The study was conducted at the Athinoula Martinos Center for Biomedical Imaging at Massachusetts General Hospital and Medical School. The list of sensations categorized as deqi
was based on the descriptors provided in TCM literature and reports by patients in clinical practice [1
] rather than descriptors based on questionnaires for pain studies. Sharp pain was regarded to result from inadvertent noxious stimulation rather than acupuncture deqi
, as evidenced by their distinct differences in hemodynamic response by fMRI [7
]. Statistical analysis including paired t tests, ANOVA, Spearman's correlation and Fisher' exact test was performed to demonstrate the prevalence of the sensations as well as the uniqueness of the deqi
The study was performed on 20–47 years old (29.0 +/- S.D 7.8), right-handed, acupuncture naïve healthy adult volunteers, 15 male and 27 female, 32 Caucasians, 6 Asians, 2 Hispanics and 2 Africans with informed consent, as approved by the Massachusetts General Hospital Subcommittee on Human Studies. The subjects were screened and excluded for major medical illnesses, history of head trauma, neuropsychiatric disorders, use of medications within one week, and contraindications for exposure to a high magnetic field. The sample size was determined by the minimum number of subjects necessary to detect activation/deactivation differences comparable to what had been observed in our previous fMRI studies, with 80% power.
The subjects were blinded to the procedures and could not see the sites undergoing stimulation from their supine position in the scanner. They were told that the acupuncture performed at different acupoints with different techniques would generate different needling sensations. Tactile (touch) stimulation was performed prior to acupuncture when the subjects were still naïve to acupuncture as a sensory comparison for the acupuncture stimulation. Thus, the comparison stimulation also took into account expectation and its placebo effects. Tactile stimulation and acupuncture were both performed at the same acupoint in 16 subjects each for LI4 and ST36, and 13 subjects for LV3. Three of the 42 subjects received tactile stimulation and acupuncture at all three acupoints; the remaining 39 subjects received acupuncture to all three acupoints, but only the paired tactile stimulation to the first of their acupoints. Analyses comparing tactile stimulation to acupuncture stimulation were performed on the paired sensory – acupuncture datasets. Data from all 3 acupoints for each subject was used in the Spearman's correlation of intensities of sensations in acupuncture.
Acupuncture and tactile stimulation control was delivered to LI4 on the hand, LV3 on the foot and ST36 on the lower leg on the right in randomized order by an acupuncturist with over 25 years of clinical experience (JL). The individual's sensitivity to needle manipulation was pretested, aiming to elicit deqi sensations without noxious pain. The stimulation paradigm is depicted in Figure . The needle was rotated approximately 180° in each direction with even motion at the rate of 60 times/min for 2 min during M1 and M2. The needle remained in place during the rest periods R1, R2 and R3. Each procedure lasted a total of ten minutes. In order to avoid excess discomfort, the subject was instructed to raise one finger if any sensation reached the intensity of 7–8 on a scale of 1–10 and 2 fingers in case of any sharp pain. When so signalled, the acupuncturist would adjust the force of stimulation so that the sharp pain would disappear within a few seconds. The acupuncture stimulation procedure was performed twice for each acupoint. Sterile, one-time use only stainless steel needles were used for LV3 (0.20 mm diameter) and ST36 (0.22 mm diameter) (KINGLI Medical Appliance Co. Wuxi, China). Silver needles (0.23 mm diameter) were used for LI4 (Matsuka, Tokyo, Japan). Superficial tactile stimulation was performed by gentle tapping with a size 5.88 von Frey monofilament, a standard method of sensory stimulation, prior to acupuncture with needling. The purpose of this design was to explore how acupuncture sensations might differ from the sensations elicited by the conventional sensory stimulus of touch. At the end of each tactile stimulation or acupuncture procedure, the subject was questioned by another researcher in the team if each of the deqi sensations (aching, pressure, soreness, heaviness, fullness, warmth, cooling, numbness, tingling, dull pain), sharp pain or any other sensations occurred during the stimulation, and to rate its intensity on the scale of 1–10 (1–3 mild, 4–6 moderate, 7–9 strong, 10 unbearable).
Figure 1 Experimental paradigm. Manual acupuncture was administered to LI4, LV3 and ST36 on the right. The subject's sensitivity to needling was pre-tested and adjusted to tolerance prior to scanning. After remaining in place for 2 min (R1), the needle was rotated (more ...)
The analyses were performed on each dataset from an acupuncture or tactile stimulation procedure. The paired t tests, ANOVA and Fisher's exact tests were performed on the average of the duplicate datasets for the acupuncture and tactile stimulation. Spearman's correlation for intensities of different sensations was performed on the individual datasets with acupuncture stimulation (41, 40, 41 datasets for LI4, ST36 LV3 respectively). The data were analysed both as continuous measures of a sensation, and also as a binary indication of its presence or absence. A sensation was determined to be present if the reported level, averaged over replicates, reached a minimal score of 1.
