Statistical Analysis Plan (SAP) for:

The effect on knee-joint load of instruction in analgesic use compared with neuromuscular exercise in patients with knee osteoarthritis

(the EXERPHARMA trial)

Collaborators: Anders Holsgaard-Larsen1, Brian Clausen2, Jens Søndergaard3, Robin Christensen4, Thomas P Andriacchi5,6

and Ewa M Roos2

Statistical advisors: Robin Christensen4

Jonas Ranstam 1Orthopaedic Research Unit, Department of Orthopaedics and Traumatology, Odense University Hospital, Institute of Clinical Research, University of Southern Denmark, Odense, Denmark 2Research Unit for Musculoskeletal Function and Physiotherapy, Institute of Sports Science and Clinical Biomechanics, University of Southern Denmark, Odense, Denmark. 3Research Unit for General Practice, Institute of Public Health, University of Southern Denmark, Odense, Denmark. 4Musculoskeletal Statistics Unit (MSU), The Parker Institute, Department of Rheumatology, Copenhagen University Hospital, Bispebjerg and Frederiksberg, Copenhagen, Denmark. 5Departments of Mechanical Engineering and Orthopaedic Surgery, Stanford University, Stanford, California, USA. 6VA Joint Preservation Center, Palo Alto, California, USA.

Table of content Trial registration ................................................................................................................................................ 3

Study Objectives and Outcomes........................................................................................................................ 3

Descriptive outcomes .................................................................................................................................... 3

Primary Objective and Outcome ................................................................................................................... 3

Secondary Objectives and Outcomes ............................................................................................................ 4

Biomechanical outcome measures............................................................................................................ 4

Functional performance measures ............................................................................................................ 4

Exploratory Outcomes ................................................................................................................................... 6

Aerobic capacity ........................................................................................................................................ 6

Adverse events .......................................................................................................................................... 6

Drug use ..................................................................................................................................................... 6

Specification of endpoints ................................................................................................................................. 6

Primary endpoint ........................................................................................................................................... 6

Study Design ...................................................................................................................................................... 7

Sample size .................................................................................................................................................... 7

Randomization and blinding .......................................................................................................................... 7

Study Population ............................................................................................................................................... 7

Subject disposition ........................................................................................................................................ 7

Statistical analysis .............................................................................................................................................. 7

Primary endpoint ........................................................................................................................................... 7

Interpretation of results ................................................................................................................................ 7

Summary of protocol deviations ................................................................................................................... 8

Implementation of Analysis Plan ....................................................................................................................... 8

References ......................................................................................................................................................... 9

Figures ............................................................................................................................................................. 11

Figure 1: Flow diagram ................................................................................................................................ 11

Tables ............................................................................................................................................................... 12

Table 1: Adverse events .............................................................................................................................. 12

Table 2: Baseline characteristics.................................................................................................................. 13

Table 4: Outcome at 8 weeks follow up ...................................................................................................... 15

Table 5: Exploratory outcomes related to drug use and adverse events, during intervention .................. 16

Study Synopsis

Knee osteoarthritis (OA) is a mechanically driven disease, and it is suggested that medial tibiofemoral knee-joint load increases with pharmacologic pain relief, indicating that pharmacologic pain relief may be positively associated with disease progression. Treatment modalities that can both relieve pain and reduce knee-joint load would be preferable. The knee-joint load is influenced by functional alignment of the trunk, pelvis, and lower-limb segments with respect to the knee, as well as the ground-reaction force generated during movement. Neuromuscular exercise can influence knee load and decrease knee pain. It includes exercises to improve balance, muscle activation, functional alignment, and functional knee stability. The primary objective of this randomized controlled trial (RCT) is to investigate the efficacy of a NEuro-Muscular EXercise (NEMEX) therapy program, compared with optimized analgesics and anti-inflammatory drug use, on the measures of knee-joint load in people with mild to moderate medial tibiofemoral knee osteoarthritis.

