The effectiveness of pulsed electrical stimulation in the management of osteoarthritis of the knee: Results of a double-blind, randomized, placebo-controlled, repeated-measures trial


  • ANZCTRN: 12607000492459.



To determine the effectiveness of subsensory, pulsed electrical stimulation (PES) in the symptomatic management of osteoarthritis (OA) of the knee.


This was a double-blind, randomized, placebo-controlled, repeated-measures trial in 70 participants with clinical and radiographically diagnosed OA of the knee who were randomized to either PES or placebo. The primary outcome was change in pain score over 26 weeks measured on a 100-mm visual analog scale (VAS). Other measures included pain on the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), function on the WOMAC, patient's global assessment of disease activity (on a 100-mm VAS), joint stiffness on the WOMAC, quality of life on the Medical Outcomes Study Short-Form 36 (SF-36) health survey, physical activity (using the Human Activity Profile and an accelerometer), and global perceived effect (on an 11-point scale).


Thirty-four participants were randomized to PES and 36 to placebo. Intent-to-treat analysis showed a statistically significant improvement in VAS pain score over 26 weeks in both groups, but no difference between groups (mean change difference 0.9 mm [95% confidence interval −11.7, 13.4]). Similarly, there were no differences between groups for changes in WOMAC pain, function, and stiffness scores (−5.6 [95% confidence interval −14.9, 3.6], −1.9 [95% confidence interval −9.7, 5.9], and 3.7 [95% confidence interval −6.0, 13.5], respectively), SF-36 physical and mental component summary scores (1.7 [95% confidence interval −1.5, 4.8] and 1.2 [95% confidence interval −2.9, 5.4], respectively), patient's global assessment of disease activity (−2.8 [95% confidence interval −13.9, 8.4]), or activity measures. Fifty-six percent of the PES-treated group achieved a clinically relevant 20-mm improvement in VAS pain score at 26 weeks compared with 44% of controls (12% [95% confidence interval −11%, 33%]).


In this sample of subjects with mild-to-moderate symptoms and moderate-to-severe radiographic OA of the knee, 26 weeks of PES was no more effective than placebo.

Electrotherapy is often used to manage symptoms of osteoarthritis (OA). It is a relatively inexpensive, noninvasive, short-term treatment option, which is recommended in evidence-based clinical guidelines (1–4). One electrotherapy treatment, pulsed electrical stimulation (PES), has been reported to significantly decrease pain and improve function in knee OA (5–8). However, anecdotal evidence and our personal observations suggest that PES is not widely used.

PES is delivered through capacitive coupling using surface electrodes and conduction gel. While often being grouped with transcutaneous electrical nerve stimulation (TENS) (9), it does differ from TENS and interferential therapy in its specific electrical current parameters and its proposed method of action (6). In particular, it is delivered at subsensory intensity. That subsensory electrical stimulation is reported to be effective in managing pain suggests a local mechanism of action. This mechanism is at present poorly understood. However, there are many pain-mediating receptors in the periphery that may be affected by an externally applied electrical field by virtue of their endogenous electrical potential and the role of polarization in receptor function and nociceptor stimulation (10). It is possible that externally applied electrical stimulation interferes with this process and thus reduces pain perception.

PES is also reported to be a potential disease modifier through its capacity to up-regulate chondrocyte activity (11–14). This assertion has yet to be tested in humans, mainly because long-term effectiveness and compliance with use have yet to be established.

Since OA of the knee is a chronic disorder, we considered the earlier randomized controlled trials of PES of 4 (8) and 12 (6) weeks' duration to be relatively short. Additionally, Farr et al (5) in a prospective, longitudinal study referred to a dose-response relationship, suggesting that increasing PES use results in better pain management. This assertion has not been tested in an independent randomized controlled trial.

Since current treatment options have moderate effect sizes at best (15) and are often limited in use by contraindications and comorbidities (16, 17), we wanted to examine whether the reported improvements with PES use continued beyond 12 weeks. By doing so, we aimed to determine whether PES could provide a useful, low-risk addition to OA management with a view to studying its potential as a disease modifier.

The primary aim of this study was to determine whether PES decreased pain in the OA knee over 26 weeks. Other outcomes included function, patient's global assessment of disease activity, quality of life, physical activity, and overall perceived effect.


