Knee-extensor strength, symptoms, and need for surgery after two, four, or six exercise sessions/week using a home-based one -exercise program: A randomized dose-response trial of knee-extensor resistance 3 exercise in patients for replacement (the

eligible knee QUADX-1 trial). To investigate firstly the efficacy of three different dosages of one home-based, knee-extensor resistance 33 exercise on knee-extensor strength in patients eligible for knee replacement, and secondly, the influence of 34 exercise on symptoms, physical function and decision on surgery.


Introduction
Trial amendments 111 Due to an oversight, the second research question and purpose were not pre-registered. Hence, we 112 consider them secondary and exploratory. All other trial amendments are reported in the trial protocol. 35  Following baseline assessment, the patients were referred to a physiotherapist in their local municipal 128 rehabilitation setting. Here the patients were instructed how to perform a single knee-extensor resistance 129 exercise at home. The knee-extensor resistance exercise was performed sitting on a chair with an exercise 130 band wrapped around the ankle and fixed behind a door for resistance. Patients were provided with a 131 personal exercise band for exercising at home and a brochure with instructional notes and illustrations. The 132 patients were randomized to one of three exercise dosage groups for twelve weeks: the two sessions/week 133 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. ; https://doi.org/10.1101/2021.04.07.21254965 doi: medRxiv preprint group, the four sessions/week group or the six sessions/week group. For all groups, training comprised only 134 the single knee-extensor resistance exercise. Patients were instructed to perform the exercise in three sets 135 of twelve repetitions with each repetition lasting eight seconds (concentric phase 3 s, isometric phase 1 s, 136 eccentric phase 4 s). The intervention was personalized to the extent where each patient was exercising 137 with an individual absolute resistance corresponding to a relative load of twelve repetition maximum (RM). 138 The patients were instructed to continue until volitional muscular failure. That is, until the knee-extensor 139 muscles were maximally fatigued, and they were not able to perform further repetitions. If volitional 140 muscular failure occurred before twelve RM, the resistance of the elastic band was adjusted so that the 141 pre-determined number of repetitions could be completed (decrease in distance between the two 142 endpoints of the elastic band). Whenever the resistance in the elastic band became too low (i.e., more than 143 twelve repetitions per set could be performed), the patients were instructed to increase the resistance in 144 the elastic band to achieve a new resistance corresponding to a relative load of twelve RM (increase in 145 distance between the two endpoints of the elastic band). Detailed intervention description can be found in 146 the trial protocol 35 and a walkthrough video of the exercise is freely available online (https://bit.ly/3i59CJn). 147 148 Assessments and outcomes 149 Outcomes were assessed: at baseline (t0), after twelve weeks of home-based exercise/before surgery (t1), 150 at hospital discharge (1-8 days after surgery) (t2) and three months after surgery (t3). Outcomes at 151 endpoints t2 and t3 were only collected for patients that underwent surgery. The primary endpoint was 152 after the exercise period (t1) and the secondary endpoints were just before hospital discharge (t2) and three 153 months after surgery (t3). After the 12-week exercise period, at endpoint t1, each patient's decision on 154 surgery was re-evaluated in a shared decision-making process between the patient and orthopedic surgeon 155 (i.e. continue with exercise therapy or schedule knee replacement). Outcome assessments were performed 156 blinded by the primary investigator and a research assistant dedicated to the trial. Primary outcome 159 The primary outcome was change in isometric knee-extensor strength from baseline to after the exercise 160 period (t0-t1). Isometric knee-extensor strength was measured using a computerized strength chair (Good 161 Strength Chair, Metitur Oy, Jyvaskyla, Finland), which is valid and reliable in the knee replacement 162 population. 40 Five measurements of maximal isometric knee-extensor strength at 60° knee flexion were 163 completed, separated by 60-s pauses. The patients were instructed to extend their knee as forcefully as 164 possible with a gradual increase in force over a 5-s period while receiving strong standardized verbal 165 encouragement. Isometric knee-extensor strength is expressed as the maximal voluntary torque per 166 kilogram body mass (Nm/kg). The highest obtained value was used for analysis. 167 168 Secondary outcomes 169 The secondary outcomes were change in performance-based function comprising six-minute walk test 170 (6MWT) and stair climb test (SCT), self-reported disability; Knee injury and Osteoarthritis Outcome Score 171 (KOOS), Oxford Knee Score (OKS), current knee pain and average knee pain during the last week (0-10 NRS), 172 "need for surgery" and objectively measured exercise adherence (t0-t1, t0-t2 and t0-t3). Other outcomes 173 were registration of adverse events and harms. 174 The "need for surgery" outcome was an assessment of the patients' self-perceived need for surgery. After 175 the 12-week exercise period at outcome assessment t1 the patients were asked by the outcome assessor: is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The patients were randomly assigned by a 1:1:1 allocation ratio. The random allocation sequence was 197 computer-generated using simple (unrestricted) randomization by a statistician otherwise not involved in 198 the trial. One hundred and forty sequentially numbered sealed opaque envelopes were generated. When a 199 patient was included in the trial a research assistant independent of the trial opened an envelope and 200 informed the patient's municipality of the exercise group allocation. 