Method/Design: ninety-three (93) men and women with mild-to-moderate medial knee osteoarthritis were recruited from general medical practices, in the community of Odense and Middelfart, Denmark, and from advertisements in local clubs, libraries, print media, and Facebook and randomly allocated (1:1) to one of two 8-week treatments, either (a) NEMEX therapy twice a week or (b) information on the recommended use of analgesics and anti-inflammatory drugs (acetaminophen and oral NSAIDs) via a pamphlet and video materials. The primary outcome is change in knee load during walking (the Knee Index, a composite score of the first external peak total reaction moment on the knee joint from all three planes based on 3D movement analysis) after 8 weeks of intervention. Secondary outcomes include changes in the external peak knee-adduction moment and impulse and functional performance measures, in addition to changes in self-reported pain, function, health status, and quality of life.

Trial registration ClinicalTrials.gov Identifier: NCT01638962 (July 3, 2012).

Study Objectives and Outcomes A study protocol elaborating the methods used in this study has been published [1]. All outcomes were obtained from all participants at baseline and follow-up (8 weeks) and selected additional measures at 52 weeks (Figure 1). The 8 weeks follow-up (primary endpoint) was finalized August 2015 and 52 weeks anticipated to be finalized July 2016.

Descriptive outcomes At screening, the participants filled out a questionnaire to record gender, age, weight, height, level of education, and working, smoking, and civil status, and radiographs were acquired for each participant to describe radiographic severity according to the Kellgren-Lawrence (KL) classification in grades 0 to 3, with grade 4 being an exclusion criterion [1-3].

Primary Objective and Outcome The primary objective of this study is to compare the efficacy of a specific neuromuscular exercise program with optimized analgesics and anti-inflammatory drug use on knee loads during walking, as well as pain and physical function in people with mild to moderate medial tibiofemoral knee OA.

Primary hypothesis: The first peak Knee Index (of the most symptomatic knee (study knee)) during walking will be reduced more by neuromuscular exercise than by analgesic use.

The primary outcome is change in knee load (i.e. Knee Index of study knee) during walking. The primary outcome will be between-group change in Knee Index immediately after intervention (Table 4).

The Knee Index (also known as the total knee reaction moment (TKRM)) is a surrogate for total load across both compartments, and has been chosen because changes in the external moment are known to occur in the sagittal plane before and without accompanying changes in the frontal and/or transverse plane [4-6]. Knee index is a novel functional variable, constructed from the maximal external moments affecting the knee in the frontal, sagittal, and transverse planes and will be reported normalized to height and weight and statistically adjusted for velocity.

Secondary Objectives and Outcomes Secondary hypotheses:

Additional biomechanical measures of knee-joint loading (Peak Knee Adduction Moment (KAM) and KAM impulse) of the study knee during walking will be reduced more by neuromuscular exercise compared with analgesic use.

Biomechanical measures of knee-joint loading (Knee Index, Peak Knee Adduction Moment (KAM) and KAM impulse) of the study knee during one-leg rise from a stool will be reduced more by neuromuscular exercise compared with analgesic use.

Functional performance (Maximum one-leg rises from stool, Maximum number of knee-bendings in 30s, One-leg hop for distance) of the study knee will be significantly improved by neuromuscular exercise in comparison with analgesic use.

Symptom and pain relief (KOOS subscales Pain, Symptoms, Activities of Daily Living, Sport and Recreation Function, Knee-related Quality of Life; UCLA activity score; Pain intensity level reported via text messages) will be equivalent or potentially superior in the group having had neuromuscular exercise.

These outcomes will only be supportive, explanatory and/or hypothesis generating, which is why multiplicity is not considered to be a problem [7].

Biomechanical outcome measures

The biomechanical measures are presented in Table 4 and were normalized to height and weight. All biomechanical outcome calculations were based on measurements taken during gait and one-leg rise before and after treatment by using a 3D Vicon MX movement analysis system with eight cameras operating at 100 Hz (Vicon, Oxford, UK) and two AMTI force-plates (AMTI, OR6-7, Watertown, MA, USA) embedded at floor level, operating at 1,000 Hz. A technician experienced in gait analysis and the Vicon system attached reflective markers that reflect infrared light according to the Vicon Plug-in-Gait marker set and model [8, 9]. To yield measures of external joint moments, inverse dynamics was applied by Plug-In Gait software. As also written in the protocol article [1] a hardware upgrade of the Vicon system during the inclusion period, resulted in the first 24 subjects were measured with an older version (3D Vicon MX movement-analysis system with six cameras operating at 100 Hz (Vicon)) at both baseline and after intervention. No systematic bias of within-subjects comparison was introduced since baseline and follow up measurements were performed by the same hardware.

