Introduction and Purpose: Compression is a tactile stimulus that can reduce the perception of pain by stimulating tactile skin receptors and the speed of nerve conduction. It is highly associated with cryotherapy and other non-pharmacological physical agents without musculoskeletal pain control. However, there is still a lack of evidence on its possible effects on the modulation of this type of pain. This study will evaluate the effect of compression by elastic bandages, on pain and on the function of belonging with knee osteoarthritis (KO). Methodology: A randomized, blinded controlled clinical trial will be conducted. Individuals with KO (n = 90; both sexes; between 40 and 75 years old), will be allocated into three groups (n = 30 / group): Compression (submitted to compression by elastic bandage on the affected knee, 20 min, in 4 days consecutive); Sham (submitted to the same protocol with elastic bandage, but without compression); and Control (waiting list, without intervention). All will be taken one day before the start and one day after the last intervention. They will also be adopted in the 12th and 24th weeks after the intervention. The main outcome will be the pain intensity (Visual Analogue Scale). The Western Ontario & McMaster Universities Osteoarthritis (WOMAC) physical function questionnaire, physical function tests (step test, sit and stand test in 30s, 40m accelerated walk test), and the perception scale global change (GRC). Data analysis: SPSS 24.0 software will be used for descriptive analysis and performance of Kolmogorov-Smirnov tests, two-way ANOVA and multiple comparison tests. A 95% confidence level and a 5% significance level will be adopted.

Study design: The study consists of a randomized controlled clinical trial that meets the guidelines recommended by CONSORT (Consolidated Standards of Reporting Trials) for non-pharmacological studies TIDieR (Template for Intervention Description and Replication) and OARSI recommendations for clinical trials with patients with knee osteoarthritis . The procedures will always be carried out at the same times for each volunteer, during the intervention and evaluation days, so that the same time interval between interventions and evaluations is respected.

The evaluations will be carried out by a "blind" evaluator with no information about the identification of the group to which the volunteer belongs. The evaluator will also be trained to maintain balance in discussions and questions raised by individuals regarding the research project and the interventions or evaluations used.

The screening, evaluation and intervention will be carried out at the School Health Unit - USE, Federal University of São Carlos (UFSCar), São Carlos. It is a non-probabilistic sampling study, for convenience and intentional. Volunteers will be recruited from public announcements and waiting lists from local and regional physiotherapy, rehabilitation, orthopedics and rheumatology outpatient clinics, as well as from an existing list of volunteers diagnosed with knee osteoarthritis, available in our laboratory.

Participants: 90 individuals will participate in the study, of both sexes, aged between 40 and 75 years, diagnosed with KO according to the clinical and radiographic criteria of the American College of Rheumatology. The individuals will perform a radiographic examination of both knees, with lateral, anteroposterior, and axial views. Radiographic exams will be performed at the University Hospital, UFSCar. The screening of volunteers will be carried out by a physiotherapist specialized in the subject and experienced in the evaluation of individuals with knee osteoarthritis.

Ethical aspects: The project was initially submitted to the Ethics and Research Committee involving human beings (Plataforma Brasil), approved under Opinion Number: 3.955.692, and submitted to the registration of clinical trials (www.clinicaltrials.gov). Then, the study activities will be carried out. Volunteers will receive a verbal and written explanation of the objectives and methodology of the study, and those who agree to participate will sign an informed consent form.

Randomization: The individuals included in the study will be stratified by sex and randomly divided using the digital tool (www.randomization.com). Three groups will be randomized, with 30 individuals each: a) Compression: compression will be applied around the entire knee with KO through tensioned elastic bandages; b) Sham will also receive the application of elastic bandages around the knee with KO, but without compression; c) Control: patients on the waiting list, who will not receive intervention. To avoid selection bias, the confidential allocation method will be adopted using an opaque, non-translucent and sealed envelope. In addition, the group to which the individual belongs will only be revealed immediately before the intervention.

