Did I ever mention that I worked on a large NIH-funded multi-center clinical trial of treatments for alcohol dependence that included naltrexone as one of the treatments? Although naltrexone had a statistically significant effect, that effect was just a small increment over the placebo effect. Trials for treatment of alcohol dependence tend to have a large placebo effect, which is probably mostly a being-in-a-study effect. Someone who volunteers to be in such a study is seeking help for their alcohol issues, and so is motivated to try to do well (versus, say, someone undergoing court-ordered treatment).
At least naltrexone did better than acamprosate, which didn’t appear to have any effect. Somewhat surprisingly, acamprosate received FDA approval while our study was in progress. (Although the approval was based on trials conducted by the pharmaceutical company concerned, the FDA must have been aware of our large study and so should have waited for our results.) Our study looked at combinations of treatments, based on the hypothesis that combinations would do better than individual treatments. That turned out not to be the case. For instance, the combination of naltrexone and cognitive behavioral therapy did no better than either one individually.
Just before our study came out in JAMA (https://pubmed.ncbi.nlm.nih.gov/16670409/), the NIH arranged for the study investigators to receive media coaching so as to ensure a unified message if their institutions put out press releases or if they were contacted by the media. I was shocked by how much of a “spin” was put on the results to make them seem newsworthy, probably because the NIH wanted to justify the tens of millions of dollars it had put into the trial. (I wasn’t the lead investigator from UNC at the time, so they didn’t try to coach me – I would have had a hard time parroting the agreed-upon talking points.)
Ironically, my more senior colleague on the study died of alcoholic liver failure a short time later. Until near the end he’d managed to keep his own alcohol dependence hidden from the other investigators, who were among the top alcohol dependence researchers in the country.
Although the effect of naltrexone was rather small, at least it was statistically significant. The other clinical trials I have worked on so far have generally not had a statistically significant effect at all. In contrast, including on contrast to many of your COIVD-related comments, based on talks I have heard by statisticians working on COVID vaccine trials, those have been much more successful than expected. (These have been talks by statisticians to statisticians on statistical issues, not “spin” for the media.)
Did I ever mention that I worked on a large NIH-funded multi-center clinical trial of treatments for alcohol dependence that included naltrexone as one of the treatments? Although naltrexone had a statistically significant effect, that effect was just a small increment over the placebo effect. Trials for treatment of alcohol dependence tend to have a large placebo effect, which is probably mostly a being-in-a-study effect. Someone who volunteers to be in such a study is seeking help for their alcohol issues, and so is motivated to try to do well (versus, say, someone undergoing court-ordered treatment).
At least naltrexone did better than acamprosate, which didn’t appear to have any effect. Somewhat surprisingly, acamprosate received FDA approval while our study was in progress. (Although the approval was based on trials conducted by the pharmaceutical company concerned, the FDA must have been aware of our large study and so should have waited for our results.) Our study looked at combinations of treatments, based on the hypothesis that combinations would do better than individual treatments. That turned out not to be the case. For instance, the combination of naltrexone and cognitive behavioral therapy did no better than either one individually.
Just before our study came out in JAMA (https://pubmed.ncbi.nlm.nih.gov/16670409/), the NIH arranged for the study investigators to receive media coaching so as to ensure a unified message if their institutions put out press releases or if they were contacted by the media. I was shocked by how much of a “spin” was put on the results to make them seem newsworthy, probably because the NIH wanted to justify the tens of millions of dollars it had put into the trial. (I wasn’t the lead investigator from UNC at the time, so they didn’t try to coach me – I would have had a hard time parroting the agreed-upon talking points.)
Ironically, my more senior colleague on the study died of alcoholic liver failure a short time later. Until near the end he’d managed to keep his own alcohol dependence hidden from the other investigators, who were among the top alcohol dependence researchers in the country.
Although the effect of naltrexone was rather small, at least it was statistically significant. The other clinical trials I have worked on so far have generally not had a statistically significant effect at all. In contrast, including on contrast to many of your COIVD-related comments, based on talks I have heard by statisticians working on COVID vaccine trials, those have been much more successful than expected. (These have been talks by statisticians to statisticians on statistical issues, not “spin” for the media.)