Functional performance measures To avoid systematic bias due to a potential learning effect, the leg to be tested first was randomized at the time of each testing. To evaluate functional performance [10], we used three tests that have shown

evidence of validity in knee OA, in the following order; Maximum number of one-leg rises from a stool [11], maximum number of knee-bendings in 30 seconds [12], and one-leg hop for distance [12] (Table 4). Functional performance measures might suffer from a flooring effect (e.g. subjects not being able to perform the test correctly). A priori we decided only to report data if more than 75% of the subjects completed the tests at baseline.

Maximum number of one-leg rises from a stool

This test evaluates the number of times the participant can rise from a stool on one leg [11, 13]. The participant sat on a stool (0.48 m) and placed one foot on the floor in a self-chosen position. One familiarization attempt was allowed, and thereafter, the foot position was changed during the test. The lifted leg was held with straight knee, and the arms hang alongside the body. The test was performed with full muscle control—that is, the sitting down phase should be performed at constant speed and the rising phase is to be performed without adding arm or trunk movement. A pause of 5 minutes was provided between testing the right and left leg. If the opposite (raised) leg touched the ground, the trial was stopped, and the performed number was noted. The number of adequately performed rises was counted and noted [11, 13] (Table 4). The number of one-leg rises from a stool of the study knee is used for further analysis. Furthermore, biomechanical measures of knee-joint loading (Knee Index, Peak Knee Adduction Moment (KAM) and KAM impulse) of the study knee during one-leg rise from a stool were analyzed and used as secondary outcome measures. Only 41 of 93 participants and 39 out of 85 participants at baseline and follow-up, respectively, were able to perform the maximum number of one-leg rises from stool. Consequently, those not able to perform the test were assigned “0” since they were not able to complete one rise.

Maximum number of knee-bendings in 30 seconds

This test evaluates the ability to perform fast changes between eccentric and concentric muscle force over the knee joint. The participant stood on one leg with fingertip support of the index fingers from a bar placed in front, with the foot placed on the vertical line of a T-mark on the floor. The participant was instructed to look down on the horizontal line and bend his or her knee, without bending forward from the hip, until he or she no longer could see the horizontal line (about 30 degrees of knee flexion) consecutively for 30 seconds [12].After 10 seconds of performing practice trials the stopwatch was started after a countdown from three. For each leg, the total number of knee bends performed in 30 seconds was noted. If the opposite (raised) leg touched the ground, the trial was stopped, and the performed number noted [12] (Table 4). The number of knee-bendings in 30 seconds of the study knee is used for further analysis.

One-leg hop for distance

This test mimics sporting activities and demands explosive muscle function, balance, and functional stability of the knee. The participant stood on the leg (wearing shoes) to be tested, hops, and lands on the same limb. The hands were placed behind the back. The participant was instructed to perform a controlled, balanced landing and to keep the landing foot in place for 2 to 3 seconds until the landing position was recorded by the tester. Failure to stand still for 2 to 3 seconds results in a disqualified hop. The distance hopped was measured in centimeters from the toe at push-off to the heel where the participant landed. Participants performed one practice trial followed by three test trials. If the participant improved more than 10 cm between the second and third hops, an additional hop was performed [12, 14]. The best trial was used for further analysis and symmetry index was be calculated ((injured side/uninjured side) × 100) [14] (Table 4). The distance of the study knee is used for further analysis.

Patient-reported outcomes

The Knee injury and Osteoarthritis Outcome Score (KOOS, www.koos.nu)[15-17] questionnaire was applied to asses self-reported outcomes in five separate subscales: Pain, Other symptoms, Activities of Daily Living

(ADL), Sport and Recreation Function, and Knee-related Quality of Life (QOL), calculated as separate subscale scores ranging from 0 to 100, worst to best (Table 4). For assessment of physical-activity level, the University of California at Los Angeles (UCLA) activity score was used [18]. It assesses self-reported current activity level on a scale of 1 to 10, worst to best.

