Effects of intensity of electroacupuncture on chronic pain in patients with knee osteoarthritis: a randomized controlled trial

Ethics approval

The protocol of this clinical trial was in adherence to the STRICTA guidelines and has been described in detail elsewhere [17]. This study was approved by the Chinese Ethics Committee of Registering Clinical Trials (reference: ChiECRCT-20140035) and registered with the Chinese Clinical Trial Registry (ChiCTR-ICR-14005411) on 31 October 2014 (http://www.chictr.org.cn/showproj.aspx?proj=9758). All patients provided written informed consent before randomization.

Study design

This was a multicenter, three-arm parallel, randomized controlled trial to compare the efficacy of the two groups of EA (weak EA and strong EA) with sham EA. Patients were enlisted through hospital-based recruitment and advertisements with posters, leaflets, and newspapers from November 2014 to March 2016.

Randomization and masking

After a 2-week washout period, patients who met our inclusion criteria were randomly assigned to one of the three groups (strong EA, weak EA, or sham EA) in a ratio of 2:1:1 using a computer-generated random allocation sequence through stratified block randomization method of SAS version 9.1.3 (SAS Institute, Cary, NC, USA). In our preliminary trial, we found that the effect of strong EA was better than that of weak EA. Moreover, the previous study investigating the efficacy of acupuncture compared with minimal acupuncture and no acupuncture in patients with KOA also used such a 2:1:1 randomization ratio [1]. For the consideration of patients’ welfare, the randomized group method was changed from 1:1:1 to 2:1:1. The randomization ratio in the register link in the Chinese Clinical Trial Registry (http://www.chictr.org.cn/showproj.aspx?proj=9758) has also been modified.

Randomization was performed by an independent research assistant who did not participate in any other section of this study. The acupuncturists were informed of the treatment assignment by mobile phone confirmation, and allocation concealment was not revealed until the final outcome analysis was reported.

Enrolled participants were only informed that they would receive one of the three acupuncture therapies; consequently, they were not aware of their treatment allocation. Acupuncturists were permitted to treat both knees if the two knees were affected by osteoarthritis. However, only the most symptomatic knee was evaluated for the outcome assessment throughout the study. Participants, clinical outcome evaluators, and statisticians were blinded to randomization, since it was not feasible to blind the acupuncturists who administered EA.

Participants

A total of 450 patients with KOA were recruited from 5 hospitals in Wuhan, China: the Combined Traditional Chinese and Western Medicine Hospital affiliated to Tongji Medical College of Huazhong University of Science and Technology; the Third Hospital of Wuhan; Central Hospital of Wuhan; Union Hospital affiliated to Tongji Medical College of Huazhong University of Science and Technology; and the Fifth Hospital of Wuhan. People aged 50 years or older who met the clinical criteria for KOA formulated by the American College of Rheumatology (ACR) were deemed eligible for inclusion [18]. We excluded patients who had ever experienced adverse reactions to acupuncture prior to our study; who had comorbidities including severe cardiovascular, cerebral, hepatic, renal, or hematopoietic diseases; who had other disorders that might affect the knee (e.g., rheumatoid arthritis, gouty arthritis); who were pregnant or attempting to become pregnant or were lactating; and who had a history of mental illness. All participants were required not to take analgesic medications and EA 48 h before each treatment session.

Interventions

Study interventions were performed by acupuncturists with at least 3 years clinical experience who were licensed Chinese medicine practitioners. To ensure standardization of the treatment protocol, each acupuncturist from the five hospitals underwent pretrial training on the study protocol, completing the case report form, treatment technique, and outcome assessment from the lead investigators who also monitored the process for this clinical study. Treatments for both the true EA groups and sham EA group consisted of ten 30-min sessions over 2 weeks. Assessments of participants were performed at baseline and at the end of the first and second weeks of the treatment phase.

During the 30-min stimulation period, participants were in a supine position with a pillow under each knee for support. Sterile disposable needles (30 gauge with an outer diameter of 0.32 mm and a length of 40 mm; Hwato, Suzhou, China) were used. Participants in the strong and weak EA groups received treatment at the same four acupoints of Neixiyan (EX-LE 5), Dubi (ST 35), Liangqiu (ST 34), and Xuehai (SP 10) unilaterally based on traditional Chinese medicine meridian theory [1, 19]. After local disinfection, needles were inserted to a depth of 25 to 40 mm vertically. De qi sensation was elicited by lifting and thrusting combined with twirling and rotating the needles. (De qi is the feeling experienced by patients at the needling site that includes fullness, heaviness, dull aching, or warmth and is indicative of effective needling.) Electrical stimulation was then applied using an EA apparatus (Shanghai Medical Electronic Instrument), with a pair of electrodes connecting acupoints EX-LE 5 with ST 35, and another pair of electrodes connecting SP 10 with ST 34 [20, 21]. Stimulation parameters were direct current, continuous wave, 2 Hz frequency, and 0.5 ms pulse width, for 30 min. After obtaining de qi sensation, the strong EA and weak EA groups received different stimulation intensity. The strong EA group received the maximum tolerable intensity of current between 2 and 5 mA. The weak EA group received low-intensity current between 0 and 0.5 mA. Once the current was felt, the participant informed the acupuncturist to stop increasing the current.

