Acupuncture as a complex intervention for depression: A consensus method to develop a standardised treatment protocol for a randomised controlled trial

Summary

Objective

To standardise a complex intervention by defining the characteristic (specific) components of treatment for a randomised controlled trial of acupuncture as an intervention for individuals who have been diagnosed with depression using a consensus method.

Methods

A nominal group technique was used. Potential components of the acupuncture intervention were generated from the literature, experts and participants. These were categorised as constant or variable, the latter including active management techniques (such as providing relevant explanations), auxiliary techniques (such as auricular acupuncture), and other aspects of patient care (such as offering life-style and dietary advice), all of which were underpinned by defined theoretical frameworks. Participants were selected on the basis of their experience and training, to encompass a diverse range of styles of traditional acupuncture practice in the UK, and all rated components in two rounds.

Results

Fifteen practitioners rated 52 variable components in the first round and 55 in the second. There was group support for 16 active management components, three auxiliary techniques and five areas of life-style support, all driven by eight theoretical diagnostic and treatment frameworks. For the 39 components that were rated twice, group support increased between rounds from 75 to 79% (z = −2.2, p = 0.03), while the absolute average deviation from the median dropped from 1.04 to 0.83 (z = −2.5, p = 0.011).

Conclusion

Standardising the characteristic components of a complex intervention for a randomised controlled trial of acupuncture for depression using a consensus approach is feasible. The method can be generalised to other clinical situations and other treatment modalities.

Introduction

Randomised controlled trials are regarded as the most rigorous method of comparing the impact of two interventions. In complex interventions built up from a number of inter-related components delivered by a therapist, defining the precise nature of the intervention can be problematic. An intervention can be described as “complex” when there is “difficulty in defining precisely what, exactly, are the ‘active ingredients’ and how they relate to each other”.¹ The Medical Research Council has identified defining the intervention as “the most challenging part of evaluating a complex intervention”, yet it gives little advice on how this should be done. Nevertheless, knowing the precise components of treatment within a randomised controlled trial is essential in order to know what it is that is being evaluated and to what one can attribute change to, therefore enabling the results of a trial to be more easily interpreted or replicated.

There has been a tendency in acupuncture research to attempt to simplify the complex nature of the intervention and focus on particular components that can then be treated as the ‘active’ ingredients of the therapy. In some trials, a particular group of acupuncture points is delivered to each patient in the same way at each treatment session. In other trials, there may be an individualised treatment strategy but each patient receives the same initial treatment at every session throughout the trial² or the same treatment for a month at a time, with changes based on reassessment.³ There is a theoretical advantage here in that the tighter the specification, the easier the interpretation of the results of trials, so that one can identify more precisely to what one can attribute the outcome to. The disadvantage of these studies is their low ecological validity when compared to normal practice, with the implication or poor generalisability. Should the outcome of a trial with a tightly specified intervention be negative for acupuncture, then interpretation might be problematic. There will be a “negative conundrum”: was it that that acupuncture did not work, or was the integrity of the intervention compromised to the extent that the treatment was simply inadequate? At the other end of the spectrum is the loosely specified trial protocol. This has the advantage that it may be more readily generalised to routine care. The downside is that if acupuncture practitioners incorporate additional therapeutic modalities into a treatment, such as herbs, nutritional supplements, massage and manipulation, only some of which are underpinned by Chinese medical theory, it will be less clear what one can attribute change to. In terms of the interpretation of a negative trial, the conundrum in this case is whether acupuncture really doesn’t work, or if there is a dilution effect due to some components being less effective than others, such that the outcome underestimates the potential impact. There remains a challenge to manage this trade off between a tightly and a loosely defined intervention (see Fig. 1).

A useful approach in standardising the components of a complex intervention can be derived from the Medical Research Council's categorisation of “constant and variable components”.¹ The “constant” components of treatment can be understood as fixed for all practitioners, all patients and every treatment. The “variable” components can be understood as treatment procedures and processes applied flexibly for each patient, yet always driven by accepted theoretical considerations, with principles that can be replicated to produce the same intended outcome.⁴

Our aim in this study was to involve practitioners in defining the parameters of a flexible standardised protocol for use in a pragmatic trial of acupuncture as an intervention for patients diagnosed with depression, with acupuncture provided as an adjunct to usual GP care. We limited our study to the components within this protocol that were seen by professional acupuncturists in the UK as specific to, and ‘characteristic’ of, acupuncture; these being the “therapeutic interactions or strategies that are theoretically derived, unique to a specific treatment and believed to be causally related to outcome”.⁵ Our intention for the final protocol was that it would provide practitioners with sufficient flexibility in their treatment options in order to match expected patient variability, yet with sufficient detail provided for replication.