Exploratory Outcomes

Aerobic capacity Aastrands test; Before the 3D movement analysis and functional performance tests, participants performed a standardized warm-up session consisting of the Aastrands test (submaximal 1-point test on a stationary bicycle) for approximately 10 minutes. The participant cycles at 60 rpm, with the workload adjusted to target a stable heart rate of 120 to 170 bpm. From pulse rate and workload, the aerobic capacity will be estimated [19].

Adverse events

Reporting of adverse events was elicited with a non-leading questionnaire at baseline and after treatment. All events were coded according to the Medical Dictionary for Regulatory Activities, as currently required by all regulatory authorities, including the US Food and Drug Administration and the European Agency for the Evaluation of Medicinal Products [20]. In addition to the questionnaire after treatment, as previously described, adverse events during treatment were monitored by text messages.

Drug use

In both treatment groups, participants completed a drug-use diary, in which they were asked to fill in date, time, type, and amount of any given drug they used every day during the treatment period. In addition, they were asked to fill in the reason for taking the drug (for example, headache or knee pain).

Specification of endpoints

Primary endpoint The primary outcome (Knee Index of the study knee) will be analyzed in intent-to-treat (ITT) and a per-protocol (PP) analysis as appropriate: In the NEMEX group, we define the per-protocol population as those participants who participated in at least 12 exercise sessions (that is, good compliance). In the Pharma group, we define the per-protocol population as those participants who used pharmacologic pain relief (acetaminophen or equivalent dose of NSAID) for more than 2,000 mg/daily on at least 28 days (that is, good compliance) during Pharma intervention. If fewer than 12 patients from each intervention group can be categorized as having good compliance, we will dichotomize the populations and compare the halves with the better compliance. The trial is designed as a superiority trial, i.e. we expect that the group allocated to NEMEX in comparison with the Pharma-group will reduce knee index more at 8 weeks follow up. The 8 weeks follow-up was chosen as the primary endpoint, since the time immediately after intervention is anticipated to provide the largest difference in Knee Index.

Study Design

Sample size For a two-sample pooled t test of a normal mean difference with a two-sided significance level of 0.05, assuming a common standard deviation of 0.8 Nm/BW×HT% [5], a sample size of 42 participants per group is required to obtain a power of at least 80% to detect a difference between the means of first peak Knee Index of 0.5 Nm/BW×HT% (corresponding to a 27% difference in means) [5]. To make up for some attrition during the trial period, we decided to include 100 participants in total (randomized 1:1). If the drop-out rate proves to be lower, the number of recruited participants will be adjusted accordingly but will not be below 84.

Randomization and blinding Eligible participants will be randomly allocated in permuted blocks of four to six generated a priori by our statistician (RC) to either the group receiving the NEMEX therapy (NEMEX-group) or the group receiving the analgesics and NSAIDs therapy (Pharma-group). Consecutively numbered opaque envelopes will be opened after the participant has been tested at baseline (BC). The patient will be informed about the allocation immediately after baseline testing. All outcomes will be assessed at baseline and after treatment (8 weeks), and the self-reported measurements will also be collected by mail at 12 months (see Figure 1). Outcome assessments at baseline and after treatment will be carried out at the motion laboratory at the Odense University Hospital by the same assessors who will remain blinded to group allocation throughout the study. Data analysis will be done in a blinded manner by the study statistician (RC) not directly involved in the study.

Study Population

Subject disposition Study procedures, including recruitment strategies and inclusion and exclusion criteria, have been published previously in a study protocol [1]. Patient included in the trial were randomized to: A) NEMEX or B) Pharma treatment. Subjects were excluded in they openly told us that they would drop out, if they were randomized to the “wrong” intervention or if subjects generally were skeptic to the drug treatment.

Statistical analysis

Primary endpoint Data will be analyzed in a blinded manner. Main comparative analyses between groups will be performed based on the full analysis set (i.e., the intention-to-treat population; all cases with available baseline data carried forward). Between-group mean differences and 95% confidence intervals will be estimated with a general linear model in which the participant’s baseline score is entered as a covariate [21]. For binary response variables (for example, text-messaging question; drug use and adverse events), statistical significance will be tested by logistic regression analysis with treatment as the fixed-factor covariate. Sensitivity analyses performed to assess the robustness of the primary analyses will include repeated-measures and multiple-imputation analyses, which use model-based approaches for missing data.

