bs_bs_banner
Pain Medicine 2014; 15: 910–920 Wiley Periodicals, Inc.
A Randomized Controlled Trial of Single Point Acupuncture in Primary Dysmenorrhea
Cun-Zhi Liu, MD, PhD,*† Jie-Ping Xie, MD,* Lin-Peng Wang, MD,† Yu-Qi Liu, MD, PhD,* Jia-Shan Song, MD,* Yin-Ying Chen, MD,* Guang-Xia Shi, MD,† Wei Zhou, MD,‡ Shu-Zhong Gao, MD, PhD,§ Shi-Liang Li, MD,¶ Jian-Min Xing, PhD,** Liang-Xiao Ma, MD, PhD,* Yan-Xia Wang, MD,†† Jiang Zhu, MD,* and Jian-Ping Liu, MD, PhD** *School of Acupuncture-Moxibustion and Tuina, Beijing University of Chinese Medicine, Beijing, China;
Potential financial conflicts of interest: None disclosed. Abstract Background. Acupuncture is often used for primary dysmenorrhea. But there is no convincing evidence due to low methodological quality. We aim to assess immediate effect of acupuncture at specific acupoint compared with unrelated acupoint and nonacupoint on primary dysmenorrhea.
†
Acupuncture and Moxibustion Department, Beijing Hospital of Traditional Chinese Medicine affiliated to Capital Medical University, Beijing, China; ‡
Acupuncture and Moxibustion Department, Huguosi Hospital of Traditional Chinese Medicine affiliated to Beijing University of Chinese Medicine, Beijing, China; §
College of Acupuncture and Massage, Shandong University of Traditional Chinese Medicine, Jinan, China; ¶
Acupuncture and Moxibustion Department, China-Japan Friendship Hospital, Beijing, China; **Center for Evidence-Based Chinese Medicine, Beijing University of Chinese Medicine, Beijing, China;
††
Acupuncture and Moxibustion Department, Dongzhimen Hospital affiliated to Beijing University of Chinese Medicine, Beijing, China
Reprint requests to: Jian-Ping Liu, MD, PhD, Center for Evidence-Based Chinese Medicine, Beijing University of Chinese Medicine, Beijing, China. Tel: 00861064286757; Fax: 00861064286760; E-mail: [email protected]. Jiang Zhu, MD, School of Acupuncture-Moxibustion and Tuina, Beijing University of Chinese Medicine, 11 Bei San Huan East Road, Chao Yang District, Beijing, China. Tel: 00861064287719; Fax: 00861084560055; E-mail: [email protected] Trial Registration: Controlled-Trials.com ISRCTN24863192. 910
Methods. The Acupuncture Analgesia Effect in Primary Dysmenorrhoea-II is a multicenter controlled trial conducted in six large hospitals of China. Patients who met inclusion criteria were randomly assigned to classic acupoint (N = 167), unrelated acupoint (N = 167), or non-acupoint (N = 167) group on a 1:1:1 basis. They received three sessions with electro-acupuncture at a classic acupoint (Sanyinjiao, SP6), or an unrelated acupoint (Xuanzhong, GB39), or nonacupoint location, respectively. The primary outcome was subjective pain as measured by a 100-mm visual analog scale (VAS). Measurements were obtained at 0, 5, 10, 30, and 60 minutes following the first intervention. In addition, patients scored changes of general complaints using Cox retrospective symptom scales (RSS-Cox) and 7-point verbal rating scale (VRS) during three menstrual cycles. Secondary outcomes included VAS score for average pain, pain total time, additional in-bed time, and proportion of participants using analgesics during three menstrual cycles. Findings. Five hundred and one people underwent random assignment. The primary comparison of VAS scores following the first intervention demonstrated that classic acupoint group was more effective both than unrelated acupoint (−4.0 mm, 95% CI −7.1 to −0.9, P = 0.010) and nonacupoint (−4.0 mm, 95% CI −7.0 to −0.9, P = 0.012) groups. However, no significant differences were detected among the three acupuncture groups for RSS-Cox or VRS outcomes. The per-protocol analysis showed similar pattern. No serious adverse events were noted. Conclusion. Specific acupoint acupuncture produced a statistically, but not clinically, significant effect compared with unrelated acupoint and
Single Point Acupuncture in Primary Dysmenorrhea nonacupoint acupuncture in primary dysmenorrhea patients. Future studies should focus on effects of multiple points acupuncture on primary dysmenorrhea. Key Words. Acupuncture; Management
Acute
Pain;
Pain
Introduction Primary dysmenorrhea refers to perceived cramping pain during menstruation in absence of an identifiable pathologic lesion. This menstrual pain could be associated with nausea, vomiting, diarrhea, and headache. It is highly prevalent and has been identified as a leading cause of recurrent short-term school or work absenteeism among female adolescents and young adults [1,2]. Current therapies for primary dysmenorrhea include nonsteroidal anti-inflammatory drugs, prostaglandin antagonists, and antispasmodic drugs. However, these drugs are only temporarily effective in relieving pain, and may induce side effects and unnecessary medical costs. Many patients may seek alternatives to conventional medicine [3]. Acupuncture is recommended as a safe treatment that is free of relevant side effects and has long been indicated for pain in China [4,5]. Unfortunately, there is insufficient convincing evidence for efficacy of acupuncture due to low methodological quality or small sample size. The natural course of disease was identified as an associated factor for results of controlled studies in chronic pain patients, but this influence could be minimized by evaluating immediate effects of a single intervention [6]. Sanyinjiao (SP6) acupoint is located on the inside of the ankle, and is the junction point of the liver, spleen, and kidney meridians. According to the theory of traditional Chinese medicine, SP6, point of the spleen meridians, is frequently used in the treatment of dysmenorrhea. We have demonstrated in our previous trial (Acupuncture Analgesia Effect in Primary Dysmenorrhoea-I [AAEPD-I]) that acupuncture was associated with significant reduction of menstrual pain severity compared with no acupuncture. However, the effects were similar to those observed with unrelated acupoint and nonacupoint acupuncture [7]. Therefore, we designed AAEPD-II with a larger sample size to investigate the immediate effects of acupuncture at a specific acupoint compared with unrelated acupoint and nonacupoint on primary dysmenorrhea patients.
of Chinese Medicine, China-Japan Friendship Hospital, the First Hospital affiliated to Tianjin University of Traditional Chinese Medicine, and the Hospital affiliated to Shandong University of Traditional Chinese Medicine) and was conducted between December 2008 and December 2009. We performed the study according to common guidelines for clinical trials (Declaration of Helsinki, International Conference on Harmonisation/WHO Good Clinical Practice standards including certification by an external audit). The Medical Ethical Committee at Dongzhimen Hospital Affiliated to Beijing University of Chinese Medicine approved the design on November 10, 2008 (Ref: ECP.J-BDY-200811-10). AAEPD-II was registered with ISRCTN24863192 at Current Controlled Trials (http://www.controlled -trials.com/ISRCTN24863192/ISRCTN24863192). Participants Participants were required to be between 15 and 30 years, with a history of regular menstrual cycles (28-day cycle ± 7 days) and primary dysmenorrhea (onset 80 mm. The VAS score was assessed by patients with the assistance of a blinded assessor. However, other form of therapy was not allowed during the study. Safety was monitored with attention paid to adverse effects.
Outcome Measures Demographic measures collected at the baseline evaluation included age, age at menarche, menstruation, and menstrual cycle. The primary outcome was subjective pain as measured by a 100-mm VAS ranging from 0 mm (no pain) to 100 mm (worst pain ever). Measurements were obtained at 0, 5, 10, 30, and 60 minutes during the second visit when acupuncture was administered. The baseline measurement was done before first treatment session; the 5, 10, and 30-minute measurements occurred during first treatment; and the 60-minute measurement occurred at 30 minutes
Single Point Acupuncture in Primary Dysmenorrhea after completion of first treatment. The Cox retrospective symptom scale (RSS-Cox) and 0- to 7-point verbal rating scale (VRS) were also employed as primary outcomes. The RSS-Cox, a menstrual symptom measure that has been shown to have high reliability, validity, and sensitivity [9], gives two scores: a Total Frequency Rating score (RSSCox 1) and an Average Severity Rating score (RSS-Cox 2). Lower scores indicate better health. The 0- to 7-point VRS, presented by Andersch and Milsom [10], defines menstrual pain according to the limitation of ability to work (unaffected = 0; rarely affected = 1; moderately affected = 2; clearly inhibited = 3), coexistence of systemic symptoms (absent = 0; present = 1), and need for analgesics (no = 1; rarely = 2; inefficacious = 3). RSS-Cox and VRS were assessed during three menstrual cycles (baseline cycle, intervention cycle, and postintervention cycle). The average VAS score for pain intensity during three menstrual cycles was the secondary outcome. When pain was recorded, patients were asked, “What was your average pain intensity over the last menstrual cycle?” The other secondary outcomes included the pain total time and additional in-bed time during three menstrual cycles. Assessments were obtained at the end of the baseline cycle, intervention cycle, and postintervention cycle. The secondary outcome measures reported in this manuscript are different from the protocol. The changes in assigned analgesic medication usage and proportion of participants using analgesics were used in our original protocol, which might have the similar meaning. Considering duplication, we chose the latter as the current secondary outcome. Assessments were obtained at the end of the intervention cycle and postintervention cycle. We would like to clarify that the baseline data were collected on the first day of the intervention cycle through talking face to face. Assessments were obtained at postintervention cycle by retrospective and phone. Sample Size Based on our design, an equation of repeated measures design was employed for sample size calculation. All the patients were measured two times before treatment (the average VAS score of baseline cycle and VAS score before the first intervention) and four times after treatment (VAS scores at 5, 10, 30, and 60 minutes following the first intervention) and P value was given 0.70. A sample size of 145 patients per group was calculated according to previous study [7] to yield two-sided 5% significance level and 80% power for detecting a mean difference of 3.2 mm with standard deviations (SDs) of 21.74 mm on the VAS scores between classic acupoint group and unrelated acupoint group from baseline to 30 minutes during the first intervention. Taken into account a dropout rate of 15%, 167 patients were needed in each group.
