Inspired by Gang et al 2022.
SMD – standardised mean difference (effect size)key to acronyms
MA – manual acupuncture
EA – electroacupuncture
TENS – transcutaneous electrical nerve stimulation
TEAS – TENS over acupuncture points
This is the ninth of the BMJ Series of papers titled Acupuncture: How to improve the evidence base. It comes with the less than self-effacing acronym FAMOUS.
It is a meta-regression analysis of acupuncture RCTs published from 2015 to 2019 inclusive. Only trials with more than 100 patients were included and they had to have at least one patient-important outcome. 584 RCTs were included, and 1304 effect estimates were extracted from the data available.
Regression was first described by Sir Francis Galton as ‘regression toward the mean’. Essentially a sample of a measured variable, if extreme, is very likely to move closer to the mean when resampled later. So, regression is used to refer to a change in a variable between sampling instances, and this can be compared with the changes in other variables. A variable could be the degree of pain reported by a patient at the start and the end of a trial or the application or not of a treatment such as acupuncture.
Variables can also be any other recorded factor related to treatment, assessment, or methodology. The team started out with 46 variables, but only 20 of these met the inclusion criterion of less than 10% missing data. A further 5 variables were dropped from the multivariate analysis due to collinearity, that is the regression lines created by the variables were too similar for the effects of each variable to be assessed independently.
Of the 15 variables that were left, 5 proved to be significant predictors of outcome:
- Type of outcome (QoL showed biggest effects)
- Acupuncture type (penetrating better than non-penetrating)
- Frequency of treatment sessions (high frequency better)
- Number of centres (single centre better than multicentre trials)
- Funding availability (no report of funding associated with better effects)
The biggest effect appeared to be in the use of quality of life as an outcome, but moderately large effects were seen in the comparison of single centre versus multicentre trials, and importantly penetration versus non-penetration in the acupuncture stimulation type. A small effect was seen in favour of high frequency versus low frequency treatment, where high frequency was more than 3 sessions a week for acupuncture in chronic conditions.
Univariable analysis was also performed on all the variables excluded from the multivariable analysis. This churned out 17 that had a significant impact on the effect of acupuncture, although these results need to be looked at with more caution.
I will just pick out a few of the highly significant factors that caught my attention from the univariable analysis and their SMDs:
- High vs low (number of treatments) 0.48
- MA vs EA 0.21
- EA vs TEAS 0.42
- Expert vs literature (acupuncture regime) -0.56
- Graduate vs short course (practitioner education) -0.22
- Chinese vs English (publication language) 0.72
Note – positive SMDs indicate the first item is associated with larger effects of acupuncture and negative SMDs indicate the second item is associated with larger effects.
Food for thought… and we can debate the degree to which these represent reality at the webinar on Wednesday.
1 Gang W-J, Xiu W-C, Shi L-J, et al. Factors Associated with the Magnitude Of acUpuncture treatment effectS (FAMOUS): a meta-epidemiological study of acupuncture randomised controlled trials. BMJ Open 2022;12:e060237. doi:10.1136/bmjopen-2021-060237