Is the placebo effect dependent on the amount of belief in it?

Question

For example, are people with habitually skeptical attitudes less likely to benefit from the placebo effect? Has this been studied? (For instance, one might seek a correlation based on professions such as law enforcement, science, etc.)

Example claim:

That is, the placebo effect is dependent not on the drug's effectiveness but solely on therapeutic intention and expectation.

Answer 1

The question is kind of ambiguous. What is "it?" Does the question mean "Is the placebo effect dependent on the amount of belief in the placebo effect?" or does it mean "Is the placebo effect dependent on the amount of belief in the effectiveness of the (alleged) treatment?" The first possibility seems pretty goofy and the discussion in the question seems not to support this interpretation. There is evidence on the second.

There is a clever paper by Anup Malani about this. Trials differ in how they split the subjects into placebo and treatment arms. Sometimes, it's 1:1. Sometimes it's 1:2. Sometimes it's other ratios. If you are in a 1:1 trial, you assess a 50% chance you are getting a sugar pill. If you are in a 2:1 trial, you assess a 33% chance you are getting a sugar pill. This is a convenient way to parameterize how strongly a person believes that they are receiving a potentially effective treatment.

Malani looked at hundreds of trials for ulcer drugs and anti-cholesterol drugs. In trials where there is a higher probability that you are assigned to treatment, you get better outcomes. Quoting:

The central finding is that a higher probability of treatment is associated with a significantly higher healing rate in H2-blocker trials and with a significantly greater reduction in LDL levels and side effects in statins trials. The coefficient estimates suggest that going from a probability 0 to a probability 1 trial, which generates the full placebo effect, increases the probability of healing an ulcer with H2-blockers by 0.06 – 0.22. It increases the reduction in LDL levels following statin treatment by roughly 45 – 92 percent. It also causes increase the probability of any side effects and usual side effects from statin therapy by at least 0.6 and 0.5, respectively.

Answer 2

There is some weak evidence that knowing that a placebo can work even if the patient knows it is a placebo.

• Kaptchuk TJ, Friedlander E, Kelley JM, Sanchez MN, Kokkotou E, et al. (2010) Placebos without Deception: A Randomized Controlled Trial in Irritable Bowel Syndrome. PLoS ONE 5(12): e15591. doi:10.1371/journal.pone.0015591

The study compared patients who received no treatment (the control) with patients who received sugar pulls and were told they were sugar pills, but that they might experience placebo effects.

Two-group, randomized, controlled three week trial (August 2009-April 2010) conducted at a single academic center, involving 80 primarily female (70%) patients, mean age 47±18 with IBS diagnosed by Rome III criteria and with a score ≥150 on the IBS Symptom Severity Scale (IBS-SSS). Patients were randomized to either open-label placebo pills presented as “placebo pills made of an inert substance, like sugar pills, that have been shown in clinical studies to produce significant improvement in IBS symptoms through mind-body self-healing processes” or no-treatment controls with the same quality of interaction with providers.

They found:

Placebos administered without deception may be an effective treatment for IBS. Further research is warranted in IBS, and perhaps other conditions, to elucidate whether physicians can benefit patients using placebos consistent with informed consent.

These were only mild effects and, by the authors own admission, this was only a "proof-of-principle" study.

There was some criticism about the strength of the result - Ed Yong summarised some in Discover Magazine's Not Exactly Rocket Science blog

In particular, it wasn't clear to what level (a) whether patient's misunderstandings of the contents of the pills affected expectations, and (b) whether patients' expectations affected the result.

This second omission makes this a less than ideal answer to the OP's question. Yong reports that a follow-up publication was expected to answer this, but I have been unable to find one. (Did I miss it?)

David Gorski of the Science-Based Medicine blog was also critical - mainly about the hype surrounding this result, but also suggesting that the patients still had an expectation of some success. Again, this undermines this as an answer to the OP's question.

