I only poked around the studies you listed a little, but there were a few things I noticed that made me more skeptical of the evidence:
I was initially surprised the SKY intervention made the list of primary recommendations. One of the studies comes from The Journal of Alternative and Complementary Medicine and the other is based off of a sample of 21 veterans with PTSD. Surely there are cheap interventions with better evidence bases?
It seems like you’re comparing apples and oranges for many of these interventions. Some interventions measure cortisol levels, some use self-reports, some use blood pressure. A 20% reduction in one will not translate to a 20% reduction in another. And presumably, what matters for the sake of this post is a reduction in stress, something that’s not measured directly by many of these studies.
It seems like you’re simply taking a raw average of all the studies within an intervention category to come up with an overall effect size. But this is invalid because some studies are higher quality and more informative than others (because of sample size, study design, etc.).
There’s definitely room for improvement with the rigour of the research analysis. I’ve discussed some of these areas in my reply to David Moss so I won’t repeat it all but to note a few quick things:
- There’s a lot more time that could be spent collating and reviewing evidence here but I’d expect that the effect sizes and recommendations would not change drastically. I definitely hope to come back and improve this article in future; as this is an early stage project I have no doubt there are improvements to be made. - I have combined different measures of interventions. Using a single measure would produce a more accurate result but would make it hard to find much evidence that presents effect sizes and enables prioritisation. In general, as I’ve noted in the article a lot of the evidence available is thin and poor quality, leaving somewhat of a choice between including interventions with poor quality evidence or leaving them out. I’ve combined measures on the assumption that this wouldn’t skew the results to a large degree but this may be wrong. - There is no direct measure for stress. Subjective stress reports are perhaps the closest but an individual’s perception of their stress levels may not correlate with their physical stress and this means they’d still likely suffer significant health effects from ongoing high stress.
I only poked around the studies you listed a little, but there were a few things I noticed that made me more skeptical of the evidence:
I was initially surprised the SKY intervention made the list of primary recommendations. One of the studies comes from The Journal of Alternative and Complementary Medicine and the other is based off of a sample of 21 veterans with PTSD. Surely there are cheap interventions with better evidence bases?
It seems like you’re comparing apples and oranges for many of these interventions. Some interventions measure cortisol levels, some use self-reports, some use blood pressure. A 20% reduction in one will not translate to a 20% reduction in another. And presumably, what matters for the sake of this post is a reduction in stress, something that’s not measured directly by many of these studies.
It seems like you’re simply taking a raw average of all the studies within an intervention category to come up with an overall effect size. But this is invalid because some studies are higher quality and more informative than others (because of sample size, study design, etc.).
There’s definitely room for improvement with the rigour of the research analysis. I’ve discussed some of these areas in my reply to David Moss so I won’t repeat it all but to note a few quick things:
- There’s a lot more time that could be spent collating and reviewing evidence here but I’d expect that the effect sizes and recommendations would not change drastically. I definitely hope to come back and improve this article in future; as this is an early stage project I have no doubt there are improvements to be made.
- I have combined different measures of interventions. Using a single measure would produce a more accurate result but would make it hard to find much evidence that presents effect sizes and enables prioritisation. In general, as I’ve noted in the article a lot of the evidence available is thin and poor quality, leaving somewhat of a choice between including interventions with poor quality evidence or leaving them out. I’ve combined measures on the assumption that this wouldn’t skew the results to a large degree but this may be wrong.
- There is no direct measure for stress. Subjective stress reports are perhaps the closest but an individual’s perception of their stress levels may not correlate with their physical stress and this means they’d still likely suffer significant health effects from ongoing high stress.