In the sham EA group, the number of acupoints, EA apparatus, and stimulation parameters were the same as for the true EA groups. However, the needles used in the sham EA group were fine and short (35-gauge needle with an outer diameter of 0.20 mm and a length of 25 mm; Hwato, Suzhou, China). The needles were inserted only superficially into non-acupoint sites, each 2 cm lateral to each of the four acupoints to an approximate depth of 5 to 10 mm. In addition, the needles were not manipulated to avoid obtaining de qi sensation. Electrical stimulation was delivered with the same low intensity as the weak EA group. For all three groups, after each session, all the needles and the EA electrodes were removed.

Outcome measurements

Participants completed questionnaires before treatment, after 1 week and after 2 weeks. Primary outcome measures were pain visual analog scale (VAS), CPM value, and Chinese translations of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [15, 22–25].

The VAS used in this study was a 10-cm line ranging from 0 (no pain) to 10 (pain as bad as it could be) that assessed peak pain intensity over the last 24 h. The WOMAC index consists of three domains, namely pain (5 items), stiffness (2 items), and physical function (17 items), and each item is scored based on a 5-point Likert numerable rating scale representing different degrees of intensity (none, mild, moderate, severe, or extreme). The final score of WOMAC was determined by adding the aggregate scores for three subscales, which ranges from 0 to 96, and a greater score indicates greater pain and dysfunction. The WOMAC has been translated and validated in Chinese [25].

In our current study, we used the terminology “CPM” instead of “DNIC,” because DNIC is a terminology used on animals. CPM represents the descending inhibitory modulation of pain. It can be assessed when two painful stimuli are applied simultaneously, the “conditioning” stimulus that typically inhibits the “test” stimulus [22]. In this study, to measure CPM, the acupuncturist applied a 180-g von Frey filament to the Ashi point (pain spot) of the affected knee within a 1-cm-diameter circle for three to five times. The participant was then asked to mark the intensity of pain on the VAS after each punch [15]. The research assistants recorded the mean value of the three VAS scores as VAS1. Next, the participant was asked to immerse the contralateral hand and wrist into cold water (10 to 12 °C) for 1 min. Afterward, the mean VAS score of pain elicited by the von Frey filament on the same Ashi point was recorded as VAS2. VAS scores were measured immediately after the conditioning stimulus since the CPM effects are generally short-lived. Percent changes were calculated for the change in CPM based on the following formula: CPM = (VAS1 − VAS2)/VAS1 × 100% [23, 24], where 0 indicated no change and higher values indicated more effective pain inhibition.

Secondary outcomes included the numeric pain rating scale (NPRS) [26], emotional scale (ES) [27], and present pain intensity (PPI) [28]. The research assistants assisted the participants in completing the survey instruments at the end of the 2-week study.

Research assistant documented severe adverse events and side effects associated with EA treatment. Participants were also asked to report side effects at the end of the study. Officers from the Scientific Research Office in the five hospitals formed the Data and Safety Monitoring Board, who periodically reviewed and evaluated the accumulated study data for participant safety, study conduct, and progress.

Sample size calculation and statistical analysis

Sample size estimation was performed to detect a minimal clinically important difference (MCID) of 1.8 units in VAS pain score (extrapolated from 18 mm MCID reported for 100 mm VAS) [29]. We also aimed to detect an MCID of 6.7 units of total WOMAC score identified as the MCID for osteoarthritis (extrapolated from 7 units of MCID reported for 0–100 normalized WOMAC total score) [30]. However, no literature reports MCID of CPM in patients with KOA. So, we calculate the sample size based on VAS and WOMAC score. The calculated sample size based on at least 80% power, 15% drop out, and a two-sided 5% significance level gave a required population of 67 subjects. Seventy-five participants in each group will give us up to 95% power to detect the true effect.

The statistical analysis plan was completed and approved by the data and safety monitoring board. Analyses of the baseline characteristics and clinical outcomes were based on the intention-to-treat (ITT) population, which included participants who had been randomized and baseline data recorded (n = 292). Baseline characteristics were presented as percentages for categorical variables and mean (SD) with 95% CIs for continuous variables. Missing data were imputed for each group separately, using chained equations with predictive mean matching. A total of 20 imputed datasets were completed based on the raw dataset, and then corresponding estimates were combined using Rubin’s rules.