The full protocol of this double-blind, randomized, placebo-controlled, repeated-measures trial is reported elsewhere (18). The trial was approved by the Curtin University Human Research Ethics Committee (HR122/2006), and all participants gave written informed consent. All recruitment tasks, screening, measurements, and instructions for use of the PES and placebo devices were completed by the same experienced musculoskeletal physical therapist (REF).


Seventy participants (mean age 70 years, 53% men) were enrolled between September 2007 and April 2009. Diagnosis of OA of the knee was in accordance with the American College of Rheumatology modified clinical classification system (19). Plain radiographs available for 64 participants confirmed the diagnosis. Persistent and stable pain (defined as not getting worse or better overall despite short-term fluctuations) for a minimum of 3 months prior to study entry was confirmed in all participants by telephone interview. All participants had a baseline pain score of at least 25 mm on a 100-mm visual analog scale (VAS). Volunteers were excluded if they had coexisting inflammatory arthropathies, contraindications to electrical stimulation, skin disorders in the vicinity of the knee to be treated, total knee replacement scheduled during the study period, and/or insufficient English to follow instructions and complete forms.

Recruitment occurred through notices in published newsletters of community organizations, letters to medical general practices, and word of mouth. Data were collected in person at the University and by mail.

Randomization and blinding.

Allocation, stratified by sex, age (<60 years, 60–75 years, and >75 years), and baseline VAS pain scores (25–40 mm, 41–60 mm, and 61–100 mm), was performed independently by an administrator, not otherwise involved in the study, using computer-generated randomization in blocks of 6. Following the randomization process, the administrator provided the serial number of an appropriate device (placebo or active), and the device was then dispensed to the participant. This process ensured that all study investigators and participants remained blinded to allocation until analysis was complete.


A commercially available TENS stimulator (Metron Digi-10s) was modified by a biomedical engineer to deliver PES current parameters as follows: pulsed, asymmetrically biphasic, exponentially decreasing waveform with a frequency of 100 Hz and pulse width of 4 msec. Current was delivered via 120 mm × 80 mm multiple-use conductive silicone electrodes inserted into larger calico pockets (175 mm × 100 mm) to increase the contact surface area and reduce current density. Electrodes, positioned over the anterior distal thigh (anode) and anterior to the knee joint itself (cathode), were coupled to the skin using hypoallergenic conduction gel and secured with specially made neoprene wraps. The placebo device was identical in appearance and method of use; however, the current flow was programmed to turn off after 3 minutes. Since this was a subsensory treatment, this change was not detectable by participants.

Identical written instructions were provided to all participants. They were asked to wear the device 7 hours daily, preferably overnight, for 26 weeks. Specifically, participants attached the device and turned the intensity up until they could feel pins and needles or a prickling sensation under one or both electrodes. After achieving sensory output, participants were instructed to turn the intensity down until they could no longer feel any electrical stimulation. At this stage, a built-in locking mechanism was engaged that prevented subsequent adjustment of intensity without restarting the device.

Participants kept a log (hours) of device use over 26 weeks. At exit, they were asked to indicate whether they thought their device was a PES device or a placebo.

Background therapy.

Participants were advised to continue their usual treatment for OA throughout the study, including prescribed medications, health professional interventions such as exercise programs, and complementary therapies. However, they were counseled against starting any new treatments. A medication diary was kept by all participants.

Outcome measures.

The primary outcome was change in pain score over 26 weeks measured on a 100-mm VAS. Participants responded to the following instruction: “Consider the amount of pain that you have experienced due to arthritis in your treated knee over the past 48 hours. Please make a vertical mark crossing the line below at a point that you consider indicates how severe your pain has been.” The left-side anchor of the line was marked as “no pain” and the right-side anchor as “extreme pain.”

In addition, physical function (Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC] [20], Likert format 3.1) and patient's global assessment of disease activity (on a 100-mm VAS [21]) were measured to complete the core set of 3 primary efficacy variables, recommended by the Outcome Measures in Rheumatology Clinical Trials group (22). Administration of the WOMAC provided an incidental pain score. These outcomes were measured at baseline and at 4, 16, and 26 weeks. Other outcome measures included quality of life (the 36-item Medical Outcomes Study Short-Form 36 version 2 [SF-36 v. 2] health survey [23]) and joint stiffness (WOMAC 3.1) measured at baseline and at 4, 16, and 26 weeks; physical activity (Human Activity Profile [24] and Actigraph GT1M accelerometers worn for 7 consecutive days) measured at baseline and at 16 weeks; and an 11-point global perceived effect scale (25) administered at 16 and 26 weeks.

