Stimulated by Chen et al 2019.
I nearly overlooked this paper when it popped up on my PubMed search a few days after being published online in the Journal of Neurology on 21st August 2019. I guess it was the comparison with propranolol, and the fact that beta-blockers have been in use for so long in migraine prophylaxis. Topiramate seems to be the main drug of choice these days, and I still have the echo of the NICE guideline in my ears – CG150 claimed from a small network meta-analysis that topiramate is twice as good as acupuncture for migraine prophylaxis.
Adrian White and I were sufficiently surprised by this guideline conclusion to publish an open letter concerning the assumption of CG150 that sham acupuncture was the same as placebo topiramate, which we now know is not true of course.
CG150 was updated in 2015, but the analysis of acupuncture was not revisited, despite the unsafe assumption that sham acupuncture is equivalent to a drug placebo, and we are left with the following as the first line recommendation for prophylaxis:
Offer topiramate or propranolol for the prophylactic treatment of migraine according to the person’s preference, comorbidities and risk of adverse events. Advise women and girls of childbearing potential that topiramate is associated with a risk of fetal malformations and can impair the effectiveness of hormonal contraceptives. Ensure they are offered suitable contraception if needed. 
I’m not sure why topiramate comes at the start of the recommendation, since as far as I know it has never proved to be superior to propranolol, or to have a better side effect profile. Moreover, it seems to be no better than placebo in patients under 18.
So why did this group pick propranolol as a comparator, and what is meant by an indirect treatment comparison meta-analysis? Well I cannot answer that with confidence, since it is not clear from the paper. But this is a mere blog, so I am free to guess, aren’t I!
I suspect they chose propranolol to achieve sufficient data for a stable network, since there is likely to be more data on older drugs than newer ones. It is the first time I have heard of an indirect treatment comparison meta-analysis, and it looked awfully like a network meta-analysis to me. So I was pleased to find, after a bit of digging,[7,8] that it is indeed a form of network meta-analysis. I guess the title reflects the fact that there have been no direct comparisons of acupuncture and propranolol, so this network can only estimate effects from the indirect comparison created in the network.
Enough of the technicalities then, what did they find? Well they included 19 RCTs (n=3656) and demonstrated that acupuncture had quite a large effect over propranolol in terms of migraine episodes (SMD -0.74). Note that this is an effect size not an absolute reduction in episodes, so it is the mean reduction in episodes divided by the standard deviation of the distribution. An SMD of 0.8 or above is considered a large effect size. I was surprised they measured such a big effect, and then pleased to see similar sized effects over other drugs, including the dreaded topiramate:
A word of caution before you keen acupuncture activists hang out the bunting! Whilst the data on acupuncture versus sham acupuncture was robust (a thick line in the network), the data for the drugs was relatively flimsy (thin lines in the network), and in fact the comparison with propranolol was via a single trial against flunarizine including less than 100 patients. Having said that, I would not expect more data to change these results substantially, and they are in line with the conclusions of Klaus Linde and colleagues in their landmark Cochrane review in 2016.
How do we explain the difference between NICE and Cochrane or that of this analysis regarding the relative effect of acupuncture and topiramate? Well it is relatively simple actually. NICE did not allow direct comparisons of acupuncture with drugs, so just compared relative efficacy over sham or placebo drugs. They therefore failed to take account of the significant difference in effectiveness between sham acupuncture and placebo drugs. Both Cochrane and this analysis did include direct comparisons with drugs. Which do you think would be the more reliable in real life?
1 Chen Y-Y, Li J, Chen M, et al. Acupuncture versus propranolol in migraine prophylaxis: an indirect treatment comparison meta-analysis. J Neurol 2019;:1–12. doi:10.1007/s00415-019-09510-x
2 NICE guideline on headaches: diagnosis and management of headaches in young people and adults. http://guidance.nice.org.uk/CG150. 2012.
3 White A, Cummings M. Inconsistent placebo effects in NICE’s network analysis. Acupunct Med 2012;30:364–5. doi:10.1136/acupmed-2012-010262
4 Meissner K, Fässler M, Rücker G, et al. Differential effectiveness of placebo treatments: a systematic review of migraine prophylaxis. JAMA Intern Med 2013;173:1941–51. doi:10.1001/jamainternmed.2013.10391
5 Diener H-C, Tfelt-Hansen P, Dahlöf C, et al. Topiramate in migraine prophylaxis–results from a placebo-controlled trial with propranolol as an active control. J Neurol 2004;251:943–50. doi:10.1007/s00415-004-0464-6
6 Le K, Yu D, Wang J, et al. Is topiramate effective for migraine prevention in patients less than 18 years of age? A meta-analysis of randomized controlled trials. J Headache Pain 2017;18:69. doi:10.1186/s10194-017-0776-4
7 Jansen JP, Fleurence R, Devine B, et al. Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 1. Value Health 2011;14:417–28. doi:10.1016/j.jval.2011.04.002
8 Hoaglin DC, Hawkins N, Jansen JP, et al. Conducting indirect-treatment-comparison and network-meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices: part 2. Value Health 2011;14:429–37. doi:10.1016/j.jval.2011.01.011
9 Linde K, Allais G, Brinkhaus B, et al. Acupuncture for the prevention of episodic migraine. Cochrane database Syst Rev 2016;:CD001218. doi:10.1002/14651858.CD001218.pub3