Acupuncture for back pain

A new study into the efficacy of acupuncture versus 'conventional' therapy has received some coverage today.

The paper found that at 6 months the response rate (33% improvement or better on 3 pain-related items on the Von Korff Chronic Pain Grade Scale or 12% improvement or better on back-specific functional status measured by the Hanover Functional Ability Questionnaire) was 47.6% in the verum acupuncture group, 44.2% in the sham acupuncture group, and 27.4% in the conventional* therapy group. The acupuncture was better than the conventional group, but real Chinese acupuncture (verum) was no better than 'sham' needling (just sticking in needles superficially).

Now this obviously suggests that (a) acupuncture might work better than conventional treatments in chronic back pain, and (b) that this is a general effect of sticking in needles and not something magic to do with 'Qi' (cf. the gate control theory of pain).

However, there are some oddities about this study. Obviously there is a real problem with blinding patients, and they were all aware that the study was into acupuncture whatever group they were assigned to, plus the physicians conducting the study were unblinded when they provided the conventional therapy (perhaps explaining why there were such low treatment rates for physio or painkillers). But the study also had some interesting criteria for responders and non-responders.

I haven't been able to figure out just why the primary analysis was 33% improvement on 3 pain-related items on the Von Korff Chronic Pain Grade Scale or 12% improvement on Hanover Functional Ability Questionnaire measues of back-specific functional status, and why these were primary outcomes, while Short Form Health Survey, or Patient Global Assessment scores were secondary, and the supplied references didn't help. But more intriguing is the requirement that no proscribed therapy be used, or the patient is then assigned to the unresponsive category.

The initial figures are 58%, 68%, and 71% responders for conventional, sham, and verum acupuncture respectively. Yet, once those who took proscribed treatments (we are not told what these might be for conventional therapy, for acupuncture it is anything other than rescue treatment for acute episodes of pain with nonsteroidal anti-inflammatory drugs to be taken on no more than 2 days per week up to the maximum daily dose during the therapy period and only 1 day per week during follow-up) are classed as non-responders, and those who missed the 6 month assessment, these figures were 27%, 44%, and 47%.

Since failure to attend follow-up was 6%, 3%, 3%, this suggests a whopping quarter of all patients were classified as non-responders for using proscribed therapies. Now given the generous allowance of painkillers allowed to the acupuncture groups we might think this is just fine since their acupuncture clearly wasn't working for them, but what were the conventional therapy subjects being excluded for? I can't read the German therapy guidelines, and the paper doesn't specify, so we can but speculate what terrible therapy was being used by these people, acupuncture perhaps? I hope not, the study does say that:

"Patients in all 3 groups were informed before randomization that
acupuncture would be offered after completion of the trial."

But I'm sure no scientist would be stupid enough to give treatment to participants within 6 months thus getting the patient classified as a non-responder (would they?) - but who knows how many patients, having got the idea of acupuncture into their heads, then went out and had some? This is especially worrying as these were chronic back pain patients who had presumably tried (and failed to respond to) conventional therapies, but were explicitly acupuncture naive, and who had signed up to a trial of acupuncture in chronic back pain.

So what we have here is a study comparing sticking needles into patients versus conventional therapy (presumably delivered by the same doctors) where only the interviewers (assessing outcome) were blinded, and where somehow, only half of patients were given analgesics in the conventional therapy group, where the acupuncture therapy group were allowed analgesics 2 days a week, and where half of conventional therapy responders were excluded for having 'proscribed' treatment that is never defined or quantified. I think I'll stick to the NSAIDs for now.

*According to German guidelines - out of 387 patients in this group: physiotherapy in 197; massage in 180, heat therapy in 157, ; electrotherapy in 65; 'back school' in 36; injections in 48; guidance in 56; infusions, yoga, hydrojet treatment, and swimming in a few, pharmacologic treatment (95% analgesia) in 183).

UPDATE
Having thought about this study for a bit - I can only assume that the study was excluding people for using additional therapy in the follow-up period. For acupuncture there were a set number of interventions (you could get more if you responded) - and for the conventional therapy this was also true:

"All interventions comprised ten 30-minute sessions, generally 2 sessions per week, and 5 additional sessions if, after the tenth session (Figure 2), patients experienced a 10% to 50% reduction in pain intensity (Von Korff Chronic Pain Grade Scale)."

But whilst things like physiotherapy interventions could be considered comparable, a consultation leading to a course of pain killers seems somewhat different since it is hard to know how long the drug therapy was allowed to continue in the conventional therapy group outside the month or so of treatment sessions. If the study is excluding people for having therapy after the study interventions (and it is not clear whether this is the case) that might explain why so many people got excluded as non-responders. The declared permitted medication for acupuncture patients was:

"For acute episodes of pain, only rescue medication was permitted in both acupuncture groups. This was strictly defined as nonsteroidal anti-inflammatory drugs to be taken on no more than 2 days per week up to the maximum daily dose during the therapy period and only 1 day per week during follow-up. Use of any additional therapies for pain during the entire study period was prohibited"

And for conventional patients:

"Patients in the conventional therapy group received a multimodal treatment program according to German guidelines. The guidelines provide the treating physician with recommendations about the treatment algorithm and assess the various therapy forms according to the degree of evidence based on a literature search and recommendations of the specialist associations. Conventional therapy included 10 sessions with personal contact with a physician or physiotherapist who administered physiotherapy, exercise, and such.

Physiotherapies were supported by nonsteroidal anti-inflammatory drugs or pain medication up to the maximum daily dose during the therapy period. Rescue medication was identical to that for the acupuncture groups."

What I worry about here is that the use of further NSAIDs after the initial treatment period but in the follow-up period of six months might have been classed as 'proscribed' in the conventional therapy group as well. In that case we are comparing an intensive acupuncture intervention with intensive physician/physiotherapy interventions, but where the available pain and anti-inflammatory medication for the conventional group is restricted to the short (1 month or so) intervention period (a period in which acupuncture patients are permitted 2/7 days per week of full NSAID therapy).

An interesting line from the study says:

"Patients in both acupuncture groups also had clinically meaningful better results for all secondary outcome measures, including medication use (Table 6)."

But Table 6 doesn't provide any figures about medication use unfortunately. You'd really hope that the acupuncture group was using less medication than the conventional therapy group though, wouldn't you?

Table 6 does, however, provide you with information on initial treatment response - that is response at 6-weeks - which, given the limitations of the unclear exclusion criteria for 'proscribed rescue medication', might provide us with the most reliable estimate of relative efficacy. They found treatment response of 56%, 59%, and 61% for conventional, sham, and verum acupuncture respectively. Given what we know about the poor efficacy of both NSAIDs and acupuncture against placebo (the Bandolier site is a good resource for information on therapy for chronic back pain) the most conservative conclusion would be that very little works for chronic back pain - probably due at least partly to its complex psychosocial nature as Ben Goldacre points out.