Two measures of physical activity were administered to enhance measurement precision. The Human Activity Profile is a valid and reliable self-report questionnaire but like all self-report questionnaires is subject to information bias (26). Accelerometers provided a direct measure of ambulatory physical activity. License agreements for the WOMAC 3.1, SF-36 v. 2 health survey, and Human Activity Profile were obtained before the study.

Sample size.

A priori calculations for sample size were based on reported VAS pain score (6), patient's global assessment of disease activity (6), and WOMAC function score (27) data.

It has previously been proposed that the minimal clinically important improvement in OA knee pain is 19.9 mm on a 100-mm VAS (28). Further, a change of 20 mm has been defined as the minimum required for classification as a primary responder by Osteoarthritis Research Society International (29, 30). Accordingly, we deemed that an improvement in the PES-treated group of 20 mm greater than that achieved in the placebo group would constitute a clinically meaningful and important difference. Allowing for withdrawal of 20% of participants, it was determined that a sample of 70 would be sufficient to detect a between-group difference of 20 mm in change on a VAS for pain, as well as differences in change equal to a clinically meaningful and important difference of 9.1 points for WOMAC function score and 18.3 mm for patient's global assessment of disease activity (28). Calculations specified a power of 80% and a 2-tailed test with an alpha level of 0.05.

Statistical analysis.

Analyses were performed on an intent-to-treat basis using SPSS, version 17.0. The last observation carried forward method was applied for participants who completed at least 1 set of followup data. Differences between groups at baseline and changes between baseline and 26 weeks were examined using independent t-tests, and 95% confidence intervals (95% CIs) are provided. To test for the fixed effect of treatment, repeated-measures analysis using a linear mixed model was performed for VAS pain score, patient's global assessment of disease activity, WOMAC scores, and SF-36 at each followup visit. Between-group comparisons of global perceived effect scale scores at 16 and 26 weeks were analyzed using independent t-tests. In secondary analyses, chi-square tests were used to compare proportions achieving a clinically meaningful and important difference in pain, function, and patient's global assessment of disease activity, as well as proportions reporting improvement in global perceived effect scale scores in each group at all followup visits. P values less than 0.05 were considered significant.


Characteristics of participants.

From September 2007 to November 2008, a total of 120 participants were provisionally vetted by telephone, and 85 were given appointments for formal screening and baseline assessment where appropriate. Seventy participants were randomized in the study (Figure 1).

Figure 1.

Flow of participants through the trial. FTA = failed to attend; PES = pulsed electrical stimulation; ITT = intent to treat. * = protocol too onerous; † = device uncomfortable to wear.

Baseline characteristics were comparable between groups with the exception of a lower body mass index (BMI) in controls (P = 0.04) (Table 1). Of the sample, 37 (53%) were men, the mean BMI was 28.1 kg/m2, and 48 (75%) had Kellgren/Lawrence (31) radiographic scores of 3 or 4. Symptoms were mild to moderate in severity. Overall, the mean baseline VAS pain score was 52 mm, with only 20 participants (29%) scoring >60 mm. By comparison, WOMAC pain scores were generally lower, with only 5 subjects (7%) scoring >60/100 on the normalized scale. WOMAC function scores also suggested low levels of disability.

Table 1. Baseline characteristics of the participants*
CharacteristicControl group (n = 36)PES-treated group (n = 34)
  • *

    Except where indicated otherwise, values are the mean ± SD. BMI = body mass index; VAS = visual analog scale; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; SF-36 v. 2 = Medical Outcomes Study Short-Form 36 version 2 health survey.

  • P = 0.04 versus controls.

  • Thirty-one subjects in the control group, 33 subjects in the pulsed electrical stimulation (PES)–treated group.

  • §

    Thirty-five subjects in the control group, 29 subjects in the PES-treated group.