201 202 Blinding 203 All outcome assessors and the data analysts were blinded to the exercise group allocation. At outcome 204 assessments the assessors started by informing the patients not to mention their exercise dosage. For 205 analysis, the data was coded to conceal group allocation, blinding the data assessors and analysts to the 206 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The primary intention-to-treat superiority analysis tested the hypothesis that an exercise dosage of four 212 knee-extensor resistance exercise sessions per week would elicit a greater change in isometric knee-213 extensor strength pre-operatively compared to two or six sessions per week. For all outcomes, between 214 group contrasts were compared using analysis of variance (one-way ANOVA). Normality assumptions of the 215 model residuals were checked to ensure that the underlying assumptions of the statistical model were met. 216 Normal distribution of data was checked by q-q plots and histograms. Analyses were adjusted for the 217 following baseline variables: isometric knee-extensor strength, KOOS symptoms, KOOS ADL, KOOS sport 218 and 6MWT. These adjustments were not prespecified. In a secondary analysis, the two sessions/week and 219 six sessions/week groups were compared and follow the same principles as the primary analysis. As 220 supplementary analyses, simple regression models were performed using the pooled exercise adherence 221 data across all three groups. The dependent variables were the primary and secondary outcomes and the 222 independent variable was exercise adherence quantified in two ways: 1) as total number of completed 223 exercise sessions and 2) as total time-under-tension (TUT) per patient. All analyses followed the ITT 224 principle and to create full datasets, missing data were imputed using multiple imputation (100 imputation 225 sets). Multiple imputation models were based on age, gender, group allocation and all previous scores in 226 relevant outcomes. Missing data break down is presented in Supplement 1. All analyses followed the pre-227 specified analysis plan 35 and were performed in SAS Enterprise Guide 7.1. 228 229 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint were assessed for eligibility. One-hundred and forty patients were included and randomized (Figure 1) is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Primary outcome: Intention-to-treat analysis did not find statistically significant differences between the 247 groups in change between baseline and following 12 weeks of exercise (primary endpoint (t0-t1)) in 248 isometric knee-extensor strength: two sessions/week group vs. four sessions/week group; 0.003 Nm/kg 249 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  Due to the large proportion of patients who postponed surgery after the exercise intervention, only 32 259 patients were available for the post-operative intention-to-treat analyses. No between group differences 260 for any outcomes were observed at these endpoints (Supplement 5). 261 262 Table 2. Mean change in all outcomes between baseline and following 12 weeks home-based exercise (t0-t1). Intention-to-treat analysis, N = 140. One-way ANOVA based on imputed data. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Data presented with mean change value and corresponding 95% confidence interval. Analyses were adjusted for the following baseline scores: isometric knee-extensor strength, KOOS symptoms, KOOS ADL, KOOS sport and 6MWT. These adjustments were not prespecified. ¥ = unadjusted numbers. Isometric knee-extensor strength reported as Nm/kg (positive change = improvement); Knee injury and Osteoarthritis Outcome Score (KOOS) subscale reported on 0-100 scale (positive change = improvement); Oxford Knee Score (OKS) reported on 0-48 scale (positive change = improvement); Pain scores reported on Numeric Raring Scales (NRS 0-10) (negative change = improvement); Six-minute walk test (6MWT) reported in meters (positive change = improvement); Star climb test (SCT) reported in seconds (negative change= improvement); The "need for surgery" outcome was an assessment of the patients self-perceived need for surgery. After the 12-week exercise period at outcome assessment t1 the patients were asked by the outcome assessor: "Based on your knee symptoms in the last week would you say that you need knee surgery?". "Need for surgery" data presented as N and corresponding %. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. ; https://doi.org/10.1101/2021.04.07.21254965 doi: medRxiv preprint (67.5%) postponed surgery, 32 (27.4%) underwent knee replacement, and 6 (5.1%) wanted surgery, but the 294 orthopedic surgeon deemed this contra-indicated due to co-morbidities (  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint likely respond differently to knee-focused exercise due to their condition and associated impairments (e.g. 321 arthrogenic muscle inhibition ) 47 -something which could interfere with the exercise dose-response 322 relationship classically seen in healthy people. 48 The results suggest that patients eligible for knee 323 replacement increase their knee-extensor strength equally when exercising with large or small dosages. 324 This is supported by the result from our recent meta-regression analysis, in which we found no relationship 325 between knee-extensor resistance exercise dosage and change in knee-extensor strength in patients 326 eligible for knee replacement (meta-regression was completed after initiation of the QUADX-1 trial). 