Hypertension 2018

The Chinese Cochrane Centre in Chengdu.

In March 1999 the Chinese Cochrane Centre opened in Chengdu. Now that it has been established for some time, we are starting to see reviews on acupuncture that include the Chinese literature. I remember reviewing one on neuropathic pain for Cochrane,[1] but this one on hypertension was a surprise when it popped up on PubMed at the end of November 2018.[2]

To get a feel of the difference between these reviews it is instructional to read the paragraph in the background section entitled: ‘How the intervention might work’. The one I reviewed from 2017 reads as follows:

The overwhelming data from basic science support the idea that acupuncture mediates its clinically relevant effects via nerves, usually, but not exclusively, in deep somatic tissue (Dhond 2008; Kim 2008; Zhang 2005). EA stimulates all fibre types, since all nerve impulses work through alterations in membrane potentials mediated via voltage‐gated channels. MA mediates a mechanical stimulus, and therefore will only stimulate mechanosensitive nerve endings (Toda 2002; Zhao 2008). Release of adenosine via both techniques may mediate a local inhibition of nociceptive fibres (Goldman 2010). Some evidence suggests that in the central nervous system acupuncture may produce an analgesic effect by the deactivation of limbic areas (Hui 2010; Shi 2015). Alternatively, descending inhibitory modulation may also be regulated by acupuncture to enable the modulation of pain (Takeshige 1992).

Ju 2017 [1]

That sounds a bit like something I might have written myself ;-).

The more recent one on hypertension reads thus:

The mechanism whereby acupuncture can lower BP is unclear. Acupuncture use is based on the TCM concept that diseases are due to disharmony in the body and there are channels (or ‘meridians’) of energy flow (called ‘qi’) within the body that can regulate the disharmony (Kalish 2004). When selected places (also called acupoints) on the meridians or collaterals are stimulated, diseases caused by dysregulation can be treated (Kaptchuk 2002). Mechanistic studies have demonstrated that acupuncture can normalize decreased parasympathetic nerve activity (Huang 2005), which is thought to result in increased excretion of sodium; reduced plasma renin, aldosterone, and angiotensin II activity; and changes in plasma norepinephrine, serotonin, and endorphin levels (Kalish 2004). For the management of chronic conditions, acupuncture can be practically administered monthly or at most weekly. Therefore, for it to be effective, it would need to reverse the elevated BP permanently or have a sustained effect for one week or more. A short‐term effect on BP for hours or even one to three days would not be sufficient to make it a practical therapy.

Yang 2018 [2]

You can imagine that I am moved to write a comment on this…

John Longhurst sadly died prematurely with his wife when their light aircraft lost power and crashed in February this year. He was responsible for some of the best physiological research on cardiovascular reflexes and electroacupuncture. If you were to read his chapter in the second edition of the textbook Medical Acupuncture – A Western Scientific Approach,[3] and then read the paragraph above on how ‘the intervention’ might work in hypertension, you would be flabbergasted, I give you my guarantee. I can only assume that the expert peer reviewers of this review came from a traditional acupuncture background – hardly useful from the perspective of a Cochrane review ie the cutting edge of modern evidence-based medicine.

The paragraph fails to mention the data on modulation of sympathetic tone, which is likely to be much more important than parasympathetic activity since it exerts direct control on peripheral resistance and thence blood pressure. The authors then go on to state that weekly treatment is the maximum possible frequency for chronic conditions, and therefore acupuncture must show an effect for at least 7 days. This is lunacy if one considers the basic science evidence. In the laboratory, acupuncture has never shown direct effects on physiological parameters for longer than 72 hours. Consequently, I teach my patients to self-administer EA every 3 days for maintenance of chronic inflammatory conditions. I have not taught patients for management of hypertension, and this may require treatment even more frequently.

The authors’ conclusions in the hypertension review are shocking, and most unlike Cochrane’s usual equanimity:

At present, there is no evidence for the sustained BP lowering effect of acupuncture that is required for the management of chronically elevated BP. The short‐term effects of acupuncture are uncertain due to the very low quality of evidence. The larger effect shown in non‐sham acupuncture controlled trials most likely reflects bias and is not a true effect. Future RCTs must use sham acupuncture controls and assess whether there is a BP lowering effect of acupuncture that lasts at least seven days.

Yang 2018 [2]

I am particularly surprised by the use of ‘most likely’. Personally I do not think the data supports more than a ‘may’. But hang on a minute, let’s have a think about this. Bias from a lack of blinding is particularly relevant to subjective outcomes, but we are talking about blood pressure here, and the patients are not all yogis… Well, let’s face it, if they were yogis, they probably would not need treatment for hypertension. But my point is that if it is all mind over matter in terms of (acupuncture for) blood pressure control, and hypertension can be controlled with placebos, why poison ourselves? The efficacy over placebo debate here is ridiculous, we need to compare effectiveness and harms of different interventions.

RCTs must use sham controls! Why? So the patients cannot bias the result by seeing the needles as opposed to the colourful pills? Who cares? Surely we should design large pragmatic studies to see whether or not acupuncture (preferably self-applied 2 to 3 times a week) can compete with modern drugs – it is certainly unlikely to compete with antihypertensive drugs in terms of NNH– number needed to harm.[4]


Bias from a lack of blinding is of course more than simply expectation effects in the patient. It is important to blind outcome assessment, and blood pressure measurement lends itself to a high degree of automation. Statistical analysis should also be performed blind to group allocation. Unfortunately, the risk of bias (RoB) assessment currently recommended by Cochrane does not separate these aspects, so even if robots were taking measurements that were not subject to expectation effects, and analysis was performed blind to allocation, the RoB assessment would be the same in a trial where the patient could see needles and pills.

  1. Ju ZY, Wang K, Cui HS, et al.Acupuncture for neuropathic pain in adults. Cochrane Database Syst Rev2017;12:CD012057. doi:10.1002/14651858.CD012057.pub2
  2. Yang J, Chen J, Yang M, et al.Acupuncture for hypertension. Cochrane Database Syst Rev 2018;11:CD008821. doi:10.1002/14651858.CD008821.pub2
  3. Longhurst JC. Acupuncture in cardiovascular medicine. In: Filshie J, White A, Cummings M, eds. Medical Acupuncture – A Western Scientific Approach. London: Elsevier 2016. 394–421.
  4. Sheppard JP, Stevens S, Stevens R, et al. Benefits and Harms of Antihypertensive Treatment in Low-Risk Patients With Mild Hypertension. JAMA Intern Med 2018;178:1626–34. doi:10.1001/jamainternmed.2018.4684

Declaration of interests MC