Age, years68.9 ± 11.470.7 ± 8.9
Men, %5650
BMI, kg/m226.8 ± 4.329.4 ± 5.9
Duration of symptoms, years11.4 ± 7.812.6 ± 12.7
Time since osteoarthritis diagnosis, years9.4 ± 10.36.9 ± 7.4
Kellgren/Lawrence radiographic grade, no.  
Clinical features of osteoarthritis, no. (%)  
 Stiffness <30 minutes in morning31 (86.1)30 (88.2)
 Crepitus33 (91.7)31 (91.2)
 Bone tenderness21 (58.3)25 (73.5)
 Bone enlargement25 (69.4)30 (88.2)
 No palpable warmth36 (100.0)32 (94.1)
Osteoarthritis laterality, %  
 Right-side treated5062
Medication use, no. (%)§  
 Complementary (glucosamine and fish oil)19 (54)15 (52)
 Analgesic18 (51)10 (34)
 Nonsteroidal antiinflammatory15 (43)13 (45)
Pain, 0–100-mm VAS52 ± 18.251 ± 17.2
Patient's global assessment of disease activity, 0–100-mm VAS47 ± 24.544 ± 19.3
WOMAC score (all normalized to 100)  
 Pain subscale36 ± 18.135 ± 16.3
 Stiffness subscale41 ± 18.745 ± 20.9
 Function subscale34 ± 16.535 ± 17.6
 Total score34 ± 14.636 ± 16.8
SF-36 v. 2 measures  
 Physical component summary score36.5 ± 9.137.0 ± 8.5
 Mental component summary score53.7 ± 11.252.7 ± 11.0
Human Activity Profile score  
 Maximum activity77 ± 8.273 ± 9.6
 Adjusted activity63 ± 12.461 ± 13.7
Accelerometer data  
 Number of days of use6 ± 0.36 ± 0.4
 Daily accelerometer count178,181 ± 82,192211,250 ± 98,574
 Daily resting time, minutes992 ± 90.5972 ± 94.8
 Daily light activity, minutes333 ± 78.6345 ± 78.3
 Daily moderate activity, minutes105 ± 56.8122 ± 63.6
 Daily hard activity, minutes0.1 ± 0.30.2 ± 0.6

Participants were physically active, with 51 (73%) classified as moderately active or active according to their Human Activity Profile–adjusted activity score. Similarly, accelerometer data showed that the average daily time spent in moderate-level activity (3–6 metabolic equivalents) exceeded the 30 minutes per day, 5 days per week recommended by Haskell et al (32) to gain health benefits (Table 1).

Participants used a variety of nonsteroidal antiinflammatory and analgesic medications as well as combinations of fish oil and glucosamine. At baseline, while there was a trend toward less analgesic use in the PES-treated group, there were no statistical differences between the groups in use of either prescribed or complementary medication (Table 1). Over the 26 weeks, there was little variation noted in either the type of medication used or the dosage in either group. Six participants failed to complete their medication diary, while 7 who did complete their diary were not using any medication for OA.

Device use, adverse effects, and blinding.

At 26 weeks, 20 (59%) of the PES-treated group and 13 (36%) of the controls achieved ≥100% target usage (P = 0.03). However, effective device use, defined as achieving 80% of the prescribed target, did not differ significantly between groups (25 [74%] in the PES-treated group, 22 [61%] in the controls; P = 0.11). The decision to define 80% of the prescribed target as effective device use closely reflected the minimum end of the accepted device use range of 6 hours per day reported in previous studies (5, 6, 8).

Twelve participants had adverse skin reactions in the form of rashes that were localized and mild. There was no difference between groups in the proportion of participants affected (6 [18%] in the PES-treated group, 6 [17%] in the controls; χ2 = 0.1, P = 0.9). Affected participants were advised to desist from device use until the rash had settled, after which they were able to resume treatment. Two participants in the control group used the device intermittently because of recurring skin reactions.

Thirty-one participants (12 in the PES-treated group and 19 in the control group) believed they knew whether their device was active or not, but their ability to identify their group correctly did not differ from chance (6 [50%] in the PES-treated group, 10 [53%] in the controls; κ = 0.02, P = 0.9). Thirty-five participants (19 in the PES-treated group, 16 in the controls) did not know whether their device was active or inactive, and 4 others did not complete the question. Thirteen of the 15 participants who believed they had used the active device reported feeling better on the global perceived effect scale. Conversely, of the 16 who believed their device was inactive, 13 reported no change or worse on the global perceived effect scale (κ = 0.68, P < 0.001). This suggests that blinding of participants was successfully maintained, since participants' opinion on which device they had been allocated was largely influenced by their outcome.

Pain and function.

