Does needling sensation (de qi) affect treatment outcome in pain? Analysis of data from a larger single-blind, randomised controlled trial
- Correspondence to Dr Peter White, School of Health Sciences, University of Southampton, Highfield, Southampton SO17 1BJ, UK;
- Published Online First 10 May 2010
Background Insertion of an acupuncture needle into an acupuncture point typically generates a range of sensations called ‘de qi’. Most acupuncturists are taught that obtaining de qi is important when treating patients with pain but this can be quite uncomfortable for patients.
Objective This study assesses the importance of the strength of de qi, on the clinical outcome in osteoarthritic pain.
Method This study was part of a larger randomised, single-blind, multifactorial trial involving three interventions: real acupuncture (RA), Streitberger needle (SN) and mock electrical stimulation for the treatment of patients with osteoarthritis (OA) of the hip and knee. Patients were treated twice a week for 4 weeks. The two outcomes relevant to this study were pain reduction assessed by visual analogue scale and the Park needling sensation questionnaire, both measured at completion of the study. Two arms of the trial were analysed (RA and SN). Reduction in pain was correlated against strength of de qi for both RA and SN. Those who felt de qi were compared with those who did not.
Results 147 patients were recruited to the study (140 completed) with a mean pain reduction of 15.2 mm and mean de qi score of 6.2. There was no significant correlation between the strength of de qi and improvement in pain (p=0.49). There was also no significant difference in pain relief (p=0.52) between those who felt de qi and those who did not using the de qi subscale of the Park questionnaire.
Conclusion These data suggest that the presence and intensity of de qi has no effect on the pain relief obtained for patients with OA. This result may have implications for both acupuncture treatment and for future trial methodology.
Increasing numbers of patients are seeking out acupuncture treatment1,–,3 but there is much that we do not know about this intervention. Furthermore, there are many different acupuncture techniques, ranging from Western to traditional Chinese, and Japanese acupuncture. Consequently, practitioners may disagree as to ‘best practice’ and the exact methodology required to maximise clinical outcome. The mechanism of acupuncture is still unclear and might involve some kind of ‘energetic’ phenomenon such as the flow of energy (‘qi’) through channels (meridians). Equally, it might be operating by activation of the nervous system through diffuse noxious inhibitory control,4 the pain gate5 or through centrally mediated mechanisms6 or, indeed, a combination of these. It is entirely possible, however, that the mechanisms described in a traditional Chinese medicine and Western scientific context are one and the same—that is, that the former might be an early attempt to describe the nervous system. The large number of placebo-controlled, randomised controlled trials (RCTs) which show minimal or no efficacy when comparing ‘placebo’ techniques with real acupuncture (RA) might suggest that placebo is the central mechanism generating clinical response. Whatever the complex underlying physiology of acupuncture an intact nervous system is required in order for it to have an effect7 and this would certainly add weight to the argument that the generation of appropriate needle sensation when treating with acupuncture might be important clinically.
Insertion and manipulation of an acupuncture needle into an acupuncture point generates a range of sensations typically described as heavy, dull or deep ache, tingling, etc.8 9 This is called ‘de qi’ and is also thought by some to be detectable by the practitioner as ‘needle grasp’—that is, felt as the resistance to movement of the needle. Most practitioners would agree that this sensation, generated during needling, is important in creating a positive clinical outcome.10,–,13 It is therefore common practice to manipulate or stimulate the needle in situ in some way in order to generate appropriate de qi for the patient. While for many patients this sensation is tolerable, others find this process quite uncomfortable.
Understanding the role of de qi is important as this has implications both for clinical practice and the design of research trials. The aim of this paper was therefore to perform a secondary analysis of the salient data taken from a larger RCT in order to assess the relationship between the presence and strength of de qi and clinical outcome in osteoarthritic pain.
The main trial was a randomised, patient-blinded, multifactorial trial involving three intervention types: RA, Streitberger needle (SN) and mock electrical stimulation for the treatment of patients with osteoarthritis (OA) of the hip and knee. The results of the main study are yet to be published. This paper focuses on reporting clinical outcome (change in pain during the study) measured by a weekly visual analogue scale (VAS) with respect to needling sensation (measured by the Park questionnaire14) for the RA and SN acupuncture groups only. This allows us to investigate if the intensity of de qi and our primary clinical outcome of pain are correlated in any way.
