Exercise therapy for knee osteoarthritis pain: how does it work? A study protocol for a randomised controlled trial

This study is substantially funded by the Research Foundations Flanders (project number G040919N), has ethics approval (B.U.N. 143201941843). The clinical trial started on 10 January 2020. The manuscript is formatted by using the Standard Protocol Items: Recommendations for Interventional Trials statement.

Trial design

A multicentre randomised controlled trial with 3-month intervention and 12-month follow-up will be conducted. The trial is designed in line with the OARSI clinical trial recommendations for rehabilitation interventions in OA patients.58

Study setting

All assessments and treatments will take place at the participating hospital (UZ Brussel) or at the participants’ homes.

Eligibility criteria

Patients should meet the following inclusion criteria:

(1) KOA according to the clinical American College of Rheumatology (ACR) criteria.59 60 The clinical ACR criteria59 for KOA are: knee pain and at least three of the six following features: age ≥50, morning stiffness <30 min, crepitus, bony tenderness, bony enlargement, no palpable warmth. KOA will be confirmed with radiographs, including anterior-posterior and medio-lateral radiographs for imaging of the tibiofemoral joint, and an axial view for imaging of the patellofemoral joint. Patients with tibiofemoral (and patellofemoral) OA will be included. Kellgren and Lawrence (K&L) grading system for OA61 62 will be applied, with K&L grade 2 or higher defined as OA; radiographic KOA is defined as definite osteophytes and possible joint space narrowing63 64; (2) pain, nominated by the patient as three or higher on a visual analogue scale on most days of the last 3 months65; (3) aged ≥50 years.

Exclusion criteria are:

  • Treatment with exercise therapy or joint infiltrations (eg, corticosteroids, hyaluronic acid) in the preceding 6 months.

  • Being on a waiting list for knee replacement.

  • Any contra-indication for exercise therapy as established by the treating physician.

  • Corticosteroid infiltrations in the last 6 months.

  • Cognitive impairment (unable to understand the test instructions and/or Mini-Mental State Examination score <23/30).

  • Unable to understand the Dutch language.

  • Inflammation unrelated to OA (eg, due to acute or chronic infection) established by CRP >10 mg/L.

  • Presence of a disorder (eg, cancer, fibromyalgia, rheumatoid arthritis) and/or medication (eg, opioids, immunotherapy, anti-epileptics) that influences pain and/or the immune system.


Muscle strength training

Muscles of both legs are trained at 3 sets of 10 repetitions at 12 repetition maximum (12RM, that is, the maximum weight that can be lifted in 12 consecutive movements). Each set consists of short bouts of intensive concentric and eccentric muscle contractions. 12RM will be assessed at baseline and the exercise intensity will be progressively increased. The 12RM assessment involves gradually increasing the resistance until the participant reaches a load that he/she can perform the exercise for 12 repetitions but not for more. By using elastic bands, the resistance can be easily modified to meet the specific needs and goals of the individual, providing a customised and adaptable training experience. Every six sessions, the 12RM will be reassessed and the resistance of the elastic bands adapted. One session will consist of different exercises containing muscles of the hip (adductors and abductors) and the knee (extensors). More details can be found in table 1 describing all treatment arms according to the Template for Intervention Description and Replication checklist and guide.66 Due to the COVID-19 crisis and the associated restrictions, we will give patients the opportunity to choose between physical interventions sessions (at campus) or digital sessions (tele-consults). As such, we can meet concerns regarding COVID-19 and guarantee therapy continuation. Tele-consults will be held by the physiotherapists using a secured platform that allows video calling.

Table 1

Consensus exercise reporting template for treatment arms behavioural graded activity (BGA) and muscle strengthening training (MST)

Behavioural graded activity

Patients will receive a behavioural treatment integrated within the concepts of operant conditioning with exercise therapy. The purpose of BGA is to increase the level of activities in a time-contingent manner and increase the level of physical activity in the patient’s daily lives.46 BGA consists of three phases: (1) education and setting the baseline, (2) treatment (pacing) and (3) integration in daily life.46 The BGA treatment protocol is detailed in available treatment manuals,67 68 and more details can be found in table 1.

Care as usual

Patients allocated to the control group will receive care as usual. They will be asked to maintain their current lifestyle and treatment (if any) and to refrain from other new interventions for 24 weeks. After completion of the study (W64) control patients can have the opportunity to receive an initiation/info session about MST or BGA. Just like the patients in the remaining study arms, control patients will also receive an email with the results of the trial and contact details of possible physiotherapists after completion of the study.