For VAS pain score, patient's global assessment of disease activity, and function, there were statistically significant within-subject changes in each group over 26 weeks except for WOMAC pain and function scores in the PES-treated participants. However, between-group mean differences in change for VAS pain score (0.9 [95% CI −11.7, 13.4]), WOMAC pain score (−5.6 [95% CI −14.9, 3.6]), patient's global assessment of disease activity (−2.8 [95% CI −13.9, 8.4]), and WOMAC function score (−1.9 [95% CI −9.7, 5.9]) were not significant (Table 2). Interestingly, mean change in VAS pain score over 26 weeks approached a clinically meaningful and important difference, unlike patient's global assessment of disease activity or function scores (Table 2). There were no between-group differences for changes in pain, patient's global assessment of disease activity, or function at any of the earlier time points (Figure 2). The proportion of participants achieving a clinically meaningful and important difference for VAS pain score at 26 weeks did not differ significantly between groups (19 [56%] for the PES-treated group and 16 [44%] for the control group; between-group proportion difference 12% [95% CI −11%, 33%]) (Table 3).

Table 2. Changes from baseline in outcome variables*
 Control group (n = 36)PES-treated group (n = 34)Between-group mean change difference (95% CI)
  • *

    Except where indicated otherwise, values are the mean ± SD change at 26 weeks. Negative values for SF-36 v. 2, Human Activity Profile, accelerometer count, and moderate and hard activity all represent improvement. For all other variables, positive values represent improvement. There were no significant differences between the groups. 95% CI = 95% confidence interval (see Table 1 for other definitions).

  • Changes in physical activity were measured from baseline to 16 weeks.

  • Thirty-four subjects in the control group, 33 subjects in the PES-treated group.

Pain, 0–100-mm VAS19 ± 31.120 ± 20.70.9 (−11.7, 13.4)
Patient's global assessment of disease activity, 0–100-mm VAS14 ± 28.011 ± 17.9−2.8 (−13.9, 8.4)
WOMAC score (all normalized to 100)   
 Pain subscale10 ± 18.45 ± 20.4−5.6 (−14.9, 3.6)
 Stiffness subscale5 ± 19.39 ± 21.53.7 (−6.0, 13.5)
 Function subscale7 ± 16.25 ± 16.5−1.9 (−9.7, 5.9)
 Total score7 ± 15.56 ± 16.0−1.3 (−8.8, 6.3)
SF-36 v. 2 measures   
 Physical component summary score−2.6 ± 7.3−1.0 ± 5.61.7 (−1.5, 4.8)
 Mental component summary score−2.4 ± 8.1−1.2 ± 9.31.2 (−2.9, 5.4)
Human Activity Profile score   
 Maximum activity1 ± 7.9−1 ± 6.8−2.0 (−5.6, 1.5)
 Adjusted activity0.5 ± 8.80.2 ± 6.2−0.3 (−4.0, 3.3)
Accelerometer data   
 Daily accelerometer count−5,419 ± 52,48812,600 ± 52,40918,020 (−7,566, 43,607)
 Daily resting time, minutes−12 ± 88.7−28 ± 85.5−15.8 (−58.3, 26.7)
 Daily light activity, minutes19 ± 67.617 ± 65.2−2.3 (−34.7, 30.1)
 Daily moderate activity, minutes−0.4 ± 36.311 ± 41.211.5 (−7.4, 30.4)
 Daily hard activity, minutes−0.2 ± 1.0−0.01 ± 0.70.1 (−0.3, 0.6)
Figure 2.

Scores for pain on a visual analog scale (VAS), patient's global assessment of disease activity, and function at each time point. Error bars indicate 95% confidence intervals. There were no between-group differences in change over time for any of these variables. WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; PES = pulsed electrical stimulation.

Table 3. Results from secondary analysis*
Outcome measureControl groupPES-treated groupBetween-group proportion difference (95% CI)
  • *

    Values are the number (%) of participants achieving 20-mm change in VAS pain score and minimal clinically important improvements (28) in patient's global assessment of disease activity (PGA) score (18.3 mm on a 0–100-mm VAS) and WOMAC function subscale score (9.1, normalized to 100) at 26 weeks. There were no significant differences between the groups. 95% CI = 95% confidence interval (see Table 1 for other definitions).

VAS pain score16 (44)19 (56)12% (−11%, 33%)
PGA score16 (44)13 (38)−6 (−28%, 16%)
WOMAC function subscale score14 (39)13 (38)−1% (−22%, 22%)

Quality of life and activity.

