement surgery (which she considered when younger but did not get due to fear of both post surgical pain and pain meds).

my thoughts: drugs that don’t do better than placebo should not be sold since all drugs have side affects and none of them are free.

Krill oil does nothing for knee pain so both krill oil and some neutral substance should and did result in nearly identical reporting of pain reduction. The placebo effect is too often referenced through the phrase “taking a placebo”. A sugar pill or vegetable oil in this case are not "placebo"s, they are props used to produce the placebo effect. Take a look at this paper about Placebo Studies and Ritual Theory. Everything about the process of using medications, even just talking to someone in a white coat results in reported improvement in well being.

I would love to see some studies in placebo effect on subjective vs objective measures.

Pain is a very objective measure. Some explanations for the reduced pain or perception thereof

  1. Feeling like you’re finally doing something. Pain can be influenced by stress, and feeling like you’re actually doing something will reduce that stress and therefore the pain.
  2. Subconsciously the subjects want to please the experimenters, and reduce their pain rating.
  3. Paying closer attention to their pain levels could might make them (subconsciously ) reduce their perception of the pain. I didn’t read this experiment, and I haven’t heard if they keep a pain diary before getting the treatment. That would mitigate this factor.
  4. And (of course) the subjects really believe it will help

There could be other factors, depending on experiment setup as well.

If the source of the knee pain is something like an injury, as opposed to something degenerative like arthritis, then we have the simple reality that the body heals itself over time.

So in the former case krill oil, random vegetable oil, and doing nothing at all would all have the same 40% reduction over the time interval.

OTOH, if we’re talking about chronic degenerative pain such as from arthritis, that isn’t getting better over time on its own, then that explanation is inapplicable.

Well, when i had small children, i mixed up a bottle of placebo (water, corn syrup, and a drop of red food coloring) and gave them a spoonful every four hours when they were sick but didn’t have a fever. It always helped.

I believe in placebos. I take vitamin C when i have a cold. Hey, it might help, and it’s pretty harmless. My prior doctor recommended glucosamine and chondroitin for knee pain. I asked her about studies that showed it didn’t work. She basically replied that the placebo effect is real, and the stuff helped her.

Perhaps some of the subjects could be given a placebo effect blocker.

The thing is, knee pain tends to wax and wane. How about a third group with no treatment. Drug group, placebo group, pain diary only group.

I read a drug study once comparing a NSAID to a placebo pill for treatment of hip pain in dogs. When the person was told the pill they were giving was expensive, their treatment evaluation (pill or placebo) improved dramatically.

Do we know that for sure though? This could be some kind of weird situation where both the test drug and the placebo are both effective for some reason we don’t know yet. Something like the ratio of omega-3 to myristic acid is in some band, which has pronounced anti-inflammatory effects and both krill oil and “combination vegetable oils” both have that ratio.

Not that I’m saying I don’t think it’s a placebo, but good science would imply that we’d need to find that out before just dismissing that result as both being placebos, I’d think.

BBC Horizon has done some episodes on placebos. Here’s one with Dr. Michael Mosely from 2018. (He’s one of my favorite British medical presenters.)

It’s all wonderfully weird how this works.

FWIW 1: studies in pain relief are notoriously difficult to perform because of high responses to placebo. (The most difficult to perform are actually studies for topical treatments). I assume the mechanisms behind this have been studied, but I’m not familiar with current thinking on this.

FWIW 2: Visual analog scales are use by patients to report pain levels - they’re like a ruler, with scores from 0 to 10 for pain levels* (no pain to worst imaginable pain), and the patient indicates the point on the scale which represents their pain. These are described as a validated assessment tool. I have no idea what this means, and have never managed to obtain an explanation of how this could be done, especially as pain reporting has to be subjective.

So: I’m not surprised by the result (having spent decades in medicines licensing)

This sounds like a famous Ig Nobel Prize winner. See 2008, medicine.


* - this study appears to have used a 0-100 scale.

Yes. That’s exactly what placebo groups are testing for. And it could be any part of the procedure that accounts for the improvement, or including that the placebo group did not receive an inert substance, but rather one that shares some characteristic with the drug being tested.

(My response here is not directed at bump specifically)

There are both technical and nontechnical uses of the word “placebo” and “placebo effect” being used here.

A drug that does no better than the placebo in a study, but both drug and placebo groups show marked improvement over a no treatment group does not mean that you could give literal sugar pills and have the same improvement because " it’s all in their heads."

In the technical sense, placebo is supposed to account for all of the actions/interventions that the experimental group receives, except for the intervention being tested. If there is a significant but more-or-less identical improvement between the experimental and control groups, it most likely means that something in the experimental design is having an effect. Many times, the cause is something identifiable, and not something like, “if you believe your injury is being healed, it will heal faster and be less painful, because…the power of the brain!” (Not saying anyone here is saying that. Just that “placebo effect” is often taken that way.)

An example is a study on a certain herbal gel that people swear by as making bruises and swelling go away faster than without it. A study backed that up, sort of. The experimental and control (placebo - using the carrier gel, but with no special herb) groups both beat the no treatment group. Further study showed that massaging the gels into the injured area was the intervention that was responsible for the improvement. So that effect was a placebo effect. It was a part of the design that was repeated in the control group to control for the non drug interventions that were part of the experiment design.

So, it worked in exactly the way a control/placebo groups is supposed to, and has zero to do with the power of the mind, or subjective pain assessments. Which is often what people think a placebo effect is about.

In college I was a paid volunteer in a clinical trial study looking at oral osmoreceptors and their effects. They wanted me to be very dehydrated, which they achieved by overnight hospitalization with no fluid intake for 24 hours.

I had to rate my dehydration on a 1-5 scale. I was a 3 after 24 hours with no fluid intake. Then they placed an IV and gave me hypertonic saline, quickly making me more dehydrated. I was thirstier than I’d ever been, so I was at a 5.

Then they made me more dehydrated. And more. I had them go back and change prior numbers. It was horrible. Eventually I was allowed to swish crushed ice around in my mouth, but I had to not swallow, but spit out the melted ice. All while having blood draws to show my serum osmolarity was unchanged, but my serum ADH levels had changed.

The study paid well, but after finishing I had to spend another 24 hours in the hospital to get back to normal.

There are actually two different pain scales in common use. I HATE that one, and never know what to say. Also, i have a vivid imagination and never score very high on it.

There’s a similar pain scale that’s calibrated:
0 no pain
2 can be ignored
4 interferes with tasks
6 interferes with concentration
8 interferes with daily needs
10 bed rest required

I’ve hit “10” on that scale, briefly. And i can answer on that scale. I think it’s much more useful.