Sunday, July 19, 2009

Acupuncture needles no good as toothpicks...

...yet funny enough toothpicks are just as good as acupuncture needles for providing relief of lower back pain.

The reference is
DC Cherkin et al. A randomized trial comparing acupuncture, simulated acupuncture, and usual care for chronic low back pain. Archives of Internal Medicine 2009 169: 858-866
but evidence-based healthcare knowledge collater, Bandolier has written a plain, easy-to-digest synopsis of the paper here.

The study looked at four therapies; individualised acupuncture, standardised acupuncture, simulated acupuncture with toothpicks and lastly, usual care. From Bandolier:

This large trial involved 638 adults, with follow up of 90% or above up to 52 weeks. Participants had an average age of about 47 years, with about 60% being women. About 70% had back pain for at least a year. The average initial RMDQ score was about 10.5 on a 0-24 scale, and average initial bothersomeness score 5 on a 0-10 scale.

The main results were these:

* There was no difference between individualised acupuncture, standardised acupuncture, or sham acupuncture.
* RMDQ scores fell from 11 to 6 for acupuncture of any sort by 52 weeks, compared with 7.9 for usual care. Any form of acupuncture was better than usual care.
* Bothersomeness scores fell from 5 to 3-5 to 4 for all four groups, with no difference between them.
* Use of medications (about 65% at baseline) fell to 47% with acupuncture, but remained at 59% with usual care.
* There was no difference in SF-36 mental and physical component scores.
* Cutting down on usual activities for more than seven days in the last month at 52 weeks was more common with usual care (18%) than with acupuncture.
* More participants with usual care missed work or school for more than a day (16%) than with acupuncture (5%-10%).
* There was no difference in total costs of back related health services between groups ($160-$221), though costs of acupuncture were not included.
* Adverse events occurred in 12/315 with real acupuncture, compared with 0/162 for simulated acupuncture, with one serious adverse event for real acupuncture.
* One patient in the usual care group went on to have back surgery.

There are a few interesting things that came out of the study.

The difference between standard care and intervention is significant, and confirms previous discussions suggesting the more theatrical the intervention the larger the placebo effect.

Secondy, the cost comparison is interesting - no real cost difference, and indeed acupuncture costs weren't included. So even the 'cost effective' argument is unfounded.

Thirdly, adverse effects - one serious adverse effect and 12 lesser, compared with zero for the toothpicks. This means about 4% of the patients had an adverse effect from a treatment which had absolutely no demonstrable benefit compared to toothpickery.

So let's see: there's no patient benefit, no cost benefit and increased risk of adverse effects. So why are NICE approving it for lower backpain?

Perhaps I'll drop them a line.

(How long do you reckon it'll be before a quackupuncturist declares that this proves acupuncture works, and the toothpicks were "accidently" letting the Qi energy move as it should?)



  1. I'm pretty sure they are already there with the toothpicks. When I was learning acupuncture, massage of the location was considered an acceptable substitute for people who were averse to needles (yes, they do sometimes consult acupuncturists) or young children.

  2. On page 157 of the full NICE guideline it says:

    "Evidence suggests that seeing an acupuncturist was better than usual care but that there is not much difference between acupuncture and sham. However, sham acupuncture is used as an active form of treatment by some practitioners, therefore this should be considered as a possible treatment."

  3. And, on page 13 of the ull guideline it says (emphasis added):

    "There are a plethora of treatments available for the treatment of non-specific low back pain. Not all of the treatments used have a strong theoretical underpinning. The differences and similarities between different therapeutic approaches are not always clearly explicated in the literature. Furthermore, for many of the individual treatment approaches used any therapeutic benefit is the result of both the specific treatment modality used and the non-specific effects of the therapist delivering the treatment. For therapist-delivered interventions the guideline development group took the pragmatic decision that it was the effect of the package of care delivered by the therapist or therapists that is of interest rather than the individual components of the treatment package."

    There is recent evidence that spinal manipulation for low back pain may be due to "non-specific effects of the therapist delivering the treatment".

    So, should placebo treatments be recommended for chronic low back pain? And, if the answers is "Yes", when should placebo treatments not be recommended?

    The evidence that supports all non-medical treatments recommended for chronic low back pain comes from RCTs that compare the treatment against non-treatment. With this trial design, there can be no blinding of treatments such as acupuncture, spinal manipulation, exercise, education, and talking therapies.

    The "non-specific effect" measured you compare treatment with no treatment, includes both placebo effects and biases such as "see what you want to see", and "say what you think they want to hear". The risk of such biases is especially great with with patient-reported outcomes such as pain and disability.

    Unless you conflate these biases with placebo effects (or ignore them), it is not correct to assume that there is god evidence for clinically important "non-specific effects".

  4. apologies for the finger troubles and typos. The 2nd last para should read:

    The "non-specific effect" measured when you compare treatment with no treatment, includes both placebo effects and biases such as "see what you want to see", and "say what you think they want to hear". The risk of such biases is especially great with with patient-reported outcomes such as pain and disability.

  5. Like Twaza I'm skeptical of the idea that the apparantly beneficial "placebo effect" seen in studies like this is clinically useful.

    Notably, in this study, although patient-reported pain seemed to get better on placebo vs usual treatment, lots of other outcomes didn't even do that. Including such important things as bothersomeness, SF-36 and cost of treatment.

    There was an effect on some other important measures like days off work, but the effect was hardly dramatic. And as Dr*T notes, this was a pretty theatrical placebo in a condition (pain) which is regarded as optimally responsive to placebos.