The Mechanism of Placebo Analgesia 1978, Levine et al (Claims Proof of Placebo)

So, a painful dental operation -
Patients were told that they might receive either morphine ('a powerful analgesic'), placebo, or naloxone (which might increase their pain).

There was blinded administration of naloxone and a placebo. It says that patients given morphine were excluded from the analysis of the results. Perhaps the reduction in pain from the placebo didn't match that of the morphine?

(Re when the D1 and D2 administrations happened. The abstract says 3 and 4 hours after surgery. The Methods says 2 and 3 hours after the start of anaesthesia (for the operation).)


Figure 1:

This figure is just for participants given placebo first. The first dotted vertical line is D1 when both sets of participants are given the placebo. The second vertical dotted line is D2 when one group is given the placebo (black dots - 17) and the other is given the naloxone (empty dots- 23).
Note that the administration of the first placebo at D1 didn't have much effect on the mean reports of pain. The trajectory of reports of pain seems very much in line with what was happening before D1. That trajectory continues for those given the placebo at D2, a slight increase in the mean. Those given the naloxone at D2 reported higher levels of pain.

So, so far, the mean results are in entirely line with what we might expect from an inert drug and a drug that blocks endorphins.

The authors then fish around in the data, identifying two groups among those given the placebo at D1 - a group who reported that their pain stayed the same or improved from just before D1 to just before D2. They called these people the 'placebo responders'. The 'placebo non-responders' were the people who reported that their pain worsened. Figure 2 is just for 23 people (the people were later given naloxone at D2). Figure 2 tells us that there were only 3 people in the placebo responders reporting an improvement of more than 2 on a scale of 1 to 10, whereas most people reported worsening pain (e.g. 10 people's reported that their pain increased by more than 2 points).

Figure 3 is again just for the 23 people who got the placebo first and naloxone second. It plots the mean pain scores for the people classed as placebo responders and as placebo non-responders throughout the study.



20 minutes after the placebo at D1, there isn't much difference between the groups. The difference in mean pain scores one hour after D1 for the two groups is just 2.5 points. And, remember, the grouping was done specifically to maximise the difference in pain scores. I think what we are seeing is probably some people assuming that they got the morphine at D1 and expecting an improvement, and more people assuming that they got the naloxone at D1 because the pain is worsening as they come off the anaesthetic. Indeed, it looks as though the pain trajectory before D1 has a lot to do with whether people reported increasing or decreasing pain. People who reported high pain levels after surgery were, at D1, less likely to report further increases in pain and more likely to report an improvement.

I think it's worth noting that people had to mark on a line from 1 to 10 where their pain was, and they weren't able to see their previous reports. So, it was hard for people to calibrate the reporting.

It's not a clean study, there are drop outs and data not reported, it's hard to know what the baseline pain is doing and there's the effect of residual pain control affecting assessments of pain trajectories. And it's not large. I don't think this study demonstrates that the placebo effect is anything more than an expectation effect.

We know that endorphins are produced when there is pain. It's entirely reasonable to expect that people given a drug that blocks the receptors for those endorphins will report higher levels of pain, although the actual impact here looks to be relatively small and possibly also short-lived.

I find it hard to believe that the whole hype of the therapeutic placebo is built on this study. I wonder if there have been later studies that present more convincing evidence. The asthma study is cleaner study of the placebo effect, and it found no real benefit. It's a much cleaner study - people with asthma given a biologically active drug, a placebo drug and nothing. People who were given either the drug or the placebo reported breathing easier, but only people given the biologically active drug actually had improved lung function.

Happy to hear if I have missed something in the study. What are your thoughts @DigitalDrifter?

 

Yes, intravenously.