When a trial reports that craniosacral therapy helped people with a particular condition, it's natural to take that at face value. And in some ways you should. A positive RCT is meaningful evidence, and dismissing it because CST is unfamiliar would be unreasonable. But reading a positive trial carefully means asking specific questions that headlines don't always answer.
Cautious reading isn't the same as sceptical dismissal. It's about understanding what a trial actually showed, how strong the showing was, and what would need to be true for the result to translate into clinical practice. That kind of reading leaves room for genuine findings while protecting against over-interpretation.
Was the primary outcome met?
Before a trial begins, researchers are supposed to declare a primary outcome — the main thing they're measuring, and what determines whether the trial succeeded. Pre-registration matters because it stops researchers from running a trial, finding nothing on the main measure, and then reporting a positive based on one of several secondary outcomes.
In practice, some trials get reported in ways that play up secondary outcomes or subgroup results even when the primary endpoint missed. The 2016 low back pain RCT had a primary endpoint that came in at p=0.060 — just outside the conventional threshold. Secondary outcomes in the same paper may have looked more favourable, but the trial's core question wasn't answered in the expected direction.
When reading a positive CST trial, check which outcome the paper is reporting as its main finding. If the headline result is a secondary outcome, or if the paper spends more space on subgroups than on the pre-specified primary, the methods section is worth a careful read.
How large was the effect?
Statistical significance tells you whether a difference is likely to be real rather than chance. Effect size tells you how large that difference is. Related but not the same. A trial can be statistically significant while reporting an effect too small to matter in practice.
Look at the actual numbers. How many points did pain scores change in the treatment group versus the control? Did the researchers compare that change to the 'minimal clinically important difference' — the standard benchmark for whether a score change matches something patients would actually notice? Many pain and disability scales have published thresholds for this. A result that looks impressive as a p-value can be underwhelming when you see the raw score changes.
This isn't a reason to dismiss CST research. It's a reason to engage with it more specifically. Some trials show effects that are both statistically significant and clinically meaningful. The 2015 neck pain trial met both. Others reach significance with effects that are modest in size.
Was the control group convincing?
The quality of the comparison matters a lot. A trial that compares CST to a waiting list or no treatment can't separate the effects of the therapy from the effects of having regular, attentive, hands-on contact with a caring practitioner. Those general therapeutic effects are real and valuable, but they aren't specific to CST. A sham-controlled trial, where the comparison group gets hands-on contact without the specific elements of CST, is a tougher test.
Not all sham controls are equally convincing. If participants can easily tell they're in the sham group, the comparison loses some of its value. Some trials ask participants afterwards whether they thought they got the real treatment, which gives a rough measure of how well the blinding worked. Where that information is reported, it's worth checking.
The other key question is follow-up length. A trial that measures outcomes immediately after treatment tells you less than one that checks at three or six months. Benefits that fade quickly may still be valuable, but sustained effects mean more clinically. Many CST trials use short follow-up periods, which limits what can be said about durability.
A carefully read positive trial is still a positive trial. Applying these questions to CST research doesn't undermine the findings that hold up — it puts them on firmer footing. The results that survive careful scrutiny are the ones worth taking seriously.