The main analysis was done including all randomly assigned patients. Because all randomized participants completed the first acupuncture treatment and had the VAS data from 0 to 60 minutes, the main analysis and per-protocol analysis include the same number of patients for VAS scores following the first intervention. Thus, the method of one-way analyses of variance (ANOVAs) with repeated measures was performed. Treatment-induced changes in the RSS-Cox and VRS scores among the three treatment groups at specific time points were analyzed by using generalized estimating equation models with the identity link for continuous outcomes. Patients with missing data were included in the generalized estimating equation model unless no primary outcomes after baseline were available. Generalized estimating equation models can accommodate missing data, for reasons unrelated to any variables in the model. In these models, we assumed an unstructured working correlation with the robust variance estimator to accommodate the correlated data. Additionally, the per-protocol analysis was done including only patients with no missing outcomes or nonadherence to the protocol (such as not meeting the inclusion criteria, receiving a proscribed co-intervention). Mauchly’s Test of Sphericity was applied to test if the data satisfied the repeated measures model assumptions. VAS scores, RSSCox scores, VRS score, the pain total time, or additional in-bed time was the dependent variable, time was the within-subject variable, and three intervention groups were the between-subject variables. When appropriate, post hoc comparisons were assessed using the least significant difference test (equal variances assumed) or Dunnett’s T3 test. The proportion of patients who reported at least 20% decrease in VAS scores for pain from baseline to 30 minutes during the first intervention, as well as the proportion of patients who used analgesics, was compared using chi-square test (two sided) for categorical variables. Resulting treatment effects are given together with 95% CI and corresponding P values as well as means and SD of the primary and secondary outcomes for each group. A level of P < 0.05 was accepted as significant. The generalized estimating equation models were run by using the SAS 9.1.3 (SAS Institute, Inc., Cary, NC, USA). The other analyses were performed with SPSS 17.0 (SPSS Inc, Chicago, IL, USA). Role of the Funding Source The Ministry of Science and Technology of China, the only source of funding, had no role in study design, data collection, analysis, interpretation, manuscript preparation, or decisions regarding publication. The needles we used were purchased from Tianjin HuaHong Medical Co., Ltd. Results
Statistical Analysis Patient Enrollment We used descriptive analysis to tabulate demographic and baseline characteristics of study participants by randomization group.
Enrollment of participants started in December 2008 and was completed in July 2009. Five hundred and one 913
Liu et al. people underwent random assignment and received the allocated intervention (Figure 2); of these, nine participants were erroneously included (four in classic acupoint group, two in unrelated acupoint group, and three in nonacupoint group) because of not meeting the inclusion criteria (four used analgesics in 24 hours before the first intervention, three had wrong diagnosis, and two had previously received acupuncture treatment). After randomization, eight participants in the three groups could not complete the study; six withdrew before the second acupuncture treatment, and two withdrew during postintervention cycle. Although one patient assigned to the classic acupoint group came only to the initial visit but did not complete the whole study because of private problem, one of the study hospitals erroneously registered for randomization. Therefore, 98% of randomly assigned patients completed treatment and the posttreatment
assessments (Figure 2). The main analysis included all 501 patients who were randomized. The per-protocol analysis excluded 16 patients who were erroneously enrolled or did not complete the study. Analysis of Baseline Data The demographic and clinical features at baseline are shown in Table 1. Most participants were unmarried women. On average, participants were 22.4 ± 2.8 years of age and had experienced pain for about 83.2 ± 40.7 months. The average baseline score on the VAS was 64.8 ± 16.8 mm for pain intensity. The classic acupoint group, the unrelated acupoint group, and the nonacupoint group were similar at baseline on demographic features, VAS scores, RSS-Cox scores, VRS score, pain total time, additional in-bed time, and patients using analgesics.
2500 (approximately) patients assessed for eligibility 1700 (approximately) not interested after receipt of information or had obvious violation of inclusion criteria 835 referred to physicians 334 excluded 174 did not meet inclusion criteria 96 declined to participate 64 other reasons 501 randomized 9 erroneously included 4 classic acupoint group* 2 unrelated acupoint group 3 nonacupoint group
167 allocated classic acupoint
163 completed study 4 discontinued
167 allocated unrelated acupoint
165 completed study 2 discontinued
167 allocated nonacupoint
165 completed study 2 discontinued
2 private problem *
1 acute gastroenteritis
1 reason unclear
1 unsatisfied
1 lost to follow up
1 lost to follow up
main analysis n = 167 per-protocol analysis n = 163
main analysis n = 167 per-protocol analysis n = 162
1 lost to follow up
main analysis n = 167 per-protocol analysis n = 160
914
Figure 2 Trial flow diagram. *Although 1 patient assigned to the classic acupoint group came only to the initial visit but did not complete the whole study because of private problem, one of the study hospitals erroneously registered for randomization.
Single Point Acupuncture in Primary Dysmenorrhea
Table 1 The baseline characteristics of patients in the three study groups, collected on the first day of the intervention cycle Variables
Classic Acupoint Group (N = 167)
Unrelated Acupoint Group (N = 167)
Nonacupoint Group (N = 167)
Age, years* Age at menarche, y* Duration of disease, m* Menstruation, d* Menstrual cycle, d* Average pain (VAS)* RSS-COX 1 score† RSS-COX 2 score* VRS score* Pain total time, min/d Additional in-bed time, min/d Patients using analgesics, n (%)
22.0 (20.0, 24.0) 13.0 (12.0, 14.0) 84.0 (60.0, 108.0) 5.0 (5.0, 6.0) 30.0 (28.0, 30.0) 66.0 (51.0, 78.0) 18.9 (7.7) 10.0 (7.0, 14.0) 3.0 (3.0, 4.0) 488.1 (340.9) 233.4 (246.1) 56 (33.5)
22.0 (20.0, 24.0) 13.0 (12.0, 14.0) 83.0 (48.0, 108.0) 5.5 (5.0, 6.5) 30.0 (28.0, 30.0) 65.0 (50.0, 80.0) 19.1 (8.0) 10.0 (7.0, 13.0) 3.0 (3.0, 4.0) 527.6 (336.5) 264.4 (271.1) 55 (32.9)
22.0 (20.0, 25.0) 13.0 (12.0, 14.0) 81.0 (49.0, 120.0) 5.0 (4.5, 6.5) 30.0 (28.0, 30.0) 66.0 (50.0, 80.0) 19.8 (7.8) 11.0 (7.0, 15.0) 3.0 (3.0, 4.0) 558.9 (371.5) 256.6 (266.1) 57 (34.1)
* Values are median (percentile 25, percentile 75); † Values are means (±standard deviations). VAS = visual analog scale; RSS = Cox retrospective symptom scale; RSS-COX 1 = frequency rating score of Cox retrospective symptom scale; RSS-COX 2 = severity rating score of Cox retrospective symptom scale; VRS = verbal rating scale; d = days; Menstruation = days in one menstrual cycle; Menstrual cycle = length between menstrual cycles; min = minutes; y = years; m = months.
At baseline, 94.0% of participants described their dysmenorrhea as moderate (80 mm > VAS ≥ 40 mm), and 6.0% described it as severe (VAS ≥ 80 mm). There were 59.7% who reported usually experiencing depression, 51.9% diarrhea, 29.9% nausea, and 7.6% vomiting in association with pain. Analysis of Outcome Variables Repeated-measures ANOVA at 0, 5, 10, 30, and 60 minutes following the first intervention (Table 2) revealed a
time effect (F = 283.0, P < 0.001) and a treatment effect (F = 4.3, P = 0.014), but no treatment × time interaction (F = 1.1, P = 0.39). The primary comparison of VAS scores demonstrated that the classic acupoint group was statistically significantly more effective both than the unrelated acupoint (mean difference = −4.0 mm, 95% CI −7.1 to −0.9, P = 0.010) and nonacupoint (mean difference = −4.0 mm, 95% CI −7.0 to −0.9, P = 0.012) groups (Figure 3), but there was no difference between the unrelated acupoint treatment and the nonacupoint treatment (mean difference = 0.1 mm, 95% CI −3.0 to 3.1, P = 0.97). Classic acupoint (167)
Figure 3 Development of the VAS scores at 0, 5, 10, 30, and 60 minutes following the start of the first intervention, in the three groups (the main analysis). Vertical bars represent standard deviations. VAS = visual analog scale.
