A randomized replication trial of physical rehabilitation versus no physical rehabilitation (DRAW2) Authors: T. Mark-Christensen, K. Thorborg, T. Kallemose, T. Bandholm Published: 2026 ABSTRACT Total hip (THA) and knee arthroplasty (TKA) are effective surgical treatments for end-stage osteoarthritis aimed at reducing pain and improving function. Although physical rehabilitation is widely recommended after THA and TKA, its added clinical benefit beyond natural recovery remains uncertain. In the DRAW1 trial, no superiority of physical rehabilitation over no physical rehabilitation was observed. We aimed to replicate the DRAW1 trial by comparing physical rehabilitation (home-based telerehabilitation, or home-based rehabilitation) to no physical rehabilitation following THA or TKA. This was a three-arm, parallel-group randomized controlled replication trial with blinded outcome assessment. The primary outcome was the Hip Disability and Osteoarthritis Outcome Score/Knee injury and Osteoarthritis Outcome Score ADL-subscale. 169 patients (76 THA/93 TKA) were randomized to home-based telerehabilitation (n = 52, 24 THA/28 TKA), home-based rehabilitation (n = 58, 27 THA/31 TKA), or no physical rehabilitation (n = 59, 25 THA/34 TKA) for six weeks. Mean group differences between physical rehabilitation and no physical rehabilitation were − 0.5 points (95% CI − 3.1 to 2.1; p = 0.70) at 6 weeks, 0.8 points (95% CI − 1.7 to 3.4; p = 0.52). These findings replicate the DRAW1 trial, with no superiority of physical rehabilitation observed following THA or TKA compared to no physical rehabilitation. Trial registration: https://clinicaltrials.gov/study/NCT04960241 (02/07/2021).Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56195-0. Introduction Exercise-based rehabilitation is commonly prescribed following hip- and knee arthroplasty (THA and TKA) for osteoarthritis, to enhance postoperative recovery, as patients often experience substantial functional limitations in the early postoperative period, including reduced muscle strength, mobility, and ability to perform activities of daily living1,2. Clinical practice varies widely, ranging from supervised rehabilitation in outpatient centers to brief instruction followed by unsupervised home exercise3,4. However, no rehabilitation strategy has shown consistent superiority across systematic reviews1,5–7, raising questions about whether structured postoperative rehabilitation provides clinically relevant benefits beyond natural recovery. Consequently, the routine provision of structured physical rehabilitation for all patients after THA or TKA warrants critical evaluation. To investigate this, two systematic reviews were conducted comparing postoperative rehabilitation to no physical rehabilitation following THA and TKA8,9. However, the limited number of trials (5 trials with 314 participants) and their heterogeneity prevented any conclusion to be drawn - and thus highlighted the need for a pragmatic, adequately powered randomized controlled trial with real-world applicability. As a consequence, a pragmatic DRAW1 trial was designed to evaluate the fundamental effectiveness of exercise-based rehabilitation in a mixed THA and TKA population10. No superiority of exercise-based rehabilitation compared to no physical rehabilitation in terms of self-reported function or secondary outcomes was observed11. These findings challenged the rehabilitation community’s pre-conceptions because post-surgical rehabilitation is an established clinical standard in many healthcare systems and widely regarded as essential for optimizing recovery (beyond natural recovery). Given the potentially practice-changing nature of the ‘Does Rehabilitation after total hip and knee Arthroplasty Work’ trial (DRAW1) trial findings11, the DRAW2 trial was designed as a direct replication using the same protocol, setting, and design, including applied interventions as previously described in the DRAW1 trial and protocol10,11. For a detailed description of the rationale for conducting the DRAW1 and DRAW2 trials, please see the supplements. Objective To replicate the DRAW1 trial by comparing physical rehabilitation (combined intervention groups) to no physical rehabilitation following THA or TKA in terms of self-reported function, using the same superiority hypothesis. Methods This is the primary trial report for the DRAW2 replication. The trial was designed as a replication trial of the DRAW1 trial11 using the DRAW1 trial protocol10 and study site. As with DRAW1, the replication trial used a superiority, three-arm, parallel-group, randomized, controlled, trial (RCT) design with blinded outcome assessments at baseline (before intervention), at 6 weeks (end of intervention, primary endpoint), and follow-ups at 3 and 12 months. The present trial was conducted with one predefined deviation: baseline assessments were allowed to be split over two visits when deemed necessary by the assessing physiotherapist. The DRAW2 trial report adheres to the Consolidated Standards of Reporting Trials (CONSORT)12 and was approved by the Ethics Committee of the Capital Region, Denmark (ID: H-18056678)13. Due to an oversight, we initially thought the DRAW2 registration was covered by the existing DRAW1 registration (NCT03750448, 14/11/2018). However, following guidance from the Ethics Committee of the Capital Region, the trial was registered as a stand-alone trial on July 2, 2021, at ClinicalTrials.gov (NCT04960241, 02/07/2021). From the time the trial was initiated to the time it was registered as a stand-alone trial, only three participants had been included. The procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 200014. Below we summarize the protocol. For details, please see the supplements or the protocol10. Participants The inclusion criteria were: Primary, unilateral THA or TKA due to osteoarthritis.Referred to receive postoperative rehabilitation at our institution, Bornholm’s Regional Municipality.Able to speak, read, and understand Danish language.Aged 18 years or more. Primary, unilateral THA or TKA due to osteoarthritis. Primary, unilateral THA or TKA due to osteoarthritis. Referred to receive postoperative rehabilitation at our institution, Bornholm’s Regional Municipality. Referred to receive postoperative rehabilitation at our institution, Bornholm’s Regional Municipality. Able to speak, read, and understand Danish language. Able to speak, read, and understand Danish language. Aged 18 years or more. Aged 18 years or more. The exclusion criteria were: Not able to comply with exercise instructions.Discharged to a nursing home facility or receiving in-home rehabilitation by home care. Not able to comply with exercise instructions. Not able to comply with exercise instructions. Discharged to a nursing home facility or receiving in-home rehabilitation by home care. Discharged to a nursing home facility or receiving in-home rehabilitation by home care. Participants were enrolled shortly after hospital discharge, with baseline assessment and intervention initiation occurring within approximately 5–7 days after discharge and no later than two weeks postoperatively. Patients with postoperative complications requiring additional rehabilitation or institutional care were not eligible for inclusion. No explicit restrictions were applied regarding comorbidities or other coexisting conditions to reflect a real-world population. Setting and rehabilitation protocols The DRAW trials were conducted in Bornholm, Denmark, where there was interest in implementing home-based telerehabilitation as a potentially cost-effective alternative to usual care. Previous studies had shown comparable outcomes between telerehabilitation and usual care15 but had not isolated the effect of a fully self-managed intervention. Therefore, the municipality sought evidence on telerehabilitation without supervision, while the research group aimed to assess the fundamental effectiveness of physical rehabilitation independent of delivery mode. Consequently, the two rehabilitation interventions were combined in the primary analysis and compared with no physical rehabilitation in a three-arm design. All had undergone unilateral primary THA or TKA and were referred to municipal, free-of-charge postoperative rehabilitation —reflecting usual care in Eastern Denmark. In practice, hospitals send a ‘rehabilitation plan’ to the local municipality before the patient is discharged, informing them of the patient’s surgery and need for rehabilitation. Before hospital discharge, patients had achieved independent basic mobility and manageable pain levels. At the first outpatient consultation following discharge, an experienced physiotherapist provided oral and written trial information, obtained informed consent, and conducted baseline assessments prior to randomization. If consent was not given, patients were offered usual care (home-based rehabilitation). Randomization was performed by an external physiotherapist using a computer-generated sequence (1:1:1), concealed in opaque, sealed envelopes. Block randomization (block size 9) ensured balance, given the limited number of available telerehabilitation units (Fig. 1). Fig. 1CONSORT flow diagram showing