Interpretation of results To minimize bias, we have a priori decided how to interpret the possible follow-up data scenarios: (a) If knee-joint load of the study knee is reduced more in the NEMEX group compared with the Pharma group, then NEMEX is the preferred treatment; (b) if knee-joint load of the study knee is reduced more in the

Pharma group compared with the NEMEX group, then Pharma is the preferred treatment; or (c) if knee-joint load of the study knee does not differ between the two treatment groups, the treatment associated with better outcomes in secondary outcomes including pain relief, functional improvement, and adverse events will be favored.

Summary of protocol deviations Only 41 of 93 participants and 39 of 85 participants at baseline and follow up, respectively, were able to perform the maximum number of one-leg rises from stool. As stated above, those not able to perform the test were assigned “0” since they were not able to complete one rise.

As also written in the protocol article [1] a hardware upgrade of the Vicon system during the inclusion period, resulted in the first 24 subjects were measured with an older version (3D Vicon MX movement-analysis system with six cameras operating at 100 Hz (Vicon)) at both baseline and after intervention. No systematic bias of within-subjects comparison was introduced since baseline and follow up measurements were performed by the same hardware.

The primary outcome (knee index), secondary biomechanical outcomes (Peak Knee Adduction Moment (KAM) and KAM impulse) and functional tests (Maximum one-leg rises from stool, Maximum number of knee-bendings in 30s, One-leg hop for distance) will be analyzed for the most symptomatic knee (study knee). This was not made explicit in the protocol article [1].

Implementation of Analysis Plan This SAP will be used as a work description for the statistician performing the analyses. All analyses will be performed by the same statistician and none of the investigators involved in this trial will perform any of the statistical analyses. The implementation of the SAP will be as follows: 1. A ‘data collection form’ will be outlined in a collaboration between the database manager, statistician and principal investigator (Anders Holsgaard-Larsen). 2. The database manager will code each treatment arm into ‘treatment A’ and ‘treatment B’ and thus leaving all others blinded from treatment during the analyses. 3. Blinded data will be delivered to the statistician according to the ‘data collection form’. 4. Primary, secondary and exploratory endpoint analyses will be made blinded from treatment 5. Results will be presented to the co-authors of the RCT-report any uncertainties will be clarified and blinded interpretations of the primary endpoint results will be conducted prior to unblinding of data.

References 1. Clausen, B., et al., The effect on knee-joint load of instruction in analgesic use compared with

neuromuscular exercise in patients with knee osteoarthritis: study protocol for a randomized, single-blind, controlled trial (the EXERPHARMA trial). Trials, 2014. 15(1): p. 444.

2. Altman, R.D. and G.E. Gold, Atlas of individual radiographic features in osteoarthritis, revised. Osteoarthritis Cartilage, 2007. 15 Suppl A: p. A1-56.

3. Schiphof, D., et al., Impact of different descriptions of the Kellgren and Lawrence classification criteria on the diagnosis of knee osteoarthritis. Ann Rheum Dis, 2011. 70(8): p. 1422-7.

4. Andriacchi, T.P., et al., A framework for the in vivo pathomechanics of osteoarthritis at the knee. Ann Biomed Eng, 2004. 32(3): p. 447-457.

5. Boyer, K.A., et al., Sensitivity of gait parameters to the effects of anti-inflammatory and opioid treatments in knee osteoarthritis patients. J Orthop Res, 2012. 30(7): p. 1118-24.

6. Kumar, D., K.T. Manal, and K.S. Rudolph, Knee joint loading during gait in healthy controls and individuals with knee osteoarthritis. Osteoarthritis Cartilage, 2013. 21(2): p. 298-305.

7. The European Agency for the Evaluation of Medicinal Products, CPMP. Points to consider on multiplicity issues in clinical trials. EMEA, 2002.

8. Kadaba, M.P., H.K. Ramakrishnan, and M.E. Wootten, Measurement of lower extremity kinematics during level walking. J Orthop Res, 1990. 8(3): p. 383-92.