Multiple linear regressions were used to compare the significant differences in mean changes from baseline between the groups for each outcome and were adjusted for basic characteristics (sex, age, duration of KOA, and body mass index) and baseline outcome score. We conducted a between-group comparison of strong EA and weak EA with sham EA as a control in the linear model, as well as comparison of the strong EA and weak EA groups. Mixed effect model was used to test the significance of change of effect between any two time points in different treatment groups. All analyses were performed using R (version 3.2.4; The R Foundation, Vienna, Austria) and its mice package, with differences considered significant if the two-tailed P value was less than 0.05.

Participants and baseline characteristics

Between September 2014 and March 2016, we screened 805 participants for eligibility among the 5 hospitals, of whom 301 were randomly assigned. Among the randomized participants, 271 (90.0%) completed the study. Multiple imputations were used for the missing data in 21 participants (9 in the strong EA group, 8 in the weak EA group, and 4 in the sham EA group). Dropouts at each stage and the number assessed for the primary end point are presented in Fig. 1. The baseline characteristics of the participants are shown in Table 1. They were comparable across the 3 groups.

Variable	Strong EA (n = 145)	Weak EA (n = 72)	Sham EA (n = 75)
Female, no. (%)	106 (73.1)	57 (79.2)	60 (80.0)
Age, mean (SD), years	64.6 (10.2)	63.7 (9.3)	61.9 (9.5)
Symptom duration, years (%)
< 0.5 years	51 (35.2)	27 (37.5)	27 (36.0)
0.5 to 3 years	60 (41.4)	27 (37.5)	23 (30.7)
3 to 5 years	14 (9.7)	8 (11.1)	15 (20.0)
≥ 5 years	20 (13.8)	10 (13.9)	10 (13.3)
Height, mean (SD), m	1.62 (0.07)	1.62 (0.07)	1.64 (0.08)
Weight, mean (SD), kg	59.8 (7.8)	59.4 (7.5)	61.8 (8.5)
BMI, mean (SD)	22.67 (2.22)	22.46 (1.87)	22.93 (1.86)
Previous treatment, no. of patients (%)
Physical therapy	25 (17.2)	11 (15.3)	12 (16.0)
Corticosteroid injections	5 (3.5)	2 (2.8)	2 (2.7)
Acupuncture	21 (14.5)	8 (10.8)	10 (13.3)
Exercise	36 (24.8)	15 (20.3)	16 (21.3)

BMI body mass index

Primary and secondary outcomes

For the primary outcome, the mean of CPM function score was 9.49 at baseline and 24.34 at week 2 in the strong EA group, 9.86 at baseline and 14.61 at week 2 in the weak EA group, and 9.46 at baseline and 10.89 at week 2 in the sham EA group (Additional file 1: Table S1). The change in CPM function score differed significantly among the three groups at 2 weeks after randomization (Table 2). The CPM function score increased in the strong EA group by 14.85, in the weak EA group by 4.75, and in the sham EA by 1.43; a greater increment of CPM function score was observed in the strong EA group than in the sham EA group (between-group difference 13.54; 95% CI 13.23 to 13.85; P < .01) and in the weak EA vs. sham EA group (3.80; 95% CI 3.45 to 4.15; P < .01). In addition, the strong EA group was also statistically different from the weak EA group (9.73; 95% CI 9.44 to 10.02; P < .01) (Table 2). As shown in Fig. 2, the mean CPM scores were similar among the three groups before treatment, with differences between true EA (strong and weak EA) and sham EA, and strong and weak EA becoming apparent after 2 weeks treatment. Moreover, at the end of week 2, VAS value, WOMAC, and all secondary outcomes (NPRS, ES, and PPI) were significantly lower in the two true EA groups than in the sham EA group (P < .01 for all comparisons), and strong EA was more effective in improving VAS, NPRS, and ES than weak EA (Table 2). Results of per-protocol analyses were included in the supporting information (Additional file 2: Table S2).