Fidelity measurements

Before initiating the data collection, therapists will be trained face-to-face in providing the therapy (either MST or BGA) by experienced researchers with a clinical background. The duration of this training varies between 3 hours (MST) and 5 hours (BGA). Every 3 months, a half-day refresher course (face-to-face or online, based on therapists’ preference) will be organised, allowing therapists to discuss any difficulties experienced. Additional measures taken to allow quality control of the provided treatment include developing and using a programme manual that depends on available guidelines and manuals. Also, in case of permission of the participants all treatment sessions 4 weeks before the 3-monthly refresher course of the therapists will be recorded.69 Independent raters will successively evaluate a random selection of the tapes of each therapist using the Cognitive Therapy Adherence and Competence Scale70 and fidelity criteria. They have a bio-/psycho-medical background and will not be involved in providing the treatments. They will be specifically trained to score the recordings of treatments. The Steering Committee, which consists of the three main investigator (IB, JN, DB) will develop a measure of fidelity criteria before the start of the interventions. The method to develop these fidelity criteria has previously been described71 and will take into account the available guidelines and manuals. The fidelity criteria will be used to a priori inform and train therapists in providing the treatment, and they will be used to score the recordings of treatment sessions.71


Outcomes were chosen in line with the OARSI recommendations for clinical trials of rehabilitation interventions for OA58 and the IMMPACT/OMERACT recommendations for the assessment of physical function in chronic pain clinical trials72 (see table 2 and online supplemental appendix 1). The collection of personal characteristics (age, sex, comorbid medical conditions, etc)73 will be carried out once at baseline. The primary outcome is pain severity and will be assessed with the pain subscale of the Knee injury and Osteoarthritis Outcome Score (KOOS).74 The KOOS includes WOMAC Osteoarthritis Index LK3.075 in its complete and original format. WOMAC (and therefore the pain and symptoms subscale of the KOOS) is a valid tool for subjects with KOA. The KOOS is proven to generate valid and reliable scores.76 77 The KOOS shows adequate content validity, internal consistency, reliability, content validity and responsiveness for age and condition relevant subscales.77

Supplemental material

Table 2

Outcomes of the Knee OsteoArthritis PAIN study

Treatment mediator: inflammation

Blood sampling will be performed at baseline (before starting the intervention), at week 2–3 during intervention (pretreatment, post-treatment and 1-hour post-treatment session), at week 10–11 during intervention (pretreatment, post-treatment and 1-hour post-treatment session), immediately after intervention (week 13, at least 48 hours after the last training session to avoid bias due to acute exercise-induced elevations of biomarkers) and at week 26 (13 weeks after intervention) for subjects allocated to the MST or BGA group.

These blood samples will be used for biomarker profiling. A panel of biomarkers for CLIP will be included: ultra-sensitivity ELISA for high-sensitivity CRP (hsCRP), IL-1β, IL-1RA, IL-6, IL-8, IL-10, MCP-1, TNFα, sTNFR1&2, CXCL-10, CX3CL-1, MIG (CXCL-9) and suPAR.

Given the involvement of advanced glycation end products (AGE, resulting from non-enzymatic condensation of glucose with an aminogroup of proteins) and the receptor for AGE (RAGE) in CLIP78 and OA,79–81 we will add carboximethyl-lysine, pentosidine and sRAGE to the inflammatory panel. The panel will be completed by a set of adipokines79 including adiponectin, leptin, visfatin and resistin. In addition, the role of cytoprotective mechanisms and their modulation by physical exercise should be considered.19 82 83 We will also include hsp27 and hsp70 (ultra-sensitivity ELISA) in our biomarker panel. Since circulating brain-derived neurotrophic factor (BDNF) is stimulated by exercise both in older healthy84 and KOA patients,85 we will add circulating BDNF to our panel. Finally, we will also measure circulating markers of collagen turnover including COMP, metalloproteinase-1 and metalloproteinase-3 to monitor the degenerative process as a possible source for inflammation.

At week 2–3 and at 10–11, blood will be sampled immediately before and immediately after an MST or BGA session. Additional sampling will be performed at 1 hour after the session. Eventually, a catheter placement may be considered so that the participant does not need to be pricked multiple times (>3) and/or when the participant has small vessels and is hard to puncture. This design allows to appraise changes in inflammatory biomarkers appearing early in the circulation (IL-6, IL-8, BDNF, MCP-1) which are considered as ‘myokines’86–88 secreted by exercising muscles, as well as biomarkers appearing later during and after exercise (IL-10, IL-1RA, sTNFR1&2), which are produced by immune cells as a response to stimulation by the myokines. All biomarkers (as mentioned above) will be determined at these timepoints, including the circulating markers of collagen turnover as exercise-induced changes (especially COMP) are closely related to knee joint loading.30 89

Subjects that are allocated to the control group, will not undergo blood sampling at week 2–3 and week 10–11. Blood will be collected at baseline and reassessed after 1 hour of rest, immediately after intervention (week 13, at least 48 hours after the last training session to avoid bias due to acute exercise-induced elevations of biomarkers) and at week 26 (13 weeks after intervention).