Pain intensity on VAS score, mm
75 70
Unrelated acupoint (167)
65
Nonacupoint (167)
60 55 50 45 40 35 30 25 20 15 10 5 0
5
10
30
60
Time(Minutes)
915
916
0.97 0.1 (−3.0, 3.1)
167 41.6 (18.9)
167 31.5 (19.1)
167 36.1 (19.9)
40.6 (16.0)
31.6 (19.9)
37.1 (23.5)
167 37.4 (17.1) 167
167 27.5 (18.1) 167
167 29.5 (19.1) 167
Additionally, the per-protocol analysis showed similar results (Table 3). For comparison of the VAS scores for average pain during three menstrual cycles, there was a time effect (F = 168.2, P < 0.001), a treatment effect (F = 3.6, P = 0.03), and a treatment × time interaction (F = 2.6, P = 0.04). The classic acupoint group was statistically significantly more effective both than the unrelated acupoint (mean difference = −3.0 mm, 95% CI −6.0 to −0.01, P = 0.05) and nonacupoint (mean difference = −3.9 mm, 95% CI −6.9 to −0.9, P = 0.01) groups while there was no significant difference between the latter two groups (mean difference = −0.9 mm, 95% CI −3.9 to 2.1, P = 0.56). However, these between-group differences were not generally considered clinically relevant in pain conditions. Meanwhile, there were no significant differences among the three groups (P = 0.170) for the proportion of patients with at least a 20% reduction in VAS pain from baseline to 30 minutes during the first intervention. The proportions in classic acupoint group, unrelated acupoint group, and nonacupoint group were 86.2%, 79.0%, and 79.6%, respectively. There were no significant differences among the three groups (P = 0.170). In addition, no significant differences for the pain total time, additional in-bed time, or the proportion of participants using analgesics were observed among the three groups (Table 3). During three menstrual cycles, no significant differences were found on the number of participants who took an aspirin among the classic acupoint group, unrelated acupoint group, and nonacupoint group (85, 85, and 83, respectively). SD = standard deviation; VAS = visual analog scale.
167 45.8 (18.8) 45.9 (16.3) 167 43.0 (17.2) 167
−4.0 (−7.0, −0.9) 0.012 167 58.5 (14.6) −4.0 (−7.1, −0.9) 0.010 58.7 (13.5) 167 56.3 (12.0) 167
Mean Difference Mean Difference Mean Difference Mean (SD) (95%CI) P value (95%CI) P value (95% CI) P value N Mean (SD) Mean (SD) N N
Unrelated Acupoint vs Nonacupoint Classic Acupoint vs Nonacupoint Classic Acupoint vs Unrelated Acupoint Classic Acupoint Unrelated Acupoint Nonacupoint
The results derived from the generalized estimating equation models on frequency and severity rating scores of RSS-Cox during three menstrual cycles (Figure 4) revealed that RSS-Cox scores decreased in three groups over time (Ps < 0.001). However, between-group effects for theses measures did not reach statistical significance (P = 0.66, P = 0.54, respectively). Similar changes in the three groups were seen in VRS score (Figure 5, time effect (P < 0.001), treatment effect (P = 0.62), and treatment × time interaction (P = 0.52).
0 minute (before first treatment) 5 minutes (during first treatment) 10 minutes (during first treatment) 30 minutes (during first treatment) 60 minutes (30 minutes after completion of first treatment)
Table 2
The main analysis on VAS scores (mm) at 0, 5, 10, 30, and 60 minutes following the first intervention for each group
Liu et al.
Safety of Acupuncture During the acupuncture treatment, one patient in the classic acupoint group reported a small hematoma, one patient in the unrelated acupoint group reported faintness on acupuncture, and two patients in the nonacupoint group, respectively, reported small hematoma and needling pain after treatment. No serious adverse events were documented. Discussion Our study is, to date, one of the largest trials of the efficacy of single-point acupuncture available. The results showed that there was a significant improvement (21–26 mm on the VAS) over time in all three groups following the first intervention, which is a clinically relevant change, although
A
Classic acupoint
29 27 25 23 21 19 17 15 13 11 9 7 5
Unrelated acupoint
18
Nonacupoint
16
RRS COX2 score
RRS COX1 score
Single Point Acupuncture in Primary Dysmenorrhea
14 12 10 8 6 4
Cycle 0
Classic acupoint 167 Unrelated acupoint 167 Nonacupoint 167
Cycle 1
165 166 165
Cycle 2
B
163 165 165
2 Cycle 0
Cycle 1
167 167 167
165 166 165
Cycle 2
163 165 165
Figure 4 Development of the RSS-Cox scores during three menstrual cycles, in the three groups (the main analysis). A. RSS-Cox 1 score. B. RSS-Cox 2 score. Vertical bars represent standard deviations. RSS-Cox 1 = frequency rating score of Cox retrospective symptom scale 1; RSS-Cox 2 = severity rating score of Cox retrospective symptom scale 2. it is crucial to note that this is an uncontrolled observation, and therefore may be entirely nonspecific. This was probably due to the therapeutic effect of the acupuncture or the general effect of being in a study (Hawthorne effect). Patients with primary dysmenorrhea who received classic point acupuncture had significantly greater immediate pain relief after a single treatment than did patients who received unrelated acupoint or nonacupoint acupuncture. It indicated that for relieving the pain of dysmenorrhea, classic acupoint acupuncture was statistically significantly more effective than unrelated acupoint and nonacupoint acupuncture. However, a consensus report characterized
decreases in individuals’ pain intensity of approximately 10 mm, or 15% to 20% represent “minimal” or “little” change [11]. In this trial, there were no significant differences among the three groups in the proportion of patients with at least a 20% reduction in pain intensity from baseline to 30 minutes. The immediate analgesic effect of classic acupoint acupuncture seems to be below a clinically relevant pain improvement. In that case, though single acupoint acupuncture reduces the pain of dysmenorrhea, the effect is quite small and not clinically relevant. On the other hand, no significant differences were detected among the three acupuncture groups for
Classic acupoint
4.8 4.5
Unrelated acupoint
VRS score
4.2 3.9
Nonacupoint
3.6 3.3 3 2.7 2.4 2.1
Figure 5 Development of the VRS scores during three menstrual cycles, in the three groups (the main analysis). Vertical bars represent standard deviations. VRS = verbal rating scale.
1.8 1.5 1.2
Cycle 0
Classic acupoint Unrelated acupoint Nonacupoint
167 167 167
Cycle 1
165 166 165
Cycle 2
163 165 165
917
918 65.1 (17.8) 56.1 (16.0) 48.6 (24.5)
3.5 (1.3) 2.7 (0.8) 2.6 (1.2)
11.1 (5.4) 9.4 (4.5) 7.5 (4.2)
19.7 (7.8) 18.3 (7.7) 14.7 (7.8)
58.6 (14.5) 46.4 (18.6) 42.0 (18.9) 31.7 (19.0) 36.3 (20.0)
Mean (SD)
−3.0 (−6.0, −0.01)
0.79 0.77
0.10
0.010
0.46
0.28
0.28
0.005
−32.7 (−68.1, 2.6)
0.07
−38.7 (−91.6, 14.2) 0.15
−3.9 (−6.9, −0.9)
−0.1 (−0.2, 0.1)
−0.5 (−1.3, 0.4)
−0.8 (−2.2, 0.7)
−4.5 (−7.6, −1.3)
0.56
0.88
0.40
0.54
1.00
P value
−3.2 (−31.9, 38.4) 0.86
14.6 (−38.0, 67.3) 0.59
−0.9 (−3.9, 2.1)
−0.01 (−0.2, 0.2)
−0.4 (−1.2, 0.5)
−0.4 (−1.9, 1.0)
0.004 (−3.1, 3.1)
Mean Difference P value (95% CI)
Unrelated Acupoint vs Nonacupoint
SD = standard deviation; VAS = visual analog scale; RSS-COX 1 = frequency rating score of Cox retrospective symptom scale; RSS-COX 2 = severity rating score of Cox retrospective symptom scale; VRS = verbal rating scale.
9 (5.5) 22 (13.6)
162 162
Participants using analgesics during two menstrual cycles, n% Intervention cycle 160 8 (5.0) 163 Postintervention cycle 160 24 (15) 163 11 (6.7) 20 (12.3)
162 260.0 (267.6) −29.5 (−64.8, 5.8) 162 182.4 (198.8) 162 157.6 (180.4)
0.37
0.05
0.56
0.81
−0.1 (−1.0, 0.8)
0.005
0.64
−0.1 (−0.2, 0.1)
Classic Acupoint vs Nonacupoint Mean Difference P value (95% CI)
−0.3 (−1.8, 1.1)
−4.5 (−7.6, −1.3)
Mean Difference (95% CI)
Classic Acupoint vs Unrelated Acupoint
162 558.5 (376.5) −24.1 (−76.9, 28.7) 162 461.7 (321.2) 162 356.1 (309.2)
162 162 162
162 162 162
162 162 162
162 162 162
162 162 162 162 162
No
Nonacupoint
Additional in-bed time during three menstrual cycles, minutes/day Baseline cycle 160 232.5 (242.2) 163 265.0 (272.7) Intervention cycle 160 134.2 (191.0) 163 161.0 (204.3) Postintervention cycle 160 135.2 (162.2) 163 164.4 (195.5)
527.9 (339.4) 438.8 (304.2) 365.8 (281.7)
3.4 (1.3) 2.7 (0.9) 2.6 (1.1)
Pain total time during three menstrual cycles, minutes/day Baseline cycle 160 492.2 (342.5) 163 Intervention cycle 160 406.1 (275.8) 163 Postintervention cycle 160 362.0 (294.0) 163
163 163 163
VRS scores during three menstrual cycles Baseline cycle 160 3.4 (1.3) Intervention cycle 160 2.6 (0.9) Postintervention cycle 160 2.5 (1.2)
10.5 (5.3) 9.2 (4.8) 7.2 (4.3)
65.3 (15.9) 52.5 (15.5) 49.4 (23.1)
163 163 163
10.5 (5.1) 8.9 (4.8) 7.2 (4.4)
19.2 (8.1) 17.7 (8.2) 14.5 (8.0)
59.1 (13.7) 46.1 (16.2) 40.9 (15.9) 31.9 (20.0) 37.0 (23.2)
Average VAS scores during three menstrual cycles, mm Baseline cycle 160 64.4 (16.4) 163 Intervention cycle 160 49.5 (15.6) 163 Postintervention cycle 160 44.2 (22.6) 163
163 163 163
163 163 163 163 163
18.8 (7.7) 16.9 (7.7) 14.7 (8.0)
RSS scores during three menstrual cycles RSS-COX 1 scores Baseline cycle 160 Intervention cycle 160 Postintervention cycle 160 RSS-COX 2 scores Baseline cycle 160 Intervention cycle 160 Postintervention cycle 160
mm 56.1 (11.8) 42.7 (17.5) 37.1 (17.2) 27.2 (18.2) 29.6 (19.2)
Mean (SD)
No
No
Mean (SD)
Unrelated Acupoint
Classic Acupoint
The per-protocol analysis on primary and secondary outcomes for each group
VAS scores following the first intervention, 0 minute (before first treatment) 160 5 minutes (during first treatment) 160 10 minutes (during first treatment) 160 30 minutes (during first treatment) 160 60 minutes (30 minutes after 160 completion of first treatment)
Table 3
Liu et al.