Outcome	Strong EA	Weak EA	Sham EA	Pairwise comparison
				Strong EA vs. sham EA		Weak EA vs. sham EA		Strong EA vs. weak EA
				Effect size (95% CI)	P value	Effect size (95% CI)	P value	Effect size (95% CI)	P value
Primary outcome
CPM, mean (SD)a
Weeks 0–1	0.64 (0.16)	0.12 (0.27)	0.44 (0.24)	0 (0 to 0)	.33	0 (− 0.31 to 0.32)	0.98	0.15 (− 0.13 to 0.42)	.29
Weeks 0–2	14.85 (0.16)	4.75 (0.28)	1.43 (0.24)	13.54 (13.23 to 13.85)	< .01	3.80 (3.45 to 4.15)	< .01	9.73 (9.44 to 10.02)	< .01
VAS, mean (SD)b
Weeks 0–1	− 1.34 (0.10)	− 1.52 (0.14)	0.65 (0.14)	− 0.58 (− 0.71 to − 0.44)	< .01	− 0.77 (− 0.93 to − 0.62)	< .01	0.19 (0.04 to 0.35)	.01
Weeks 0–2	− 2.97 (0.10)	− 2.75 (0.15)	1.19 (0.14)	− 1.58 (− 1.75 to − 1.4)	< .01	− 1.36 (− 1.57 to − 1.16)	< .01	− 0.22 (− 0.42 to − 0.03)	.03
WOMAC, mean (SD)c
Weeks 0–1	− 13.03 (0.56)	− 13.06 (0.87)	− 3.47 (0.80)	− 9.35 (− 10.75 to − 7.96)	< .01	− 9.66 (− 11.26 to − 8.07)	< .01	0.18 (− 1.23 to 1.59)	.80
Weeks 0–2	− 20.92 (0.56)	− 20.55 (0.89)	− 8.87 (0.81)	− 11.70 (− 12.52 to − 10.89)	< .01	− 11.55 (− 12.52 to − 10.58)	< .01	− 0.15 (− 1.06 to 0.75)	.74
Secondary outcome
NPRS, mean (SD)c
Weeks 0–1	− 1.59 (0.10)	1.12 (0.14)	0.72 (0.14)	− 0.93 (− 1.1 to − 0.76)	< .01	− 0.69 (− 0.9 to − 0.49)	< .01	− 0.11 (− 0.3 to − 0.07)	.24
Weeks 0–2	− 2.97 (0.11)	2.48 (0.14)	1.40 (0.14)	− 1.55 (− 1.76 to − 1.33)	< .01	− 1.07 (− 1.31 to − 0.83)	< .01	− 0.47 (− 0.7 to − 0.24)	< .01
ES, mean (SD)c
Weeks 0–1	− 1.69 (0.12)	− 1.36 (0.16)	0.71 (0.13)	− 0.91 (− 1.09 to − 0.73)	< .01	− 0.67 (− 0.88 to − 0.45)	< .01	− 0.3 (− 0.5 to − 0.1)	< .01
Weeks 0–2	− 3.28 (0.13)	− 2.56 (0.16)	1.24 (0.13)	− 1.88 (− 2.12 to − 1.65)	< .01	− 1.22 (− 1.49 to − 0.95)	< .01	− 0.67 (− 0.92 to − 0.42)	< .01
PPI, mean (SD)c
Weeks 0–1	− 0.70 (0.07)	− 0.68 (0.09)	0.29 (0.09)	− 0.44 (− 0.58 to − 0.29)	< .01	− 0.45 (− 0.62 to − 0.28)	< .01	0.04 (− 0.11 to 0.19)	.59
Weeks 0–2	− 1.46 (0.07)	− 1.39 (0.09)	0.65 (0.09)	− 0.78 (− 0.96 to − 0.6)	< .01	− 0.82 (− 1.03 to − 0.6)	< .01	0.04 (− 0.16 to 0.24)	.72

EA electroacupuncture, VAS visual analog scale, CPM conditioned pain modulation, WOMAC Western Ontario and McMaster Universities Osteoarthritis Index, NPRS numeric pain rating scale, ES emotional scale, PPI present pain intensity

^aHigher values indicate better status

^bRating scales 0 to 10, with 0 being no pain and 10 being excruciating

^cLower values indicate better status

During the treatment course (the first week), there was no significant difference among the three groups in the CPM value (P > .05 for all comparisons) (Table 2). On the other hand, compared with the sham EA group, both strong and weak EA groups showed a significant improvement in all pain-related scores of VAS, WOMAC, NPRS, ES, and PPI (P < .01 for all comparisons). Furthermore, no differences were observed between the strong EA group and the weak EA group in the WOMAC, NPRS, or PPI scores (P > .05 for all comparisons) except for the VAS and ES scores at week 1.

Safety

During the 2-week trial, 10 participants experienced serious AEs (5 coronary heart disease, 1 stroke, 1 fracture, 1 pulmonary embolism, 1 lung infection, and 1 nephritis). All individuals were admitted to a hospital, and their conditions were considered unrelated to the study or intervention. Among 145 participants (15.2%) who received at least 1 week of strong EA treatment, 22 AEs were recorded (15 subcutaneous hemorrhage or bleeding, 7 needling pain and nausea); 72 participants (13.9%) receiving weak EA reported 10 side effects (7 subcutaneous hemorrhage and 3 needling pain or nausea); and in 75 participants (14.7%) who underwent sham EA treatment, 11 side effects were recorded (7 subcutaneous hemorrhage or bleeding and 4 needling pain).