Ethics approval was gained from the Southampton and South West Hampshire and the Salisbury and South Wiltshire Research Ethics Committees (approval number 170/03/t). Informed consent was obtained for each patient enrolled and all data were anonymised. STRICTA guidelines were used in designing and reporting this trial.15 Patients were recruited from joint replacement waiting lists at Southampton General and Salisbury District Hospitals.
Inclusion and exclusion criteria for the main trial
Patients were aged between 18 and 80 with chronic stable pain, predominantly from a single joint (hip or knee). All patients were diagnosed with OA and were awaiting joint replacement, scoring >30 mm during the baseline week on a 100 mm VAS, were not undergoing any active physical treatment (eg, physiotherapy). Patients were excluded if pregnant, had serious comorbidity (including severe back pain), a history of prolonged or current steroid use, were awaiting hip/knee revision, had needle phobia or allergy to sticking plaster.
After consent and 1 week of baseline pain recordings, eligible patients were randomised to one of three treatment types—that is, RA, a non-penetrating needle, the SN or a mock electrical stimulation of acupuncture points. For the purposes of this study, results from the mock electrical stimulation group will not be included. Randomisation was via an independent third party through a computer-generated list (Medical Statistics Department, University of Southampton) and stratified for joint (hip/knee) and sex. Sealed and numbered envelopes containing the individual treatment allocations were opened sequentially.
A ‘Western’ acupuncture approach16 was used. A range of prespecified acupuncture points was drawn up by consensus after consultation with various practitioners, experts and acupuncture texts17 18 before the start of the study. Local and distal points were specified for each joint (hip or knee) and practitioners were free to pick appropriate points from the agreed list depending on the progress of the patient and their presentation, this made the treatments much more pragmatic than using a completely prescriptive approach. An average of six points was used at each treatment, with needles staying in for 20 min. De qi was elicited typically three times during each treatment, when appropriate (ie, if the sensation had ceased) by oscillating the needles (rotation). Wherever possible, the aim was to elicit de qi on each needle. Some patients needed very little stimulation in order to engender de qi, whereas others required concerted effort; this individualised approach was felt to reflect clinical practice. Treatment sessions were twice a week for 4 weeks. Needles were single use, blister packed 30 mm×0.3 mm and 40 mm×0.25 mm, depending on the area, needles and body size. Needles were supplied by ‘AsiaMed’ Germany and ‘Cloud and Dragon’ China. A list of points used can be found in table 1.
Placebo acupuncture (SN)
The guidelines for point prescription, treatment duration, manipulation and treatment frequency were exactly the same as those for RA in all aspects. SNs19 were used instead of RA needles. These work rather like a stage dagger such that the shaft moves into the handle rather than into the body. These needles have been shown to be valid and indistinguishable from RA needles by patients.20,–,22
Three appropriately qualified and experienced practitioners were involved in the study. One was a member of the Acupuncture Association of Chartered Physiotherapists with 10 years experience and the other two members of the British Acupuncture Council with 10 and 3 years' experience, respectively.
Interested patients attended an initial assessment session and were then asked to consent, if appropriate. They were given a daily pain diary (100 mm VAS) and a record of their analgesia to complete for 7 pretreatment days. This was posted back and those who scored a mean daily pain of 30/100 were randomised and completed baseline questionnaires and treatment started. Once the course of treatment was ended, patients once more completed a daily diary for 7 days.
This was measured on a 100 mm VAS in the 7 days (ie, daily) immediately after the course of treatment. Patients were asked to ‘mark an X on the line at the point which represents how much hip/knee pain you have had on average in the last 24 hours.’ The anchors ‘No pain’ and ‘Worst pain imaginable’ were used at either end of the line. Mean pain scores were calculated for both pre- and post-treatment pain. The change in pain was calculated by subtracting mean post-treatment pain from mean pre-treatment pain.