Single Point Acupuncture in Primary Dysmenorrhea RSS-Cox or VRS outcomes. To some extent, this finding is consistent with a recent systematic review. It assessed the analgesic effect of acupuncture in 13 trials (3,025 patients) involving a variety of pain conditions [12], and found that there was a 4 mm on a 100 mm VAS difference between acupuncture and placebo acupuncture. In this study, patients were naive to acupuncture treatment, and treatments were quite similar, all three treatment locations distant from location of pain, both of which should facilitate blinding. Acupuncture is likely to have a nonspecific effect, but whether it has a more specific physiological effect remains unclear. Apart from acupoint specificity, issues such as needling sensation, psychological factors, and needle retaining time also have relevant influence on the effectiveness of acupuncture [13–15]. In the largest clinical acupuncture studies for back and knee pain to date [16,17]—the German Acupuncture Trials (GERAC) trials—acupuncture was found to be almost twice as effective as standard therapy over 6 months. However, there were no differences between traditional Chinese and sham acupuncture given to nonacupoints, suggesting that the benefits may have been a placebo effect. Functional studies have shown that there are differences between the neural effects of expectation and acupuncture’s therapeutic analgesia in experimental settings [18,19]. The present study did show that there was clinical benefit of obtaining acupuncture treatment, and this benefit may be due to both nonspecific effects (e.g., empathy from clinical staff, doing a positive action, belief in the intervention, etc.) and specific effects of the treatment. A general physiological effect of needling may be the reason why significant differences between sham and specific needling are not detected [20]. Our study has several limitations. The nonacupoint seemed to be very close to unrelated acupoint GB39. There is continuing uncertainty about the size and area of acupuncture points. The acupuncture point was considered as an area, about 2 cm in diameter [21]. Because of spread of the electrical current between the acupuncture points, use of electrical stimulation may make it even more difficult to isolate activation of an acupuncture point from other points or nonpoints of the limb. There is also the possibility of the E-stim at 2 hz resulting in diffuse noxious inhibitory control/pain inhibition independent of acupoint stimulation. Both of the mechanisms—inadvertant acupoint stimulation and diffuse noxious inhibitory control, the supposedly inert control—may have had physiologic effect. This may impact on choice of control condition for future studies. Our statistical analysis made assumptions about missing values, specifically that the values were missing completely at random. Although this assumption is difficult to formally test, we found no evidence that patient dropout was treatment related (Tables S1–4 and Figure S1). All subjects received the acupuncture treatment starting on the first day of intervention cycle for 3 consecutive days. Response at the beginning of the menstrual cycle may differ from the end of the cycle, as pain is likely to improve toward the end of the menstrual cycle. This may impact on the effect of acupuncture.
In summary, specific acupoint acupuncture produced a statistically, but not clinically, significant effect compared with unrelated acupoint and nonacupoint acupuncture in primary dysmenorrhea patients. In the current study, we examined the effects of single-point acupuncture, not the effects of multiple points acupuncture. Study on single point effects may have more theoretical significance to identify point specificity. However, in daily clinical practice, multiple points acupuncture may provide better pain relief than single puncture acupuncture. Future studies should focus on effects of multiple points acupuncture on primary dysmenorrhea. Acknowledgments We thank the research staff at Dongzhimen Hospital Affiliated to Beijing University of Chinese Medicine: Ji-Ping Zhao, Ling Tang; Beijing Hospital of Traditional Chinese Medicine affiliated to Capital Medical University: Ya-Li Wen, Jing-Dao Li; Huguosi Hospital of Traditional Chinese Medicine affiliated to Beijing University of Chinese Medicine: Zhi-Liang Li, Xu Chen; China-Japan Friendship Hospital: Min Li, Ying Wang; The First Hospital affiliated to Tianjin University of Traditional Chinese Medicine: JingXian Han, Meng-Meng Wu; Shandong University of Traditional Chinese Medicine: Yu-Xia Ma, Kun Lv; and Beijing University of Chinese Medicine: Yan-Fen She, Hui-Juan Cao. The study was funded by the National Basic Research Program of China (973 Program, reference number: 2012CB518506). Author Contributions Conceived and designed the experiments: Zhu J, Liu J-P, Liu C-Z, Xie J-P. Performed the experiments: Wang L-P, Zhou W, Gao S-Z, Li S-L, Wang Y-X. Analyzed the data: Song J-S, Chen Y-Y. Wrote the article Liu C-Z, Liu Y-Q, Shi G-X, Ma L-X. Biostatistics analysis: Xing J-M. Study supervision: Liu C-Z, Xie J-P. References 1 Patel V, Tanksale V, Sahasrabhojanee M, Gupte S, Nevrekar P. The burden and determinants of dysmenorrhoea: A population-based survey of 2,262 women in Goa, India. BJOG 2006;113:453– 63. 2 Dawood MY. Primary dysmenorrhea: Advances in pathogenesis and management. Obstet Gynecol 2006;108(2):428–41. 3 Durain D. Primary dysmenorrhea: Assessment and management update. J Midwifery Womens Health 2004;49(6):520–8. 4 Kaptchuk TJ. Acupuncture: Theory, efficacy, and practice. Ann Intern Med 2002;136(5):374–83. 5 Vincent C. The safety of acupuncture. BMJ 2001; 323(7311):467–8. 919
Liu et al. 6 Lim S. WHO standard acupuncture point locations. Evid Based Complement Alternat Med 2010; 7(2):167–8.
17 Manheimer E, Linde K, Lao L, Bouter LM, Berman BM. Meta-analysis: Acupuncture for osteoarthritis of the knee. Ann Intern Med 2007;146(12):868–77.
7 Liu CZ, Xie JP, Wang LP, et al. Immediate analgesia effect of single point acupuncture in primary dysmenorrhea: A randomized controlled trial. Pain Med 2011; 12(2):300–7.
18 Dhond RP, Kettner N, Napadow V. Do the neural correlates of acupuncture and placebo effects differ? Pain 2007;128(1–2):8–12.
8 Han JS. Acupuncture: Neuropeptide release produced by electrical stimulation of different frequencies. Trends Neurosci 2003;26(1):17–22, 6.
19 Cho ZH, Oleson TD, Alimi D, Niemtzow RC. Acupuncture: The search for biologic evidence with functional magnetic resonance imaging and positron emission tomographytechniques. J Altern Complement Med 2002;8(4):399–401.
9 Cox DJ, Meyer RG. Behavioral treatment parameters with primary dysmenorrhea. J Behav Med 1978; 1(3):297–310.
20 Pyne D, Shenker NG. Demystifying acupuncture. Rheumatology (Oxford) 2008;47(8):1132–6.
10 Andersch B, Milsom I. An epidemiologic study of young women with dysmenorrhea. Am J Obstet Gynecol 1982;144(6):655–60.
21 Jing YC. The measurement on the area of acupuncture point by noninvasive way. J Guiyang College Tradit Chin Med 1985;6(3):58.
11 Dworkin RH, Turk DC, Wyrwich KW, et al. Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. J Pain 2008;9(2):105–21. 12 Madsen MV, Gøtzsche PC, Hróbjartsson A. Acupuncture treatment for pain: Systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no classic acupoint groups. BMJ 2009;338:a3115. 13 Paterson C, Dieppe P. Characteristic and incidental (placebo) effects in complex interventions such as acupuncture. BMJ 2005;330(7501):1202–5. 14 Moore A, McQuay H. Acupuncture: Not just needles? Lancet 2005;366(9480):100–1. 9–15. 15 White AR, Filshie J, Cummings TM. Clinical trials of acupuncture: consensus recommendations for optimal treatment, sham controls and blinding. Complement Ther Med 2001;9(4):237–45. 16 Manheimer E, White A, Berman B, Forys K, Ernst E. Meta-analysis: Acupuncture for low back pain. Ann Intern Med 2005;142(8):651–63.
920
Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Figure S1 The flow diagram of dropout patients in the study. Table S1 The reasons for dropout patients with missing data. Table S2 The baseline characteristics of patients with missing data, collected on the first day of the intervention cycle. Table S3 The primary outcomes of patients with missing data, collected during the intervention and postintervention cycles. Table S4 The secondary outcomes of patients with missing data, collected during the intervention and postintervention cycles.