The SF-36 physical component summary measures were slightly below the OA population norm, and the mental component summary measures were slightly above (33) (Table 1). Only small improvements in physical and mental component summary measures occurred over 26 weeks, with no statistical differences between groups (Table 2). Patterns of change observed in the subscale scores were similar to those seen in the summary scores (Table 4). Changes in both Human Activity Profile and accelerometer physical activity measures at followup were small, with no significant differences between groups (Table 2).

Table 4. Mean normalized values of SF-36 v. 2 subscale and component summary scores by time*
 Osteoarthritis population norm, mean ± SDControl groupPES-treated group
Baseline4 weeks16 weeks26 weeksBaseline4 weeks16 weeks26 weeks
  • *

    PCS = physical component summary score; MCS = mental component summary score (see Table 1 for other definitions).

  • Ware and Kosinski (33).

Physical functioning38.97 ± 12.7533.7133.1236.3436.3434.0734.0736.0534.94
Roles, physical41.20 ± 11.9639.1739.7140.4641.4839.7841.4141.3741.66
Bodily pain40.77 ± 9.8640.0742.0742.2745.8640.7443.2842.5944.92
General health42.99 ± 10.7049.4448.5250.7049.4346.9146.3046.6344.89
Vitality45.31 ± 10.0748.2848.1950.7050.6248.5149.3449.7048.79
Social functioning43.69 ± 12.5446.8547.6149.7349.4348.1948.6747.8748.35
Roles, emotional45.57 ± 12.7245.4148.0048.4349.6245.0346.5145.2646.62
Mental health47.56 ± 10.6451.7351.6552.8253.3749.5950.7552.2451.99
PCS38.85 ± 11.8136.4736.3338.2839.1236.9637.7038.2837.95
MCS48.72 ± 10.9853.6554.8155.8456.0752.6853.7653.4353.85

Global perceived effect scale.

Similar to most other indices, global perceived effect scale scores at both 16 and 26 weeks did not differ between groups (at 16 weeks, mean difference 0.11 [95% CI −0.83, 1.04]; at 26 weeks, mean difference 0.78 [95% CI −0.22, 1.78]).

Adjustment for covariates including sample characteristics, baseline measures, and amount of device use did not alter any of the findings for any of the variables.


In patients with OA of the knee, PES treatment over 26 weeks was no better than placebo for reducing pain and improving physical function. Our results were consistent across all time points and outcome measures. Importantly, individual outcome results were reflected by the participants' overall perception of their response to treatment on the global perceived effect scale. An unacceptably low proportion of participants using PES achieved the clinically meaningful and important difference for pain, patient's global assessment of disease activity, and function. Moreover, the control group responses were comparable.

Investigators in previous randomized controlled trials of PES, conducted under the auspices of Bionicare, the commercial supplier, have reported favorable responses to PES compared with placebo (6, 8). Our results are clearly different.

Both our PES and placebo electrical parameters and method of application were comparable with those of Bionicare in frequency (100 Hz) and waveform (spiked, exponentially decreasing shape). Zizic et al (8) and Garland et al (6) described their current as monophasic, but ours was slightly biphasic to avoid skin irritation (34). Comparison of our 16-week mean change data with the 12-week data reported by Garland et al (6) reveals strikingly similar “PES treatment group” outcomes. The mean ± SD change for VAS pain score in our study was 12.0 ± 22.6 mm, while that reported by Garland et al was 14.7 ± 23.1 mm. However, while our placebo device response was consistent with that expected in OA clinical trials (35), with a mean ± SD change of 14.4 ± 27.4 mm, the placebo group of Garland et al showed very little change (mean ± SD 2.3 ± 22.0 mm). This comparison was consistent across the 3 primary efficacy variables of pain, patient's global assessment of disease activity, and function and indicates that the difference in study outcomes appears to be due to differences in placebo responses rather than differences in PES characteristics or equipment.

The possibility should be considered that the 3 minutes of placebo treatment could be therapeutic. However, since the placebo device used by Zizic et al (8) and Garland et al (6) also delivered 3 minutes of treatment and did not show a therapeutic effect, this is unlikely to be the case.

A comparison between the sample characteristics reported by Garland et al (6) and Zizic et al (8) and among the participants in our study is limited by the different outcome measures used and characteristics reported. However, in both the Garland et al (6) and Zizic et al (8) studies, higher scores in function outcome measures at baseline were recorded, meaning that OA was affecting the health status of their participants to a greater extent than that of our study participants. Additionally, while the VAS pain scores reported by Garland et al (6) were similar to ours, their WOMAC pain scores were higher. Where reported, age and years since diagnosis were similar, while participants in the study by Garland et al (6) had higher BMIs and a higher proportion of women (66% versus 47%). It is unlikely that these differences account for the contrasting results.