Needle sensation/de qi
Needle sensation/de qi was monitored using the Park questionnaire.14 This was adapted from a measure devised by Vincent et al,8 which in turn was developed from the McGill pain questionnaire.23 It contains 25 descriptors and each is scored on an ordinal scale from 0 to 3 (with 3 being the most intense). A total sensation score was calculated by a simple summation of the scores for each patient giving a possible range from 0 to 75. MacPherson and Asghar11 also examined this measure using opinion from 29 international experts. They concluded that seven descriptors defined de qi and a further nine suggested acute pain at the site of needling. Patients completed the measure immediately after their last treatment session but were asked to record the nature and intensity of sensations they had felt over the entire course of treatment They were also asked the question ‘Do you think the treatment you had was real?’ yes/no, this was used to check that blinding had been successful.
Sample size and statistical analysis
An analysable sample of 70 subjects per group provided 80% power to detect a moderate correlation of r=0.30 using a one-sided 5% significance.
The analysis was on an intention-to-treat basis. Total sensation scores were correlated against change in mean pain using Spearman's correlation coefficient for the two treatment groups separately and combined.
Between January 2004 and August 2007 a total of 147 patients were recruited to the two relevant arms of the study (74 to RA and 73 to SN) as part of the main trial. Eighty-four were female, 88 had knee pain and the mean age was 67.0 years (SD=8.5). Seven patients dropped out of treatment before any outcome measures were collected (see CONSORT diagram, figure 1).
Blinding was successful, with 96% and 93% of people from the RA and SN groups, respectively, believing that their treatment had been real. Conversely, 4% and 7% of patients from RA and SN groups thought their treatments were not real. These results show no significant difference between groups (χ2, p=0.43)
The median scores for pain reduction and total sensation can be seen in table 2. A total of 24.3% of patients experienced an increase in pain, 1.4% stayed the same and 74.3% improved. It can be seen that those receiving RA tended to feel stronger sensations. This was statistically significant (Mann–Whitney, p<0.01).
The majority of patients tended to feel between one and six different types of sensations, with the mode being 3 (see table 3), although it should be noted that one patient recorded 18 different sensations.
The type of sensations recorded and the proportion of patients noting that particular sensation are shown in table 4. It should be noted that these figures reflect the fact that the majority of patients recorded more than one sensation.
A simple scattergraph suggests that there is no relationship between reduction in pain and strength of sensation (figure 2).
This was confirmed by the statistical analysis where correlation coefficients were calculated (table 5).
It can be seen that there were no correlations between change in pain and intensity of needling sensation, suggesting that the intensity of de qi during treatment does not relate to treatment outcome for OA pain. Those receiving RA tended to feel stronger sensations than those receiving SN (table 2). In an attempt to take account of potential ‘background noise’ of the questionnaire, which contains some descriptors relating to ‘pain’ rather than de qi, data were then analysed using the criteria for de qi set out in MacPherson's study11 (ie, using aching, dull, heavy, numb, radiating, spreading and tingling to designate de qi). Using this method, there was still no significant correlation between intensity of de qi and reduction in OA pain (Spearman coefficient=0.106, p=0.38). In order to ascertain whether patients were responding to ‘pain’ during the needling process, a correlation was performed using MacPherson's nine descriptors for ‘pain’ (burning, hot, hurting, pinching, pricking, sharp, shocking, stinging and tender). Again there was no significant correlation when needling pain was plotted against reduction in OA pain (Spearman's correlation coefficient=0.016, p=0.89).
An analysis of covariance was carried out to ascertain whether the presence of any of the de qi sensations, using MacPherson's criteria, is important for pain outcome in OA; this was not significant (p=0.317). These data, however, indicated an outlier; a patient who felt no de qi sensations but experienced a substantial fall in pain. Therefore a robust regression analysis was carried out but there was no significant difference between the two groups (presence of de qi vs total absence of de qi) and pain outcome (p=0.52). This would suggest that the presence of de qi has no impact on treatment outcome for OA pain.