Pain Medicine 2014; 15: 910–920 Wiley Periodicals, Inc.
A Randomized Controlled Trial of Single Point Acupuncture in Primary Dysmenorrhea
Cun-Zhi Liu, MD, PhD,*† Jie-Ping Xie, MD,* Lin-Peng Wang, MD,† Yu-Qi Liu, MD, PhD,* Jia-Shan Song, MD,* Yin-Ying Chen, MD,* Guang-Xia Shi, MD,† Wei Zhou, MD,‡ Shu-Zhong Gao, MD, PhD,§ Shi-Liang Li, MD,¶ Jian-Min Xing, PhD,** Liang-Xiao Ma, MD, PhD,* Yan-Xia Wang, MD,†† Jiang Zhu, MD,* and Jian-Ping Liu, MD, PhD** *School of Acupuncture-Moxibustion and Tuina, Beijing University of Chinese Medicine, Beijing, China;
Potential financial conflicts of interest: None disclosed. Abstract Background. Acupuncture is often used for primary dysmenorrhea. But there is no convincing evidence due to low methodological quality. We aim to assess immediate effect of acupuncture at specific acupoint compared with unrelated acupoint and nonacupoint on primary dysmenorrhea.
†
Acupuncture and Moxibustion Department, Beijing Hospital of Traditional Chinese Medicine affiliated to Capital Medical University, Beijing, China; ‡
Acupuncture and Moxibustion Department, Huguosi Hospital of Traditional Chinese Medicine affiliated to Beijing University of Chinese Medicine, Beijing, China; §
College of Acupuncture and Massage, Shandong University of Traditional Chinese Medicine, Jinan, China; ¶
Acupuncture and Moxibustion Department, China-Japan Friendship Hospital, Beijing, China; **Center for Evidence-Based Chinese Medicine, Beijing University of Chinese Medicine, Beijing, China;
††
Acupuncture and Moxibustion Department, Dongzhimen Hospital affiliated to Beijing University of Chinese Medicine, Beijing, China
Reprint requests to: Jian-Ping Liu, MD, PhD, Center for Evidence-Based Chinese Medicine, Beijing University of Chinese Medicine, Beijing, China. Tel: 00861064286757; Fax: 00861064286760; E-mail: [email protected]. Jiang Zhu, MD, School of Acupuncture-Moxibustion and Tuina, Beijing University of Chinese Medicine, 11 Bei San Huan East Road, Chao Yang District, Beijing, China. Tel: 00861064287719; Fax: 00861084560055; E-mail: [email protected] Trial Registration: Controlled-Trials.com ISRCTN24863192. 910
Methods. The Acupuncture Analgesia Effect in Primary Dysmenorrhoea-II is a multicenter controlled trial conducted in six large hospitals of China. Patients who met inclusion criteria were randomly assigned to classic acupoint (N = 167), unrelated acupoint (N = 167), or non-acupoint (N = 167) group on a 1:1:1 basis. They received three sessions with electro-acupuncture at a classic acupoint (Sanyinjiao, SP6), or an unrelated acupoint (Xuanzhong, GB39), or nonacupoint location, respectively. The primary outcome was subjective pain as measured by a 100-mm visual analog scale (VAS). Measurements were obtained at 0, 5, 10, 30, and 60 minutes following the first intervention. In addition, patients scored changes of general complaints using Cox retrospective symptom scales (RSS-Cox) and 7-point verbal rating scale (VRS) during three menstrual cycles. Secondary outcomes included VAS score for average pain, pain total time, additional in-bed time, and proportion of participants using analgesics during three menstrual cycles. Findings. Five hundred and one people underwent random assignment. The primary comparison of VAS scores following the first intervention demonstrated that classic acupoint group was more effective both than unrelated acupoint (−4.0 mm, 95% CI −7.1 to −0.9, P = 0.010) and nonacupoint (−4.0 mm, 95% CI −7.0 to −0.9, P = 0.012) groups. However, no significant differences were detected among the three acupuncture groups for RSS-Cox or VRS outcomes. The per-protocol analysis showed similar pattern. No serious adverse events were noted. Conclusion. Specific acupoint acupuncture produced a statistically, but not clinically, significant effect compared with unrelated acupoint and
Single Point Acupuncture in Primary Dysmenorrhea nonacupoint acupuncture in primary dysmenorrhea patients. Future studies should focus on effects of multiple points acupuncture on primary dysmenorrhea. Key Words. Acupuncture; Management
Acute
Pain;
Pain
Introduction Primary dysmenorrhea refers to perceived cramping pain during menstruation in absence of an identifiable pathologic lesion. This menstrual pain could be associated with nausea, vomiting, diarrhea, and headache. It is highly prevalent and has been identified as a leading cause of recurrent short-term school or work absenteeism among female adolescents and young adults [1,2]. Current therapies for primary dysmenorrhea include nonsteroidal anti-inflammatory drugs, prostaglandin antagonists, and antispasmodic drugs. However, these drugs are only temporarily effective in relieving pain, and may induce side effects and unnecessary medical costs. Many patients may seek alternatives to conventional medicine [3]. Acupuncture is recommended as a safe treatment that is free of relevant side effects and has long been indicated for pain in China [4,5]. Unfortunately, there is insufficient convincing evidence for efficacy of acupuncture due to low methodological quality or small sample size. The natural course of disease was identified as an associated factor for results of controlled studies in chronic pain patients, but this influence could be minimized by evaluating immediate effects of a single intervention [6]. Sanyinjiao (SP6) acupoint is located on the inside of the ankle, and is the junction point of the liver, spleen, and kidney meridians. According to the theory of traditional Chinese medicine, SP6, point of the spleen meridians, is frequently used in the treatment of dysmenorrhea. We have demonstrated in our previous trial (Acupuncture Analgesia Effect in Primary Dysmenorrhoea-I [AAEPD-I]) that acupuncture was associated with significant reduction of menstrual pain severity compared with no acupuncture. However, the effects were similar to those observed with unrelated acupoint and nonacupoint acupuncture [7]. Therefore, we designed AAEPD-II with a larger sample size to investigate the immediate effects of acupuncture at a specific acupoint compared with unrelated acupoint and nonacupoint on primary dysmenorrhea patients.
of Chinese Medicine, China-Japan Friendship Hospital, the First Hospital affiliated to Tianjin University of Traditional Chinese Medicine, and the Hospital affiliated to Shandong University of Traditional Chinese Medicine) and was conducted between December 2008 and December 2009. We performed the study according to common guidelines for clinical trials (Declaration of Helsinki, International Conference on Harmonisation/WHO Good Clinical Practice standards including certification by an external audit). The Medical Ethical Committee at Dongzhimen Hospital Affiliated to Beijing University of Chinese Medicine approved the design on November 10, 2008 (Ref: ECP.J-BDY-200811-10). AAEPD-II was registered with ISRCTN24863192 at Current Controlled Trials (http://www.controlled -trials.com/ISRCTN24863192/ISRCTN24863192). Participants Participants were required to be between 15 and 30 years, with a history of regular menstrual cycles (28-day cycle ± 7 days) and primary dysmenorrhea (onset 80 mm. The VAS score was assessed by patients with the assistance of a blinded assessor. However, other form of therapy was not allowed during the study. Safety was monitored with attention paid to adverse effects.
Outcome Measures Demographic measures collected at the baseline evaluation included age, age at menarche, menstruation, and menstrual cycle. The primary outcome was subjective pain as measured by a 100-mm VAS ranging from 0 mm (no pain) to 100 mm (worst pain ever). Measurements were obtained at 0, 5, 10, 30, and 60 minutes during the second visit when acupuncture was administered. The baseline measurement was done before first treatment session; the 5, 10, and 30-minute measurements occurred during first treatment; and the 60-minute measurement occurred at 30 minutes
Single Point Acupuncture in Primary Dysmenorrhea after completion of first treatment. The Cox retrospective symptom scale (RSS-Cox) and 0- to 7-point verbal rating scale (VRS) were also employed as primary outcomes. The RSS-Cox, a menstrual symptom measure that has been shown to have high reliability, validity, and sensitivity [9], gives two scores: a Total Frequency Rating score (RSSCox 1) and an Average Severity Rating score (RSS-Cox 2). Lower scores indicate better health. The 0- to 7-point VRS, presented by Andersch and Milsom [10], defines menstrual pain according to the limitation of ability to work (unaffected = 0; rarely affected = 1; moderately affected = 2; clearly inhibited = 3), coexistence of systemic symptoms (absent = 0; present = 1), and need for analgesics (no = 1; rarely = 2; inefficacious = 3). RSS-Cox and VRS were assessed during three menstrual cycles (baseline cycle, intervention cycle, and postintervention cycle). The average VAS score for pain intensity during three menstrual cycles was the secondary outcome. When pain was recorded, patients were asked, “What was your average pain intensity over the last menstrual cycle?” The other secondary outcomes included the pain total time and additional in-bed time during three menstrual cycles. Assessments were obtained at the end of the baseline cycle, intervention cycle, and postintervention cycle. The secondary outcome measures reported in this manuscript are different from the protocol. The changes in assigned analgesic medication usage and proportion of participants using analgesics were used in our original protocol, which might have the similar meaning. Considering duplication, we chose the latter as the current secondary outcome. Assessments were obtained at the end of the intervention cycle and postintervention cycle. We would like to clarify that the baseline data were collected on the first day of the intervention cycle through talking face to face. Assessments were obtained at postintervention cycle by retrospective and phone. Sample Size Based on our design, an equation of repeated measures design was employed for sample size calculation. All the patients were measured two times before treatment (the average VAS score of baseline cycle and VAS score before the first intervention) and four times after treatment (VAS scores at 5, 10, 30, and 60 minutes following the first intervention) and P value was given 0.70. A sample size of 145 patients per group was calculated according to previous study [7] to yield two-sided 5% significance level and 80% power for detecting a mean difference of 3.2 mm with standard deviations (SDs) of 21.74 mm on the VAS scores between classic acupoint group and unrelated acupoint group from baseline to 30 minutes during the first intervention. Taken into account a dropout rate of 15%, 167 patients were needed in each group.