While we are piling on weak evidence, there was an older study with no control group that showed with no control group that neurotics with no control group openly given placebos with no control group improved, with no control group. It was described by Dr Ben Goldacre in his Bad Science column.

Conclusion: I present no strong evidence that skeptics can or can't benefit from a known placebo, but there have been some interesting exploratory studies that suggest that, just perhaps, they can.

Answer 3

The way the term placebo is used in medicial studies it's a catch-all term for different effects. Let's look at an example:

A patient asks for a painkiller and a nurse gives a patient a painkiller as a pill. What effects are there? Without claiming to have a complete list:

1. The chemical in the painkiller binds to the targets that the pharmacompany who designed the painkiller identified as good targets for reducing pain.
2. The chemical in the painkiller binds to an offtarget or otherwise does something that the company that produced the drug doesn't expect.
3. The patient has mindfulness about what he feels when he contemplates the decision whether or not to ask for a painkiller.
4. The patient made a conscious decision to do something that reduces his pain.
5. The patient gets caring attention from the nurse.
6. The patient drinks a glas of water to take the pill.
7. The patient has an expectation that asking the nurse for help will help his problems.
8. The patient has an expectation that a pailkiller pill will reduce his pain because he believes in pills.
9. The nurse suggests to the patient that he will feel less pain.
10. The patient pays more attention to the feeling because he wants to know whether the pill works.
11. When the person is asked, whether he has less pain, he wants to report that the efforts the nurse made has been effective or that the nurse should do something else.
12. When the person is asked, whether he has less pain, he thinks he's expected to report less pain because he got a pill.

The effects from 3-12 are all something that would appear as placebo in a study. Whether or not the water has an effect depend much on beliefs.

Nancy Dougherty found in a self experiment that #3/4 had a high influence when she too placebos to improve her emotional state.

As far as 9 goes you have studies within the domain of hypnosis that investigate similar effects. Different people seem to have a different level of suggestibility and the effects are not the same for everyone.

Answer 4

At least one recent study shows that placebos work equally well when you know you are taking them

A study published in July 2017 compared the clinical effectiveness of several ways of administering placebos. No treatment (NT) was compared to blinded placebo (DP, for Deceptive Placebo where the patient thinks they are getting real medicine), openly labelled placebo with and without an explanation of the point of the treatment (OPR+ and OPR- *Open-label Placebo with or without Rationale). The study used a standard method for assessing pain control in a laboratory.

The key conclusion was that the subjective results showed that OPR+ and DP had about the same effectiveness implying that it doesn't matter whether you know you are taking a placebo but that it does matter whether you know what the treatment if intended to do. Objective measurements didn't differ much among the groups.

As the abstract describes it:

We conducted baseline and posttreatment measurements of heat pain threshold and tolerance. Apart from the NT, all groups received an application of a placebo cream. Primary outcomes were planned comparisons of heat pain tolerance and the corresponding intensity and unpleasantness ratings. Objective posttreatment pain tolerance did not differ among groups. However, for subjective heat pain ratings at the posttreatment tolerance level, groups with a rationale (OPR+ and DP) reported diminished heat pain intensity (t(146) = -2.15, P = 0.033, d = 0.43) and unpleasantness ratings (t(146) = -2.43, P = 0.016, d = 0.49) compared with the OPR-group. Interestingly, the OPR+ and the DP groups did not significantly differ in heat pain intensity (t(146) = -1.10, P = 0.272) or unpleasantness ratings (t(146) = -0.05, P = 0.961) at the posttreatment tolerance level. Our findings reveal that placebos with a plausible rationale are more effective than without a rationale. Even more, open-label placebos did not significantly differ in their effects from DPs.

The conclusion is interesting but there are some caveats. It was a study based on perceived pain tolerance (which isn't nothing but is harder to assess than, say, reducing the effect of a burn). And it was a small study with only a couple of hundred participants. But the essential point that the apparent effect of a placebo is equally present when you know you are taking it, seems notable.