A therapeutic response to placebo treatment is well documented in the OA literature (35, 36). A number of characteristics of our study have previously been noted to contribute to a robust placebo response (35, 37–39). Blinding was apparent throughout, and the level of commitment required to participate was considerable and over an extended period of time. Furthermore, pain was the primary outcome. Participants had also been informed that previous trials of the modality had produced encouraging results, so their expectations concerning improvement, along with their desire to contribute in an affirmative way, may have contributed to the positive response to the placebo control device.

Men accounted for just over 50% of the sample, whereas OA of the knee is usually more prevalent in women in the age group represented by this sample (40, 41). The mild-to-moderate baseline pain scores, mild levels of disability, and high physical activity levels are also not typical. Thus, the sample may not be representative of the OA population.

It may be that PES is more effective in some subgroups of people with OA. It is well recognized that OA is a heterogeneous disease (42–44) and that causes of pain and pain mechanisms in OA are multifactorial (45–47). PES may be a more appropriate treatment modality in those patients in whom local pain mediators, which rely on membrane ion channels that may be affected by externally applied electrical stimulation, are the main cause of pain. In contrast, those in whom biomechanical changes or psychosocial factors are the main contributors to pain production may be less responsive. These latter factors were not measured in this study. Therefore, while the outcome of this study is decisive, it may not be possible to generalize the results to the wider OA population.

Both accelerometer data and Human Activity Profile scores confirmed moderate levels of activity of the cohort at baseline. This may have limited the scope for further improvement and suggests that study participants were managing the functional impact of their disease quite well. However, since accelerometer data were recorded and reported cumulatively, it was not possible to determine whether moderate physical activity was performed in blocks of at least 10 minutes, as recommended by Haskell et al (32) for general health improvement.

The study was designed so that reporting of results conforms to the Consolidated Standards of Reporting Trials group statement (48). Both the subsensory nature of PES and robust allocation concealment meant that blinding was a major strength of this study. All participants were screened, assessed, and managed over 26 weeks by 1 experienced musculoskeletal physical therapist, thus avoiding investigator bias.

The recruitment of highly motivated volunteers may have resulted in a sample with characteristics different from those of the OA population in general. Additionally, the strong sense of commitment and desire to please noted in these participants may have enhanced the placebo response. It is unknown what influence these sample characteristics have had on the overall study outcome.

A priori calculation of sample size based on the achievement of a 20-mm improvement in the PES group over that achieved by the placebo group had the potential to limit interpretation of the results, given that a minimum baseline VAS pain score of 25 mm was an inclusion criterion for the study. That is, if a substantial number of PES group participants were to achieve a final VAS pain score of zero mm and a similar number in the placebo group were to achieve VAS pain scores <20 mm, the capacity to detect a 20-mm difference between the 2 groups would have been compromised. However, since only 1 person in the whole sample (a PES group participant) achieved a VAS pain score of zero mm at 26 weeks, a true difference between the groups could be calculated and reported.

So as not to disadvantage those in the control group over such a lengthy period, participants were instructed to continue with their usual OA management. Apart from medication no concurrent treatments were recorded. Consequently, there may have been unknown confounding factors that might have influenced the outcome.

In this sample of subjects with mild-to-moderate symptoms and moderate-to-severe radiographic evidence of OA of the knee, PES was no more effective than placebo in achieving improvements in pain, function, quality of life, or physical activity. Therefore, results of this study do not support more widespread use of this modality.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Fary had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Fary, Carroll, T. G. Briffa, N. K. Briffa.

Acquisition of data. Fary.

Analysis and interpretation of data. Fary, Carroll, T. G. Briffa, N. K. Briffa.


We thank all participants who enrolled in the study. Mr. Chris Tingley, Senior Biomedical Engineer, Department of Medical Technology and Physics, Sir Charles Gairdner Hospital, Nedlands, Western Australia developed the experimental equipment and placebo circuit. Dr. Ritu Gupta, Statistician, Curtin University contributed to the statistical analysis in the study design phase and developed the computer-generated randomization software used in the study. Dr. Richard Parsons, Statistician, Curtin University assisted with statistical analysis of data collected.