The main analysis was done including all randomly assigned patients. Because all randomized participants completed the first acupuncture treatment and had the VAS data from 0 to 60 minutes, the main analysis and per-protocol analysis include the same number of patients for VAS scores following the first intervention. Thus, the method of one-way analyses of variance (ANOVAs) with repeated measures was performed. Treatment-induced changes in the RSS-Cox and VRS scores among the three treatment groups at specific time points were analyzed by using generalized estimating equation models with the identity link for continuous outcomes. Patients with missing data were included in the generalized estimating equation model unless no primary outcomes after baseline were available. Generalized estimating equation models can accommodate missing data, for reasons unrelated to any variables in the model. In these models, we assumed an unstructured working correlation with the robust variance estimator to accommodate the correlated data. Additionally, the per-protocol analysis was done including only patients with no missing outcomes or nonadherence to the protocol (such as not meeting the inclusion criteria, receiving a proscribed co-intervention). Mauchly’s Test of Sphericity was applied to test if the data satisfied the repeated measures model assumptions. VAS scores, RSSCox scores, VRS score, the pain total time, or additional in-bed time was the dependent variable, time was the within-subject variable, and three intervention groups were the between-subject variables. When appropriate, post hoc comparisons were assessed using the least significant difference test (equal variances assumed) or Dunnett’s T3 test. The proportion of patients who reported at least 20% decrease in VAS scores for pain from baseline to 30 minutes during the first intervention, as well as the proportion of patients who used analgesics, was compared using chi-square test (two sided) for categorical variables. Resulting treatment effects are given together with 95% CI and corresponding P values as well as means and SD of the primary and secondary outcomes for each group. A level of P < 0.05 was accepted as significant. The generalized estimating equation models were run by using the SAS 9.1.3 (SAS Institute, Inc., Cary, NC, USA). The other analyses were performed with SPSS 17.0 (SPSS Inc, Chicago, IL, USA). Role of the Funding Source The Ministry of Science and Technology of China, the only source of funding, had no role in study design, data collection, analysis, interpretation, manuscript preparation, or decisions regarding publication. The needles we used were purchased from Tianjin HuaHong Medical Co., Ltd. Results
Statistical Analysis Patient Enrollment We used descriptive analysis to tabulate demographic and baseline characteristics of study participants by randomization group.
Enrollment of participants started in December 2008 and was completed in July 2009. Five hundred and one 913
Liu et al. people underwent random assignment and received the allocated intervention (Figure 2); of these, nine participants were erroneously included (four in classic acupoint group, two in unrelated acupoint group, and three in nonacupoint group) because of not meeting the inclusion criteria (four used analgesics in 24 hours before the first intervention, three had wrong diagnosis, and two had previously received acupuncture treatment). After randomization, eight participants in the three groups could not complete the study; six withdrew before the second acupuncture treatment, and two withdrew during postintervention cycle. Although one patient assigned to the classic acupoint group came only to the initial visit but did not complete the whole study because of private problem, one of the study hospitals erroneously registered for randomization. Therefore, 98% of randomly assigned patients completed treatment and the posttreatment
assessments (Figure 2). The main analysis included all 501 patients who were randomized. The per-protocol analysis excluded 16 patients who were erroneously enrolled or did not complete the study. Analysis of Baseline Data The demographic and clinical features at baseline are shown in Table 1. Most participants were unmarried women. On average, participants were 22.4 ± 2.8 years of age and had experienced pain for about 83.2 ± 40.7 months. The average baseline score on the VAS was 64.8 ± 16.8 mm for pain intensity. The classic acupoint group, the unrelated acupoint group, and the nonacupoint group were similar at baseline on demographic features, VAS scores, RSS-Cox scores, VRS score, pain total time, additional in-bed time, and patients using analgesics.
2500 (approximately) patients assessed for eligibility 1700 (approximately) not interested after receipt of information or had obvious violation of inclusion criteria 835 referred to physicians 334 excluded 174 did not meet inclusion criteria 96 declined to participate 64 other reasons 501 randomized 9 erroneously included 4 classic acupoint group* 2 unrelated acupoint group 3 nonacupoint group
167 allocated classic acupoint
163 completed study 4 discontinued
167 allocated unrelated acupoint
165 completed study 2 discontinued
167 allocated nonacupoint
165 completed study 2 discontinued
2 private problem *
1 acute gastroenteritis
1 reason unclear
1 unsatisfied
1 lost to follow up
1 lost to follow up
main analysis n = 167 per-protocol analysis n = 163
main analysis n = 167 per-protocol analysis n = 162
1 lost to follow up
main analysis n = 167 per-protocol analysis n = 160
914
Figure 2 Trial flow diagram. *Although 1 patient assigned to the classic acupoint group came only to the initial visit but did not complete the whole study because of private problem, one of the study hospitals erroneously registered for randomization.
Single Point Acupuncture in Primary Dysmenorrhea
Table 1 The baseline characteristics of patients in the three study groups, collected on the first day of the intervention cycle Variables
Classic Acupoint Group (N = 167)
Unrelated Acupoint Group (N = 167)
Nonacupoint Group (N = 167)
Age, years* Age at menarche, y* Duration of disease, m* Menstruation, d* Menstrual cycle, d* Average pain (VAS)* RSS-COX 1 score† RSS-COX 2 score* VRS score* Pain total time, min/d Additional in-bed time, min/d Patients using analgesics, n (%)
22.0 (20.0, 24.0) 13.0 (12.0, 14.0) 84.0 (60.0, 108.0) 5.0 (5.0, 6.0) 30.0 (28.0, 30.0) 66.0 (51.0, 78.0) 18.9 (7.7) 10.0 (7.0, 14.0) 3.0 (3.0, 4.0) 488.1 (340.9) 233.4 (246.1) 56 (33.5)
22.0 (20.0, 24.0) 13.0 (12.0, 14.0) 83.0 (48.0, 108.0) 5.5 (5.0, 6.5) 30.0 (28.0, 30.0) 65.0 (50.0, 80.0) 19.1 (8.0) 10.0 (7.0, 13.0) 3.0 (3.0, 4.0) 527.6 (336.5) 264.4 (271.1) 55 (32.9)
22.0 (20.0, 25.0) 13.0 (12.0, 14.0) 81.0 (49.0, 120.0) 5.0 (4.5, 6.5) 30.0 (28.0, 30.0) 66.0 (50.0, 80.0) 19.8 (7.8) 11.0 (7.0, 15.0) 3.0 (3.0, 4.0) 558.9 (371.5) 256.6 (266.1) 57 (34.1)
* Values are median (percentile 25, percentile 75); † Values are means (±standard deviations). VAS = visual analog scale; RSS = Cox retrospective symptom scale; RSS-COX 1 = frequency rating score of Cox retrospective symptom scale; RSS-COX 2 = severity rating score of Cox retrospective symptom scale; VRS = verbal rating scale; d = days; Menstruation = days in one menstrual cycle; Menstrual cycle = length between menstrual cycles; min = minutes; y = years; m = months.
At baseline, 94.0% of participants described their dysmenorrhea as moderate (80 mm > VAS ≥ 40 mm), and 6.0% described it as severe (VAS ≥ 80 mm). There were 59.7% who reported usually experiencing depression, 51.9% diarrhea, 29.9% nausea, and 7.6% vomiting in association with pain. Analysis of Outcome Variables Repeated-measures ANOVA at 0, 5, 10, 30, and 60 minutes following the first intervention (Table 2) revealed a
time effect (F = 283.0, P < 0.001) and a treatment effect (F = 4.3, P = 0.014), but no treatment × time interaction (F = 1.1, P = 0.39). The primary comparison of VAS scores demonstrated that the classic acupoint group was statistically significantly more effective both than the unrelated acupoint (mean difference = −4.0 mm, 95% CI −7.1 to −0.9, P = 0.010) and nonacupoint (mean difference = −4.0 mm, 95% CI −7.0 to −0.9, P = 0.012) groups (Figure 3), but there was no difference between the unrelated acupoint treatment and the nonacupoint treatment (mean difference = 0.1 mm, 95% CI −3.0 to 3.1, P = 0.97). Classic acupoint (167)
Figure 3 Development of the VAS scores at 0, 5, 10, 30, and 60 minutes following the start of the first intervention, in the three groups (the main analysis). Vertical bars represent standard deviations. VAS = visual analog scale.