Serum and plasma will be collected and stored at −80°C until assayed (simultaneously for all time points). Biomarkers will be measured using ultra-sensitivity ELISA and/or multiplex (Luminex platform) assays. To reduce bias due to intra-assay and inter-assay variability for each participant, samples of all time-points will be assayed on the same plate.

Treatment mediator: features of central sensitisation

Features of central sensitisation will be assessed by evaluating (1) electrical detection thresholds (EDTs) and electrical pain thresholds (EPTs),90 (2) temporal summation (TS),91 92 (3) conditioned pain modulation (CPM),92–94 (4) offset analgesia94–96 and (5) event-related potentials following electrical stimulation97 by a single evaluator. All these assessments will be done at baseline (before starting the intervention), after the intervention (week 13) and at week 26 (13 weeks after intervention). A subset of these assessments (ie, EDTs, EPTs and TS) will be performed at week 2–3 and at week 10–11 during the intervention (pretreatment, post-treatment and 1-hour post-treatment) in a separate test moment, not together with the blood sampling and self-reported measures).

EDTs and EPTs

EDTs and EPTs will be measured (Surpass LT stimulator, EMS Biomedical, Korneuburg, Austria) at the knee (peripatellar region: 3 cm medial to the midpoint of the medial edge of patella98) of the symptomatic leg, and at the median nerve ipsilateral to symptomatic leg side (as remote site). The order of test locations will be randomised in everyone. Each stimulus will be a constant current rectangular pulse train consisting of 5 pulses delivered at a frequency of 250 Hz, each lasting 1 ms.99 Stimulation will start at 0 mA and will be gradually increased using steps of 0.5 mA100 until the patient is experiencing a faint sensation (=EDT) and further until the stimulus is experienced as painful (=EPT). The mean EDT and mean EPT of the respectively three measurements with 30 s interval will be used for further analyses. EPTs has demonstrated to be reliable for the assessment of the sensitivity of the spinal nociceptive pathways in people with chronic pain.101

Endogenous pain facilitation assessed by the TS paradigm

TS will be assessed by delivering 20 stimuli100 at the previously determined intensity of the EPT, with an inter-stimulus interval of 0.5 s. The same order of test location as during EDT and EPT will be used. Each test region will be assessed three times, interspersed with a 30 s rest. The patient will be asked to give a verbal numeric rating scale (VNRS) score ranging from 0 (=no pain) to 10 (=worst possible pain)100 after the 1st, the 10th and the 20th stimulus. The outcome measures for TS will be the differences between the 10th and 1st VNRS score,92 102 the 20th and 10th VNRS score and 20th and 1st VNRS score.

Endogenous pain inhibition assessed by offset analgesia

Offset analgesia can be described as a disproportionately large decrease in perceived pain following slight decreases in noxious thermal or electrical intensity.103 Traditionally noxious thermal intensity has been used by researchers in order to evoke off-set analgesia.103 However, in this study, noxious electrical intensity will be used which has been validated to be used same as noxious thermal intensity.94

Electrical stimuli will be applied as a train of rectangular pulses (frequency: 100 Hz; pulse duration: 1 ms) delivered by a constant current simulator. The test site will be located and marked 3 cm distal to the elbow joint on the volar site of the dominant hand side forearm. The stimulation intensity will be stated according to electrical pain perception values (EPP) of each individual. EPP will be calculated average of three trials of giving electrical stimuli from a baseline of 0.5 mA in steps of 0.1 mA with interpulse intervals of 5 s until the participants reported the stimulus as painful stimuli. Then patients will be given the painful stimuli into three time intervals; T1 (5 s), T2 (5 s) and T3 (20 s). Intensity of the painful stimuli will be %150 of EPP at T1, %180 of EPP at T2 and %150 of EPP at T3. Finally, during each application (T1, T2 and T3) participants need to report their intensity of pain according to the visual analogue scale; 0 means no pain and 10 means maximum pain.94

Second, a control paradigm will be conducted with painful stimuli %150 of EPP for 30 s. It is well known that a pain reduction due to prolonged pain stimuli can be caused by the adaptation of primary afferents known to occur during prolonged stimulation, and this control paradigm was conducted to account for the adaptation. All trials will be separated by a 5 min interval in an attempt to minimise the carry-over effects on the site of the stimulation, as primary afferents have adaptive behaviours.95 104

Endogenous pain inhibition assessed by the CPM paradigm

To assess the efficacy of endogenous analgesia the CPM paradigm will be used. CPM is an established way of measuring endogenous analgesia.105–107 The cold pressor task will be used as conditioning stimulus and electrical stimulation will be used as the test stimulus. Electrical stimulation (test stimulus) will consist of 20 stimuli,108 with a variable interstimulus interval of 8–12 s, at 1.4 times EPT109 and will be applied before and during the application of the conditioning stimulus at the knee and remote site. After 20 stimulations a VNRS score is obtained on a scale from 0 (=no pain) to 10 (=worst pain possible). In order to apply the conditioning stimulus, patients will have to put their hand (the one contralateral to symptomatic leg side) up to the wrist in a cold-water bath (VersaCool, Thermo Scientific, Thermo Fisher Scientific, Waltham, MA, USA). The bath area will be filled with distilled water circulating at 12°C110 up to the wrist.