Sample size: The sample size was calculated preliminarily using the G * Power software (version 3.1.3; University of Trier, Germany). Two calculations were performed, the first considering pain (assessed by VAS) and the second, function (assessed by the WOMAC questionnaire). The calculation was based on the application of an F-test for the difference between three independent means (three groups). The effect size considered for this calculation, based on a previous study, was d = 0.45 for VAS, and d = 0.39 for WOMAC, which after conversion represent respectively f = 0.225 and f = 0.195). The effect sizes of f are between small and moderate, and match the rating range for the displayed d values. The level of significance was 5% and the power 95%. The calculations indicated a total of 54 individuals by the VAS and 72 individuals by the WOMAC questionnaire. The calculation to be considered will be related to the WOMAC questionnaire, with 24 individuals per group, making a total of 72 individuals. Considering a possible dropout rate of 20%, 29 participants must be allocated to each group, but to facilitate calculations and randomization, 30 participants will be allocated per group, totalling 90 individuals included in the study. A more reliable calculation will be performed after a preliminary analysis of the data collected from a pilot study.

Intervention: The elastic bandages used will be (Selecta® 13cm x160 cm, composed of 45% cotton, 20% elastodiene and 27% polyamide) involving the entire knee surface, positioned considering anatomical aspects: covering the femoral condyles and the anterior tuberosity of the tibia. The bandage will involve the knee from the distal (tibial tuberosity) to the proximal (femoral condyles), respecting the blood flow of the venous return. The level of compression was defined according to recommendations in the literature on compression interventions in lymphedema and venous changes, and should be kept between 30 mmHg and 60 mmHg. The occurrence of any sign of venous stasis (flushing and / or edema) may also indicate the need for a reduction in the level of compression or interruption of the procedure.

To standardize the level of compression presented, a previous reliability study (n = 10) was performed, with the aid of a pressure gauge (Stabilizer® - Chattanooga Group), positioned on the knee, between the patient's skin and the elastic bandage . The manometer was inflated to 40mmHg, a value indicated by the manufacturer as the resting pressure of the pneumatic bag47, and we then began to wrap the knee with the bandage. The number of turns that the tensioned elastic bandage allowed to wrap the knee was collected to reach the indicated compression range (30 mmHg ≤ x ≥ 60 mmHg), and if necessary, more than one band could be used. It was also evaluated what level of compression the Sham group would receive with the non-tensioned bandage. In both groups, the circumference of the knees was collected at three points (popliteal fossa, 10 cm above and 10 cm below) and the number of turns that were taken with the bandage.

This first step was performed in the test-retest format, with an interval of seven days, so that it was possible to calculate the intra-rater reliability level, which indicated a Kappa Coefficient48 of 0.625, considered a substantive agreement. From the reliability study, it was possible to calculate the average number of turns that should be performed both in the Compression group and in the Sham group so that the level of compression is maintained in the stipulated range. In the Compression group, all patients used a bandage, with an average of 5.7 turns (ranging from 5 to 7 turns) and the pressure level maintained at 48 mmHg (ranging from 46 to 52 mmHg). In the Sham group, everyone also used a bandage, with an average of 4 turns (ranging from 3 to 5 turns) and the pressure level maintained at 00 mmHg.

After the completion of the evaluations, the individuals of the three groups will receive face-to-face training, consisting of therapeutic exercises recommended for the treatment of knee osteoarthritis (KO), as well as a booklet with the proposed exercises, prepared by our research group, so that they can also perform the exercises without supervision. After completing the evaluations, participants in the Sham and Control groups will be able to receive the same treatment available to the Compression group, if they so request. In addition, they will remain on our list of volunteers for further studies in KO. The applications of the interventions will be performed by a physiotherapist previously trained on these procedures performed on the knee joint. The therapist will also be trained to maintain balance in explaining the study's general outlook, in discussions and questions raised by individuals about the research project and about the interventions or assessments used.

Data analysis: The registration of the data collected in the evaluations will be carried out through digital forms (Google Forms) and automatically stored in an electronic database in the Cloud (Google Drive), protected by password, to guarantee the security of the data and the participants. After completion of the collections, the data will be analysed by a "blind" biostatistician, with no information on the identification of the groups, using the SPSS 24.0 software (SPSS Inc, Chicago, IL).