49 327 Patients with knee OA might not need large exercise dosages to improve muscle strength -something also 328 suggested in the recent START trial. 45 In the START trial, high-intensity strength training was not superior to 329 low-intensity strength training, nor to an attention-control in knee OA. 45 It suggests that a classic exercise-330 dose-response relationship may not exist in knee OA, and that some of the effect, believed to be exercise-331 specific, may in fact be caused by "unspecific" or "contextual" factors. 45,50 332 Another factor that may contribute to our finding of no dose-response relationship is adherence to the 333 prescribed dosages. As seen in figure 3 there is some overlap between the completed exercise sessions 334 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. ; https://doi.org/10.1101/2021.04.07.21254965 doi: medRxiv preprint 1.8 and -1.1 changes in avg. knee pain last week (NRS 0-10), respectively, while the six sessions/week did 358 not. 56 Finally, for the SCT no MCID is known while the minimal detectable change is reported to be 2.6 359 seconds. 57 No groups reached this for neither the up or down stair climbing assessment. 360 In general, the two and four sessions/week groups reached the MCID for the outcomes more often than 361 the six sessions/week group. This could be explained by the larger exercise dose with more frequent 362 sessions leaving less time to recover between sessions -something that could lead to increase in knee pain 363 and decreased physical function. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022.  table 4 showing that patients who believe they need surgery, undergo surgery, while those 394 who "don't know" or do not believe they need surgery postpone it. 395 In the Enhanced Recovery After Surgery (ERAS) concept it is assumed that exercise therapy before planned 396 surgery (exercise-based pre-habilitation) is always followed by surgery. [69][70][71][72][73] We have previously argued that 397 the premise for exercise therapy before potential surgery -to enhance post-surgical outcomes in patients 398 eligible for knee replacement -should be questioned 49 as several systematic reviews conclude no clinically-399 relevant effect post-operatively. 49,74-83 Instead of being a predetermined care pathway (leading to surgery), 400 exercise therapy before potential surgery could be used to inform the shared decision-making process 401 when planning a care pathway, 8,17,18 which complies with guideline recommendations while being cost-402 effective. 5-10,84 Based on the results from the QUADX-1 trial, we suggest using simple (one exercise) home-403 based resistance exercise therapy within the ERAS concept to "pre-evaluate" the need for surgery in 404 patients with severe knee OA rather than to "prepare" patients for surgery. 405 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint   is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Mean change between group two and six sessions/week for all outcomes between baseline and following 12 weeks home-based exercise (t0-t1). Intention-to-treat analysis, N = 140. One-way ANOVA based on imputed data.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. ; https://doi.org/10.1101/2021.04.07.21254965 doi: medRxiv preprint Supplement 5. Results for the post-surgical assessments (t2 and t3) 723 724 Primary analyses (t0-t2) 725 eTable 7. Mean change in all outcomes between baseline and acutely after surgery (t0-t2). Intention-to-treat analysis, N = 32. One-way ANOVA based on imputed data.
Mean change (95% CI) from baseline within groups (effect = time) Mean change (95% CI) from baseline between groups (effect = time*group) Data presented with mean change value and corresponding 95% confidence interval. Analyses were adjusted for the following baseline scores: isometric kneeextensor strength, KOOS symptoms, KOOS ADL, KOOS sport and 6MWT. These adjustments were not prespecified. ¥ = unadjusted numbers. Isometric knee-extensor strength reported as Nm/kg (positive change = improvement); Pain scores reported on Numeric Raring Scales (NRS 0-10) (negative change = improvement). Too few observations to run the One-way ANOVA model (model did not converge) for the outcomes: Six-minute walk test (6MWT) and Star climb test (SCT). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. ; https://doi.org/10.1101/2021.04.07.21254965 doi: medRxiv preprint Secondary analyses (Mean change between group two and six sessions/week) 736 eTable 8. Mean change between group two and six sessions/week for all outcomes between baseline and acutely after surgery (t0-t2). Intention-to-treat analysis, N = 32. One-way ANOVA based on imputed data. Data presented with mean change value and corresponding 95% confidence interval. Analyses were adjusted for the following baseline scores: isometric knee-extensor strength, KOOS symptoms, KOOS ADL, KOOS sport and 6MWT. These adjustments were not prespecified. ¥ = unadjusted numbers. Isometric knee-extensor strength reported as Nm/kg (positive change = improvement); Pain scores reported on Numeric Raring Scales (NRS 0-10) (negative change = improvement). Too few observations to run the One-way ANOVA model (model did not converge) for the outcomes: Six-minute walk test (6MWT) and Star climb test (SCT).

737
Primary analyses (t0-t3) 738 eTable 9. Mean change in all outcomes between baseline and three months after surgery (t0-t3). Intention-to-treat analysis, N = 32. One-way ANOVA based on imputed data. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted February 26, 2022. ; https://doi.org/10.1101/2021.04.07.21254965 doi: medRxiv preprint