Pain intensity on VAS score, mm
75 70
Unrelated acupoint (167)
65
Nonacupoint (167)
60 55 50 45 40 35 30 25 20 15 10 5 0
5
10
30
60
Time(Minutes)
915
916
0.97 0.1 (−3.0, 3.1)
167 41.6 (18.9)
167 31.5 (19.1)
167 36.1 (19.9)
40.6 (16.0)
31.6 (19.9)
37.1 (23.5)
167 37.4 (17.1) 167
167 27.5 (18.1) 167
167 29.5 (19.1) 167
Additionally, the per-protocol analysis showed similar results (Table 3). For comparison of the VAS scores for average pain during three menstrual cycles, there was a time effect (F = 168.2, P < 0.001), a treatment effect (F = 3.6, P = 0.03), and a treatment × time interaction (F = 2.6, P = 0.04). The classic acupoint group was statistically significantly more effective both than the unrelated acupoint (mean difference = −3.0 mm, 95% CI −6.0 to −0.01, P = 0.05) and nonacupoint (mean difference = −3.9 mm, 95% CI −6.9 to −0.9, P = 0.01) groups while there was no significant difference between the latter two groups (mean difference = −0.9 mm, 95% CI −3.9 to 2.1, P = 0.56). However, these between-group differences were not generally considered clinically relevant in pain conditions. Meanwhile, there were no significant differences among the three groups (P = 0.170) for the proportion of patients with at least a 20% reduction in VAS pain from baseline to 30 minutes during the first intervention. The proportions in classic acupoint group, unrelated acupoint group, and nonacupoint group were 86.2%, 79.0%, and 79.6%, respectively. There were no significant differences among the three groups (P = 0.170). In addition, no significant differences for the pain total time, additional in-bed time, or the proportion of participants using analgesics were observed among the three groups (Table 3). During three menstrual cycles, no significant differences were found on the number of participants who took an aspirin among the classic acupoint group, unrelated acupoint group, and nonacupoint group (85, 85, and 83, respectively). SD = standard deviation; VAS = visual analog scale.
167 45.8 (18.8) 45.9 (16.3) 167 43.0 (17.2) 167
−4.0 (−7.0, −0.9) 0.012 167 58.5 (14.6) −4.0 (−7.1, −0.9) 0.010 58.7 (13.5) 167 56.3 (12.0) 167
Mean Difference Mean Difference Mean Difference Mean (SD) (95%CI) P value (95%CI) P value (95% CI) P value N Mean (SD) Mean (SD) N N
Unrelated Acupoint vs Nonacupoint Classic Acupoint vs Nonacupoint Classic Acupoint vs Unrelated Acupoint Classic Acupoint Unrelated Acupoint Nonacupoint
The results derived from the generalized estimating equation models on frequency and severity rating scores of RSS-Cox during three menstrual cycles (Figure 4) revealed that RSS-Cox scores decreased in three groups over time (Ps < 0.001). However, between-group effects for theses measures did not reach statistical significance (P = 0.66, P = 0.54, respectively). Similar changes in the three groups were seen in VRS score (Figure 5, time effect (P < 0.001), treatment effect (P = 0.62), and treatment × time interaction (P = 0.52).
0 minute (before first treatment) 5 minutes (during first treatment) 10 minutes (during first treatment) 30 minutes (during first treatment) 60 minutes (30 minutes after completion of first treatment)
Table 2
The main analysis on VAS scores (mm) at 0, 5, 10, 30, and 60 minutes following the first intervention for each group
Liu et al.
Safety of Acupuncture During the acupuncture treatment, one patient in the classic acupoint group reported a small hematoma, one patient in the unrelated acupoint group reported faintness on acupuncture, and two patients in the nonacupoint group, respectively, reported small hematoma and needling pain after treatment. No serious adverse events were documented. Discussion Our study is, to date, one of the largest trials of the efficacy of single-point acupuncture available. The results showed that there was a significant improvement (21–26 mm on the VAS) over time in all three groups following the first intervention, which is a clinically relevant change, although
A
Classic acupoint
29 27 25 23 21 19 17 15 13 11 9 7 5
Unrelated acupoint
18
Nonacupoint
16
RRS COX2 score
RRS COX1 score
Single Point Acupuncture in Primary Dysmenorrhea
14 12 10 8 6 4
Cycle 0
Classic acupoint 167 Unrelated acupoint 167 Nonacupoint 167
Cycle 1
165 166 165
Cycle 2
B
163 165 165
2 Cycle 0
Cycle 1
167 167 167
165 166 165
Cycle 2
163 165 165
Figure 4 Development of the RSS-Cox scores during three menstrual cycles, in the three groups (the main analysis). A. RSS-Cox 1 score. B. RSS-Cox 2 score. Vertical bars represent standard deviations. RSS-Cox 1 = frequency rating score of Cox retrospective symptom scale 1; RSS-Cox 2 = severity rating score of Cox retrospective symptom scale 2. it is crucial to note that this is an uncontrolled observation, and therefore may be entirely nonspecific. This was probably due to the therapeutic effect of the acupuncture or the general effect of being in a study (Hawthorne effect). Patients with primary dysmenorrhea who received classic point acupuncture had significantly greater immediate pain relief after a single treatment than did patients who received unrelated acupoint or nonacupoint acupuncture. It indicated that for relieving the pain of dysmenorrhea, classic acupoint acupuncture was statistically significantly more effective than unrelated acupoint and nonacupoint acupuncture. However, a consensus report characterized
decreases in individuals’ pain intensity of approximately 10 mm, or 15% to 20% represent “minimal” or “little” change [11]. In this trial, there were no significant differences among the three groups in the proportion of patients with at least a 20% reduction in pain intensity from baseline to 30 minutes. The immediate analgesic effect of classic acupoint acupuncture seems to be below a clinically relevant pain improvement. In that case, though single acupoint acupuncture reduces the pain of dysmenorrhea, the effect is quite small and not clinically relevant. On the other hand, no significant differences were detected among the three acupuncture groups for
Classic acupoint
4.8 4.5
Unrelated acupoint
VRS score
4.2 3.9
Nonacupoint
3.6 3.3 3 2.7 2.4 2.1
Figure 5 Development of the VRS scores during three menstrual cycles, in the three groups (the main analysis). Vertical bars represent standard deviations. VRS = verbal rating scale.
1.8 1.5 1.2
Cycle 0
Classic acupoint Unrelated acupoint Nonacupoint
167 167 167
Cycle 1
165 166 165
Cycle 2
163 165 165
917
918 65.1 (17.8) 56.1 (16.0) 48.6 (24.5)
3.5 (1.3) 2.7 (0.8) 2.6 (1.2)
11.1 (5.4) 9.4 (4.5) 7.5 (4.2)
19.7 (7.8) 18.3 (7.7) 14.7 (7.8)
58.6 (14.5) 46.4 (18.6) 42.0 (18.9) 31.7 (19.0) 36.3 (20.0)
Mean (SD)
−3.0 (−6.0, −0.01)
0.79 0.77
0.10
0.010
0.46
0.28
0.28
0.005
−32.7 (−68.1, 2.6)
0.07
−38.7 (−91.6, 14.2) 0.15
−3.9 (−6.9, −0.9)
−0.1 (−0.2, 0.1)
−0.5 (−1.3, 0.4)
−0.8 (−2.2, 0.7)
−4.5 (−7.6, −1.3)
0.56
0.88
0.40
0.54
1.00
P value
−3.2 (−31.9, 38.4) 0.86
14.6 (−38.0, 67.3) 0.59
−0.9 (−3.9, 2.1)
−0.01 (−0.2, 0.2)
−0.4 (−1.2, 0.5)
−0.4 (−1.9, 1.0)
0.004 (−3.1, 3.1)
Mean Difference P value (95% CI)
Unrelated Acupoint vs Nonacupoint
SD = standard deviation; VAS = visual analog scale; RSS-COX 1 = frequency rating score of Cox retrospective symptom scale; RSS-COX 2 = severity rating score of Cox retrospective symptom scale; VRS = verbal rating scale.
9 (5.5) 22 (13.6)
162 162
Participants using analgesics during two menstrual cycles, n% Intervention cycle 160 8 (5.0) 163 Postintervention cycle 160 24 (15) 163 11 (6.7) 20 (12.3)
162 260.0 (267.6) −29.5 (−64.8, 5.8) 162 182.4 (198.8) 162 157.6 (180.4)
0.37
0.05
0.56
0.81
−0.1 (−1.0, 0.8)
0.005
0.64
−0.1 (−0.2, 0.1)
Classic Acupoint vs Nonacupoint Mean Difference P value (95% CI)
−0.3 (−1.8, 1.1)
−4.5 (−7.6, −1.3)
Mean Difference (95% CI)
Classic Acupoint vs Unrelated Acupoint
162 558.5 (376.5) −24.1 (−76.9, 28.7) 162 461.7 (321.2) 162 356.1 (309.2)
162 162 162
162 162 162
162 162 162
162 162 162
162 162 162 162 162
No
Nonacupoint
Additional in-bed time during three menstrual cycles, minutes/day Baseline cycle 160 232.5 (242.2) 163 265.0 (272.7) Intervention cycle 160 134.2 (191.0) 163 161.0 (204.3) Postintervention cycle 160 135.2 (162.2) 163 164.4 (195.5)
527.9 (339.4) 438.8 (304.2) 365.8 (281.7)
3.4 (1.3) 2.7 (0.9) 2.6 (1.1)
Pain total time during three menstrual cycles, minutes/day Baseline cycle 160 492.2 (342.5) 163 Intervention cycle 160 406.1 (275.8) 163 Postintervention cycle 160 362.0 (294.0) 163
163 163 163
VRS scores during three menstrual cycles Baseline cycle 160 3.4 (1.3) Intervention cycle 160 2.6 (0.9) Postintervention cycle 160 2.5 (1.2)
10.5 (5.3) 9.2 (4.8) 7.2 (4.3)
65.3 (15.9) 52.5 (15.5) 49.4 (23.1)
163 163 163
10.5 (5.1) 8.9 (4.8) 7.2 (4.4)
19.2 (8.1) 17.7 (8.2) 14.5 (8.0)
59.1 (13.7) 46.1 (16.2) 40.9 (15.9) 31.9 (20.0) 37.0 (23.2)
Average VAS scores during three menstrual cycles, mm Baseline cycle 160 64.4 (16.4) 163 Intervention cycle 160 49.5 (15.6) 163 Postintervention cycle 160 44.2 (22.6) 163
163 163 163
163 163 163 163 163
18.8 (7.7) 16.9 (7.7) 14.7 (8.0)
RSS scores during three menstrual cycles RSS-COX 1 scores Baseline cycle 160 Intervention cycle 160 Postintervention cycle 160 RSS-COX 2 scores Baseline cycle 160 Intervention cycle 160 Postintervention cycle 160
mm 56.1 (11.8) 42.7 (17.5) 37.1 (17.2) 27.2 (18.2) 29.6 (19.2)
Mean (SD)
No
No
Mean (SD)
Unrelated Acupoint
Classic Acupoint
The per-protocol analysis on primary and secondary outcomes for each group
VAS scores following the first intervention, 0 minute (before first treatment) 160 5 minutes (during first treatment) 160 10 minutes (during first treatment) 160 30 minutes (during first treatment) 160 60 minutes (30 minutes after 160 completion of first treatment)
Table 3
Liu et al.