9. Davis, R.B., et al., A Gait Analysis Data-Collection and Reduction Technique. Human Movement Science, 1991. 10(5): p. 575-587.

10. Dobson, F., et al., Measurement properties of performance-based measures to assess physical function in hip and knee osteoarthritis: a systematic review. Osteoarthritis Cartilage, 2012. 20(12): p. 1548-62.

11. Thorstensson, C.A., et al., Reduced functional performance in the lower extremity predicted radiographic knee osteoarthritis five years later. Ann Rheum Dis, 2004. 63(4): p. 402-7.

12. Bremander, A.B., L.L. Dahl, and E.M. Roos, Validity and reliability of functional performance tests in meniscectomized patients with or without knee osteoarthritis. Scand J Med Sci Sports, 2007. 17(2): p. 120-7.

13. Thorstensson, C.A., et al., The effect of eight weeks of exercise on knee adduction moment in early knee osteoarthritis--a pilot study. Osteoarthritis Cartilage, 2007. 15(10): p. 1163-70.

14. Noyes, F.R., S.D. Barber, and R.E. Mangine, Abnormal lower limb symmetry determined by function hop tests after anterior cruciate ligament rupture. Am J Sports Med, 1991. 19(5): p. 513-8.

15. Roos, E.M. and L.S. Lohmander, The Knee injury and Osteoarthritis Outcome Score (KOOS): from joint injury to osteoarthritis. Health Qual Life Outcomes, 2003. 1: p. 64.

16. Roos, E.M., et al., Knee injury and osteoarthritis outcome score (KOOS) - Development of a self-administered outcome measure. Journal of Orthopaedic & Sports Physical Therapy, 1998. 28(2): p. 88-96.

17. Roos, E.M. and Toksvig-Larsen, Knee injury and Osteoarthritis Outcome Score (KOOS) - validation and comparison to the WOMAC in total knee replacement. Health and Quality of Life Outcomes, 2003. 1(17).

18. Terwee, C.B., et al., Instruments to assess physical activity in patients with osteoarthritis of the hip or knee: a systematic review of measurement properties. Osteoarthritis Cartilage, 2011. 19(6): p. 620-33.

19. Borg, G., G. Ljunggren, and R. Ceci, The Increase of Perceived Exertion, Aches and Pain in the Legs, Heart-Rate and Blood Lactate during Exercise on a Bicycle Ergometer. European Journal of Applied Physiology and Occupational Physiology, 1985. 54(4): p. 343-349.

20. Riecke, B.F., et al., Comparing two low-energy diets for the treatment of knee osteoarthritis symptoms in obese patients: a pragmatic randomized clinical trial. Osteoarthritis Cartilage, 2010. 18(6): p. 746-54.

21. Pocock, S.J., et al., Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practice and problems. Stat Med, 2002. 21(19): p. 2917-30.

Figures

Figure 1: Flow diagram

Assessed for eligibility by general practitioner or physician (n=…)

Excluded (n=…): Failed inclusion criteria (n=…) Declined to participate (n=…) Other reasons (n=…)

Assessed for eligibility by researcher and x-ray (n=…)

Excluded (n=…): Failed inclusion criteria (n=…) Declined to participate (n=…) Other reasons (n=…)

Allocated to NEMEX-group (n=…):

Received allocated intervention (n=…) Did not receive allocated intervention (n=…):

Randomized (n=…)

Allocated to Pharma-group (n=…):

Received allocated intervention (n=…) Did not receive allocated intervention (n=…):

Baseline assessment 3D motion analysis (gait and one-leg raise from stool) Physical performance tests Patient-reported outcomes (questionnaires)

Post intervention assessment (8 weeks) 3D motion analysis (gait and one-leg rise from stool) Physical performance tests Patient-reported outcomes (questionnaires) X-ray

12-month follow-up Patient-reported outcomes (questionnaires)

Wash-out period 1 week

During intervention: - Drug use diary - Text message questions:

Pain, drug use and adverse events

Tables

Table 1: Adverse events Variable NEMEX

N= Pharma N=

Risk difference (95%CI)

Abdominal and intestinal symptoms:

Nausea Diarrhoea Constipation Wind/Flatulence Epigastric pain Vomiting Abdominal pain Heartburn Biliary symptoms Musculoskeletal symptoms: Cramps Joint paint Back pain Swollen joints Sciatic pain

Central nervous system and psychiatric symptoms:

Dizziness Headache Anxiety Sleeplessness Fatigue Mood changes Depressive tendencies

Skin and subcutaneous symptoms:

Dry skin Allergic rash Redness Eczema Perianal itching Skin irritation Urticaria

Miscellaneous symptoms:

Sensitive to cold Influenza Hair loss Bad breath Toothache

Data is presented as proportions no%; mean. Difference is estimated via the risk difference. Abbreviations: NEMEX, Neuromuscular exercise; Pharma, Pharmacological pain-relief. *P< 0.05.

Table 2: Baseline characteristics Baseline characteristics NEMEX Pharma

Women, n (%) Age (years), mean (SD) Weight (kg), mean (SD) Body Mass Index, mean (SD)

Study knee in right knee, n (%)

Radiographic knee OA severity of study knee (Kellgren-Lawrence), n (%) Grade 1 Grade 2 Grade 3 Grade 4 Living alone, n (%) College education or equivalent, n (%)

Social economic status

Working full-time or part time (%) Sick leave, n (%) Pensioner, n (%) Living alone, n (%) College education or equivalent, n (%)

Prior experience with exercise as treatment of knee OA, n (%)

Yes, with effect, n (%) Yes, without effect, n (%) No, n (%)

KOOS

Pain Symptoms ADL Sport/Rec QOL

Generic health states

EQ-5D Index

EQ VAS

SF36

Functional Performance, mean (SD)

Maximum number of one-leg rise from a stool (study knee)

Maximum number of knee-bendings in 30 seconds (study knee)

One-leg hop for distance (study knee)

Åstrand Vo2max UCLA

Biomechanical outcomes, walking

Knee Index (study knee)

Knee Adduction Moment (KAM) (study knee)

KAM impulse (study knee)

Biomechanical outcomes, one-leg rise from a stool

Knee Index (study knee) Knee Adduction Moment (KAM) (study knee)

KAM impulse (study knee)

Medication usage

Lsjdls Skljl Slj Aksj Æks Alskhf Askjf Sklh Sklf Aæksjf aæksjf

Table 4: Outcome at 8 weeks follow up Nemex

(change from baseline) Pharma (change from baseline)

Between-group difference from base-line

Mean (weeks) follow-up after start of treatment (95% CI

Number of participants with good complicance (n = %)*

Primary outcome

mean change in Knee index (study knee) (95% CI)

Secondary outcomes

KOOS subscales scores (95% CI)

Pain Symptoms ADL Sport/Rec QOL

Functional Performance, mean (95% CI)

Maximum number of one-leg rises from a stool (study knee)

Maximum number of knee-bendings in 30 seconds (study knee)

One-leg hop for distance (study knee)

Åstrand Vo2max (ml O2 bw-

1

UCLA

Biomechanical outcomes of walking (95%CI)

Knee Adduction Moment (KAM) (study knee)

KAM impulse (study knee)

Biomechanical outcomes of one-leg rise from a stool (95%CI)

Knee Index (study knee) Knee Adduction Moment (KAM) (study knee)

KAM impulse (study knee)

*NEMEX; Participants who participated in at least 12 exercise sessions. Pharma; those participants who used pharmacologic pain relief (acetaminophen or equivalent dose of NSAID) for more than 2,000 mg/daily on at least 28 days (that is, good compliance)

Table 5: Exploratory outcomes related to drug use and adverse events, during

intervention Nemex

Pharma

Between-group difference

Medication usage

Participants demonstrating compliance*, n (%)

Paracetamol, % (95%CI)$ NSAID, % (95%CI)$ Opioid, % (95%CI)$ Text-message evaluation Pain level (0-4) Drug use

Adverse events

* NEMEX group, participants who participated in at least 12 exercise sessions (that is, good compliance). Pharma group, participants who used pharmacologic pain relief (acetaminophen or equivalent dose of NSAID) for more than 2,000 mg/daily on at least 28 days (that is, good compliance). $ Average number of days participants used pain relief