The independent variables of interest in the study are group (Compression, Sham and Control) and time [pre- (assessment 1) and post-intervention (assessments 2, 3 and 4)]. The dependent variables are VAS (pain intensity), WOMAC (total score), Step test (seconds), Sit and stand test in 30 seconds (number of repetitions), and 40 m accelerated walk test (speed in m / s). The data distribution, or normality, will be tested by the Kolmogorov-Smirnov test and, according to the result, parametric or non-parametric tests will be used.

Initially, descriptive analyses will be performed, using measures of central tendency and dispersion: mean and standard deviation when following a normal distribution, and the median, minimum and maximum, when non-normal distribution. For normal data, the two-way ANOVA with mixed design will be the parametric test chosen to compare the means of the dependent variables, considering both factors simultaneously, one from repeated measurements (pre- and post-intervention, and follow up of 12 and 24 weeks) and another of independent samples (Compression, Sham and Control). If significant differences are found, tests of multiple comparisons (Post-Hoc) will be performed to assess the differences. For distribution of non-normal data, the possible reasons for non-normality, analysis of possible correction and then non-parametric tests can be applied, using both repeated (time) and non-repeated (groups) comparisons with Bonferroni adjustments or similar.

For all variables, a 95% confidence level will be determined and a significance level of 5% will be considered statistically significant. Added to this, the difference between the groups will be compared to the MDCI values defined for each variable. When the MDCI values are not available, Cohen's d coefficient will be calculated (effect size:> 0.8 large, close to 0.5 moderate and ≤ 0.2 small). Finally, to preserve the benefit of randomization, allowing for a balanced distribution of prognostic factors in the compared groups and, consequently, the observed effect, an intention-to-treat analysis will be adopted using the expectation-maximization imputation method.

Inclusion Criteria:

• Diagnosis of knee osteoarthritis according to the clinical and radiographic criteria of the American College of Rheumatology: signs of osteoarthritis in at least one of the compartments of the knee joint (tibiofemoral and / or in the patellofemoral joint); grade 2 or 3, according to Kellgren & Lawrence criteria in the radiographic examination;
• Minimum score of 04 cm on the Visual Analogue Scale (EVA, total of 10 cm);
• Accept and sign the informed consent form.

Exclusion Criteria:

• Regular practice of moderate or intense physical activity for more than 45 minutes a week;
• Having started physical activity or undergone any physical therapy treatment in the previous 3 months;
• Use of corticosteroid infiltration in the knee (previous 6 months);
• Performing previous surgery on the knee or hip;
• Presence of some clinical restriction that makes it impossible for them to participate in the proposed evaluations or interventions (cardiorespiratory, neurological, musculoskeletal, vascular changes, and / or the presence of skin lesions when applying the bandage).

The registration of the data collected in the evaluations will be carried out by means of digital forms and automatically stored in an electronic database in the Cloud, protected by password, to guarantee the security of the data and the participants. The project manager and the research assistant will regularly monitor the study data sets and make recommendations on how necessary protocol modifications or the termination of all or part of the study. The data of the participants that substantiate the results reported in this article will be shared after masking (text, tables, figures, appendices), immediately after publication. In addition, the study protocol and the clinical trial report (both with the planned statistical analysis) will be made available by the researchers who proposed the methodology. Data requests or any form of analysis should be sent to ferrari.angelicaviana@gmail.com or tania@ufscar.br. Requesters must sign a data access agreement.

Participant data that underlie the results reported in this paper will be shared after blinding (text, tables, figures, appendices), immediately following publication.

Request for data or any form of analysis should be directed to ferrari.angelicaviana@gmail.com or tania@ufscar.br. Requesters will be asked to sign a data access agreement.

The evaluations will be carried out by a "blind" evaluator with no information about the identification of the group to which the volunteer belongs.

After the completion of the collections, the data will appear by a "blind" biostatistician, with no information on the identification of the groups.