The datasets were divided into 3 groups 1) deqi
, 2) mixed (deqi
+ sharp pain), and 3) no sensations (neither deqi
nor sharp pain) according to the sensations recorded at the end of each experimental procedure. None of the participants experienced acute pain without deqi
. While dull pain was included as an important component of deqi
, sharp pain of different forms such as stabbing, burning or pricking, was regarded as inadvertent noxious stimulation, and the co-occurrence of sharp pain with deqi
was classified as a 'mixed response'. This scheme of categorization was based on the distinct differences in the hemodynamic response of the brain to these two categories of psychophysical response as evidenced by neuroimaging in prior studies. The pain neuromatrix was inhibited in deqi
but activated in the presence of sharp pain [7
Characterization of the deqi response
We characterized the sensory responses elicited by the stimulation paradigm through a number of statistical approaches as described below. The set of observations and analyses we present are organized in support of three objectives: a) characterization of the prevalence of sensations elicited by the acupuncture stimulation; b) characterization of the uniqueness of the sensations associated with the deqi experience; and c) exploration of characteristic of sensations that could be used as a 'deqi composite', a single-valued summary of the reported simultaneous sensations. Each of these objectives was explored using techniques as described below.
Frequency of overall response groups
Fisher's exact test was conducted on the data from individual acupoints and on the data pooled from all points. We first tested for differences in the overall response (i.e. deqi, mixed, or no sensations) in acupuncture versus tactile stimulation control. There is as yet no consensus in regard to the number of sensations or magnitude of response to define the overall sensory experience as a deqi response. We have set two thresholds based on the sum of the scores for all sensations: T = 1, a minimal experience of sensations; and T = 3, a more stringent requirement. The threshold selected for data analysis will depend on the purpose of the analysis.
Frequency of sensations
We compared the frequency of sensations that reached the thresholds within the same subjects, between an acupoint and the corresponding tactile stimulation control. In order to do this, we tabulated the presence or absence of a sensation for each of the 10 sensations separately, with acupuncture versus the corresponding tactile stimulation control for each acupoint.
Intensity of sensations
We compared the mean sensation intensity between acupuncture and tactile stimulation control separately for each acupoint using paired t-tests. Simple paired t-tests using all of the data would be unreliable, due to the many zero scores reported for a number of the sensations. We therefore performed the t-tests on mean sensations only for subjects who reported a sensation above the threshold of 1 in acupuncture regardless of its presence or absence in tactile stimulation control. Thus, the p-values are for comparison of the intensity for acupuncture vs. tactile stimulation, conditional on the subject experiencing measurable sensation during acupuncture.
Correlation between the intensities of sensations
Spearman's correlation was performed on normalized data of all acupuncture procedures to determine if a correlation existed between the intensities of the different sensations that occurred during acupuncture for each acupoint, conditioned that the sensation was experienced at the minimal total score of 1.
Specificity of individual sensations for acupuncture vs. tactile stimulation
This analysis tests the null hypothesis in statistics that specific individual sensations co-occur in acupuncture and tactile stimulation. Rejecting this hypothesis informs us that if a subject felt a sensation in acupuncture then they did not feel it in tactile stimulation or vice versa. This comparison was performed by Fisher's exact test, pooled and for each acupoint. The reason for doing the test both ways is that the use of data pooled from all acupoints assumes that any association is independent of acupoint, but we will present evidence below to suggest that the nature of deqi is probably different for different acupoints. Doing Fisher's exact test separately for each acupoint avoids the independence assumption, but this approach might suffer from lack of power due to the small counts for certain sensations.
Comparison of the frequency and the intensity by which acupuncture exceeds tactile stimulation control for each sensation between acupoints
We applied Fisher's exact test to all datasets to determine the differences in frequency between acupuncture and tactile stimulation control, for each sensation, between the acupoints. We applied ANOVA to test for the differences in intensity between acupuncture and tactile stimulation control, for each sensation, between the acupoints. The ANOVA analysis was applied to the thresholded data that only included subjects who experienced that sensation during acupuncture as well as to all datasets.
Deqi composite metric
Rank ordering of individual sensations
One difficulty in comparing combinations of sensations between groups and between experimental conditions is that a sensation profile is 10-dimensional. Hence a very large number of subjects would be required to adequately and reliably explore this high-dimensional space of sensation combinations. It is very desirable to reduce a sensation profile to a single number, a 'deqi composite', which can then be used to characterize and compare deqi using standard univariate statistical procedures. We explored reduction of the set of sensations to a single value as follows: (1) for each subject determine the mean difference in intensity between acupuncture and tactile stimulation control for each sensation/acupoint combination; (2) average these differences over subjects; and (3) normalize the averaged differences so that the sum over sensations equals 1. The sets of values thus defined will be referred to as 'deqi weights'.
'Deqi composite': differentiation of sensation and acupuncture stimulation
Given the deqi weights defined above, we can ask if using the observed patterns of sensations, the weighted index, 'deqi composite' calculated by a summation over all sensations of each sensation score multiplied by its weight provides additional evidence for differentiation between the tactile and acupuncture stimulation conditions. ANOVA was used to discriminate acupuncture from tactile stimulation and to test differences between the acupoints, using both the 'simple' average and the 'weighted' average of the deqi sensations.