In this retrospective matched-cohort study, Chinese researchers investigated the association of acupuncture treatment for insomnia with the risk of dementia. They collected data from the National Health Insurance Research Database (NHIRD) of Taiwan to analyze the incidence of dementia in patients with insomnia who received acupuncture treatment.
The study included 152,585 patients, selected from the NHIRD, who were newly diagnosed with insomnia between 2000 and 2010. The follow-up period ranged from the index date to the date of dementia diagnosis, date of withdrawal from the insurance program, or December 31, 2013. A 1:1 propensity score method was used to match an equal number of patients (N = 18,782) in the acupuncture and non-acupuncture cohorts. The researchers employed Cox proportional hazards models to evaluate the risk of dementia. The cumulative incidence of dementia in both cohorts was estimated using the Kaplan–Meier method, and the difference between them was assessed through a log-rank test.
Patients with insomnia who received acupuncture treatment were observed to have a lower risk of dementia (adjusted hazard ratio = 0.54, 95% confidence interval = 0.50–0.60) than those who did not undergo acupuncture treatment. The cumulative incidence of dementia was significantly lower in the acupuncture cohort than in the non-acupuncture cohort (log-rank test, p < 0.001).
The researchers concluded that acupuncture treatment significantly reduced or slowed the development of dementia in patients with insomnia.
They could be correct, of course. But, then again, they might not be. Nobody can tell!
As many who are reading these lines know: CORRELATION IS NOT CAUSATION.
But if acupuncture was not the cause for the observed differences, what could it be? After all, the authors used clever statistics to make sure the two groups were comparable!
The problem here is, of course, that they can only make the groups comparable for variables that were measured. These were about 20 parameters mostly related to medication intake and concomitant diseases. This leaves a few hundred potentially relevant variables that were not quantified and could thus not be accounted for.
My bet would be lifestyle: it is conceivable that the acupuncture group had acupuncture because they were generally more health-conscious. Living a relatively healthy life might reduce the dementia risk entirely unrelated to acupuncture. According to Occam’s razor, this explanation is miles more likely that the one about acupuncture.
So, what this study really demonstrates or implies is, I think, this:
- The propensity score method can never be perfect in generating completely comparable groups.
- The JTCM publishes rubbish.
- Correlation is not causation.
- To establish causation in clinical medicine, RCTs are usually the best option.
- Occam’s razor can be useful when interpreting research findings.