Single Point Acupuncture in Primary Dysmenorrhea RSS-Cox or VRS outcomes. To some extent, this finding is consistent with a recent systematic review. It assessed the analgesic effect of acupuncture in 13 trials (3,025 patients) involving a variety of pain conditions [12], and found that there was a 4 mm on a 100 mm VAS difference between acupuncture and placebo acupuncture. In this study, patients were naive to acupuncture treatment, and treatments were quite similar, all three treatment locations distant from location of pain, both of which should facilitate blinding. Acupuncture is likely to have a nonspecific effect, but whether it has a more specific physiological effect remains unclear. Apart from acupoint specificity, issues such as needling sensation, psychological factors, and needle retaining time also have relevant influence on the effectiveness of acupuncture [13–15]. In the largest clinical acupuncture studies for back and knee pain to date [16,17]—the German Acupuncture Trials (GERAC) trials—acupuncture was found to be almost twice as effective as standard therapy over 6 months. However, there were no differences between traditional Chinese and sham acupuncture given to nonacupoints, suggesting that the benefits may have been a placebo effect. Functional studies have shown that there are differences between the neural effects of expectation and acupuncture’s therapeutic analgesia in experimental settings [18,19]. The present study did show that there was clinical benefit of obtaining acupuncture treatment, and this benefit may be due to both nonspecific effects (e.g., empathy from clinical staff, doing a positive action, belief in the intervention, etc.) and specific effects of the treatment. A general physiological effect of needling may be the reason why significant differences between sham and specific needling are not detected [20]. Our study has several limitations. The nonacupoint seemed to be very close to unrelated acupoint GB39. There is continuing uncertainty about the size and area of acupuncture points. The acupuncture point was considered as an area, about 2 cm in diameter [21]. Because of spread of the electrical current between the acupuncture points, use of electrical stimulation may make it even more difficult to isolate activation of an acupuncture point from other points or nonpoints of the limb. There is also the possibility of the E-stim at 2 hz resulting in diffuse noxious inhibitory control/pain inhibition independent of acupoint stimulation. Both of the mechanisms—inadvertant acupoint stimulation and diffuse noxious inhibitory control, the supposedly inert control—may have had physiologic effect. This may impact on choice of control condition for future studies. Our statistical analysis made assumptions about missing values, specifically that the values were missing completely at random. Although this assumption is difficult to formally test, we found no evidence that patient dropout was treatment related (Tables S1–4 and Figure S1). All subjects received the acupuncture treatment starting on the first day of intervention cycle for 3 consecutive days. Response at the beginning of the menstrual cycle may differ from the end of the cycle, as pain is likely to improve toward the end of the menstrual cycle. This may impact on the effect of acupuncture.
In summary, specific acupoint acupuncture produced a statistically, but not clinically, significant effect compared with unrelated acupoint and nonacupoint acupuncture in primary dysmenorrhea patients. In the current study, we examined the effects of single-point acupuncture, not the effects of multiple points acupuncture. Study on single point effects may have more theoretical significance to identify point specificity. However, in daily clinical practice, multiple points acupuncture may provide better pain relief than single puncture acupuncture. Future studies should focus on effects of multiple points acupuncture on primary dysmenorrhea. Acknowledgments We thank the research staff at Dongzhimen Hospital Affiliated to Beijing University of Chinese Medicine: Ji-Ping Zhao, Ling Tang; Beijing Hospital of Traditional Chinese Medicine affiliated to Capital Medical University: Ya-Li Wen, Jing-Dao Li; Huguosi Hospital of Traditional Chinese Medicine affiliated to Beijing University of Chinese Medicine: Zhi-Liang Li, Xu Chen; China-Japan Friendship Hospital: Min Li, Ying Wang; The First Hospital affiliated to Tianjin University of Traditional Chinese Medicine: JingXian Han, Meng-Meng Wu; Shandong University of Traditional Chinese Medicine: Yu-Xia Ma, Kun Lv; and Beijing University of Chinese Medicine: Yan-Fen She, Hui-Juan Cao. The study was funded by the National Basic Research Program of China (973 Program, reference number: 2012CB518506). Author Contributions Conceived and designed the experiments: Zhu J, Liu J-P, Liu C-Z, Xie J-P. Performed the experiments: Wang L-P, Zhou W, Gao S-Z, Li S-L, Wang Y-X. Analyzed the data: Song J-S, Chen Y-Y. Wrote the article Liu C-Z, Liu Y-Q, Shi G-X, Ma L-X. Biostatistics analysis: Xing J-M. Study supervision: Liu C-Z, Xie J-P. References 1 Patel V, Tanksale V, Sahasrabhojanee M, Gupte S, Nevrekar P. The burden and determinants of dysmenorrhoea: A population-based survey of 2,262 women in Goa, India. BJOG 2006;113:453– 63. 2 Dawood MY. Primary dysmenorrhea: Advances in pathogenesis and management. Obstet Gynecol 2006;108(2):428–41. 3 Durain D. Primary dysmenorrhea: Assessment and management update. J Midwifery Womens Health 2004;49(6):520–8. 4 Kaptchuk TJ. Acupuncture: Theory, efficacy, and practice. Ann Intern Med 2002;136(5):374–83. 5 Vincent C. The safety of acupuncture. BMJ 2001; 323(7311):467–8. 919
Liu et al. 6 Lim S. WHO standard acupuncture point locations. Evid Based Complement Alternat Med 2010; 7(2):167–8.
17 Manheimer E, Linde K, Lao L, Bouter LM, Berman BM. Meta-analysis: Acupuncture for osteoarthritis of the knee. Ann Intern Med 2007;146(12):868–77.
7 Liu CZ, Xie JP, Wang LP, et al. Immediate analgesia effect of single point acupuncture in primary dysmenorrhea: A randomized controlled trial. Pain Med 2011; 12(2):300–7.
18 Dhond RP, Kettner N, Napadow V. Do the neural correlates of acupuncture and placebo effects differ? Pain 2007;128(1–2):8–12.
8 Han JS. Acupuncture: Neuropeptide release produced by electrical stimulation of different frequencies. Trends Neurosci 2003;26(1):17–22, 6.
19 Cho ZH, Oleson TD, Alimi D, Niemtzow RC. Acupuncture: The search for biologic evidence with functional magnetic resonance imaging and positron emission tomographytechniques. J Altern Complement Med 2002;8(4):399–401.
9 Cox DJ, Meyer RG. Behavioral treatment parameters with primary dysmenorrhea. J Behav Med 1978; 1(3):297–310.
20 Pyne D, Shenker NG. Demystifying acupuncture. Rheumatology (Oxford) 2008;47(8):1132–6.
10 Andersch B, Milsom I. An epidemiologic study of young women with dysmenorrhea. Am J Obstet Gynecol 1982;144(6):655–60.
21 Jing YC. The measurement on the area of acupuncture point by noninvasive way. J Guiyang College Tradit Chin Med 1985;6(3):58.
11 Dworkin RH, Turk DC, Wyrwich KW, et al. Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. J Pain 2008;9(2):105–21. 12 Madsen MV, Gøtzsche PC, Hróbjartsson A. Acupuncture treatment for pain: Systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no classic acupoint groups. BMJ 2009;338:a3115. 13 Paterson C, Dieppe P. Characteristic and incidental (placebo) effects in complex interventions such as acupuncture. BMJ 2005;330(7501):1202–5. 14 Moore A, McQuay H. Acupuncture: Not just needles? Lancet 2005;366(9480):100–1. 9–15. 15 White AR, Filshie J, Cummings TM. Clinical trials of acupuncture: consensus recommendations for optimal treatment, sham controls and blinding. Complement Ther Med 2001;9(4):237–45. 16 Manheimer E, White A, Berman B, Forys K, Ernst E. Meta-analysis: Acupuncture for low back pain. Ann Intern Med 2005;142(8):651–63.
920
Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Figure S1 The flow diagram of dropout patients in the study. Table S1 The reasons for dropout patients with missing data. Table S2 The baseline characteristics of patients with missing data, collected on the first day of the intervention cycle. Table S3 The primary outcomes of patients with missing data, collected during the intervention and postintervention cycles. Table S4 The secondary outcomes of patients with missing data, collected during the intervention and postintervention cycles.