This secondary analysis of data sought to examine whether there is a relationship between strength of de qi and improvement in pain for patients with OA of the lower limb. Our data suggest that the strength of de qi bears no relation to clinical outcome for OA pain in this study. Furthermore, the data imply that the presence of de qi during acupuncture treatment has no effect on acupuncture for pain relief.
We chose to include data from two arms of a larger three-arm, placebo-controlled trial because both of these interventions yielded ‘de qi’-like sensations, though the RA arm produced stronger sensation as would be expected. It should be emphasised that SN did not involve skin penetration and it might therefore be argued that this does not constitute acupuncture. Therefore we analysed the RA data separately using both the whole Park questionnaire and the subsets for pain and de qi defined by MacPherson and Asghar11.
When data from tables 2 and 3 are viewed together, it can be seen that the majority of patients tended to feel relatively few sensations (to varying intensities). The Park needling sensation questionnaire was, at the time, the only measure available. It should be noted, however, that this measure does have limitations insomuch as it was originally derived from a pain scale and might not adequately reflect the full range of sensations necessary to measure de qi. Consequently, the inclusion of descriptors that relate specifically to pain might ‘water down’ the data collected with this measure. MacPherson and Asghar11 also noted that this questionnaire contains a cluster of adjectives that related to neither pain nor de qi. This might also have the effect of introducing ‘background noise’ into the outcome. The Park instrument is rather a ‘blunt stick’ with a great deal of redundancy in it. Because of this, the maximum permitted score of 75 would almost never be attained in reality, and this has implications for the sensitivity of the measure. Lastly, patients were asked to rate their sensation only at the end of the course of treatment, which might have further decreased the sensitivity of this measure. Patients were urged to record the ‘average’ sensation they had felt over the entire course for all of the points used. Clearly, this has limitations as they were recording de qi retrospectively and also this might not reflect differences in de qi felt between points or between treatment sessions (some might elicit more de qi than others), all of which also have implications for the sensitivity of this measure. Other much more relevant and well-validated measures are now available9 to assess de qi and it is recommended that future studies should use these outcomes as well as recording de qi at each acupuncture consultation to reduce recall bias.
It might also be argued that recording of pain is subject to difficulties, which might have reduced sensitivity of this outcome. However, we felt that the use of daily recording for 1 week helped to ‘smooth over’ some of the natural fluctuations in pain symptoms. Furthermore, patients were asked to record the average amount of pain they had felt each day rather than peak pain and while this has the advantage of levelling off the waxing and waning of daily symptoms, it also has implications for the sensitivity of the measure and this is a limitation of the study.
There was not even a trend towards significance for either the intensity or presence of de qi in relation to pain outcome, which suggests that no such relationship exists between de qi and treatment outcome. These results must be viewed with caution, particularly in light of the limitations discussed above. Our data imply, however, that practitioners and researchers should be far less intent on obtaining what might be described as painful de qi when treating painful conditions. This would certainly increase patient comfort as manipulating an acupuncture needle in situ can be very painful. We would also recommend that a larger study is undertaken before acupuncture practitioners abandon the notion that the intensity or indeed presence of de qi is related to clinical improvement in pain. Such a study should have frequent recording of de qi, using a much more sensitive measure. It might also be prudent to record any de qi noted by the practitioners—that is, needle grasp. However, the data we have presented certainly suggest that the strength and presence of de qi is not related to treatment outcome when using acupuncture to treat pain.
▶. It is widely believed that acupuncture's effect depends on eliciting de qi sensation
▶. We conducted a study of acupuncture for osteoarthritic hip and knee (n=140)
▶. Pain relief did not correlate with de qi sensation
Funding PW and this study was funded by a Department of Health Postdoctoral Research Award. GL's post is partially funded by the Rufford Maurice Laing Foundation. The Southampton Complementary Medicine Research Trust contributed funding to help with the running of this trial.
Competing interests None.
Ethics approval This study was conducted with the approval of the Southampton and South West Hampshire and the Salisbury and South Wiltshire Research ethics Committees (approval number 170/03/t).
Provenance and peer review Not commissioned; externally peer reviewed.