After applying the CPM paradigm at one body site, the participant is allowed to withdraw the hand from the cold-water bath for 15 consecutive minutes. Next, the CPM protocol will be repeated for the remaining anatomical location. The order of test locations will be the same as the randomised order in which EDTs and EPTs were determined. For each of the two anatomical locations (knee and remote site) the VNRS score after electrical stimulation will be recorded before and during immersion of the participant’s hand into cold water. The CPM effect will be quantified as the absolute difference in VNRS score before and during the cold pressor test.99

EEG recordings during offset analgesia and CPM

During the assessment of the offset analgesia and CPM, the nociceptive evoked potentials elicited by the contact electrical stimulus will be recorded by means of scalp EEG (Sienna Digital EEG, EMS Biomedical, Korneuburg, Austria) from 32 active Ag/Au electrodes attached on the participant’s head (Headcap, EMS Biomedical, Korneuburg, Austria). EEG will be used to measure voltage fluctuations along the scalp resulting from the ionic current flows within the neurons of the brain,111 allowing us to examine differences in brain responses and brain localisation to painful stimulation between test conditions and group allocations. Participants will be asked to close their eyes and sit as still as possible during the EEG recording procedures. Sienna EEG software will be used to set high and low bandpass filters, remove remaining artefacts and perform independent component analysis.112 The time window from −100 ms to 500 ms post-stimulation will be analysed with the accent on late evoked potentials and the maximum sampling frequency will be set at 1024 Hz. Further analyses on the averaged evoked potentials with the Brain Electrical Source Analysis software will be used to obtain 3D brain mapping with specific coordinates for each brain area. A ratio between the dorsal anterior cingulate cortex and pregenual anterior cingulate cortex/ventromedial prefrontal cortex will be calculated to reflect a balance between pain supporting and pain suppressing systems.


First, patients will be recruited in primary care. Therefore, all family physicians working within a 15 min travel distance from the participating hospitals will be contacted personally and invited to participate. When agreeing, physicians will receive recruitment leaflets which are displayed in the waiting areas. Interested patients can register by contacting the study team.

Second, patients will be recruited within the participating hospital (UZ Brussel), including the Department of Orthopaedics. Physicians can refer interested patients to the central study team and provide them an information leaflet of the study.

Finally, advertisements and announcements in local newspapers, pharmacies and (online or printed) publications from patient support groups will be used as additional recruitment strategies.

Patients will be informed that they will receive a gift voucher of €50 as travel reimbursement. Patients will receive the gift voucher of €50 in two parts: a gift voucher of €30 when they complete the 12-week intervention period, and a gift voucher of €20 after they completed the test moment at week 26. Patients who are willing to participate, will be screened by phone by a study nurse and subsequently in the hospital by a doctor for eligibility based on the predefined inclusion and exclusion criteria. During the screening by the doctor at the hospital, a blood sample will be taken to determine the hsCRP levels of the patient.

Blinding (masking)

Patients will not know which one the experimental intervention is and which one the control intervention is; however, they will of course be aware of the intervention received and whether or not they are part of the control group. Outcome assessors (LLeemans, SP) will be blinded to the maximal extent possible. About this, patients will be asked not to communicate with the assessors (LLeemans, SP) about the intervention received. The therapists providing BGA (LLie, IR) will not be involved in providing MST (EB, CDC), and vice versa. The interventions will take place at different locations on the Brussels Health Campus (Jette, Belgium) to minimise the contamination between groups, MST sessions will be organised in the UZ Brussel (Department of Physical Medicine) and BGA sessions will be held at the Erasmus Hogeschool Brussel (Jette, Belgium). Patients will not see each other in the hospital waiting rooms. Furthermore, because of the COVID-19 crisis and the associated restrictions, patients will have the opportunity to opt for tele-consults that will be held by the physiotherapists using a secured platform that allows video calling. As such, concerns regarding COVID-19 can be met, and therapy continuation will be guaranteed. The statistician will be blinded to the allocation of the treatment groups.

This post was originally published on https://bmjopen.bmj.com