participant enrollment, allocation, follow-up, and analysis across the three arms of the DRAW2 trial. aNumber of imputations in intention- to-treat analysis: 5. CONSORT flow diagram showing participant enrollment, allocation, follow-up, and analysis across the three arms of the DRAW2 trial. aNumber of imputations in intention- to-treat analysis: 5. Interventions The trial compared three post-discharge strategies: home-based physical telerehabilitation, home-based physical rehabilitation, and no physical rehabilitation, in a mixed population of THA and TKA patients. Exercise programs included in both physical rehabilitation interventions comprised of four standardized exercises tailored to THA or TKA, performed daily (3 sets of 10 repetitions) at 15 RM intensity, progressing biweekly. Exercise adherence was defined as completion of at least 80% of the prescribed exercise sessions during the 6-week intervention period. Participants below this threshold were classified as non-adherent. While the core program was identical across participants, progression was pragmatically adapted based on individual functional capacity and symptom response, reflecting usual care at Bornholm’s Regional Municipality. Full intervention details following TIDieR16 and CERT guidelines17 are publicly available via Harvard Dataverse18. For a detailed description of the rationale for the intervention types in the DRAW1 and DRAW2 trials, please see the supplements. This approach, first applied in DRAW1, was retained in DRAW2. All interventions were initiated in the first or second visit during outpatient consultations following discharge, all within two weeks of the primary surgery. All participants received identical pamphlets with information about their surgery, expected discomforts, complications, and advice on returning to daily activities. For the two rehabilitation groups, exercise instructions were included as add-ons. Physical telerehabilitation Patients randomized to this intervention were initially instructed in the exercise programs face-to-face by a physiotherapist and received a home-based exercise programme, supported by interactive virtual instructions via a mobile app using ICURA sensor technology (www.icura.dk). The system combines motion sensors that monitor exercise quality and quantity with real-time visual feedback through the app. A key feature is that physiotherapists can remotely supervise and adjust patient adherence and progression. The ICURA system is already in use across several Danish rehabilitation settings and reflects current clinical practice (Picture 1). The system is designed to be user-friendly, and patients were instructed to contact the ICURA support hotline in case of technical issues or questions. Picture 1Features of home-based telerehabilitation (ICURA). Features of home-based telerehabilitation (ICURA). Physical rehabilitation Patients in this group were initially instructed in the exercise programs face-to-face by a physiotherapist and then followed the same home-based exercise program as the telerehabilitation group, but received this only in a written version, which reflects usual care in Bornholm’s Regional Municipality. The program was developed using Exorlive templates (www.exorlive.com) and included links to short instructional videos for each exercise (Picture 2). Picture 2Home-based rehabilitation (Exorlive). Home-based rehabilitation (Exorlive). No physical rehabilitation Patients in this group received no therapeutic exercise or follow-up beyond advice to resume activities of daily living (e.g., walking, vacuuming) at their own pace, as symptoms and functional capacity permitted. The rationale for this approach and terminology is detailed in the published trial protocol10. Outcome The primary outcome was the between-group difference in the mean HOOS/KOOS ADL subscale score at 6 weeks19,20. This subscale was selected based on patient input and its clinical relevance to daily functioning, the target of the rehabilitation effort. The scores range from 0 to 100, with higher scores indicating better function. Due to their similar psychometric properties, HOOS and KOOS ADL scores were pooled for analysis21. A difference of 10 points is considered clinically important and was used as the predefined superiority margin22,23. Secondary outcomes included between-group differences at all follow-ups in HOOS/KOOS subscales for pain, symptoms, and quality of life. Secondary outcomes further included a patient global assessment of function (0–100 VAS), the 30-s chair stand test, and the 4 × 10 m fast-paced walk test, as recommended by OARSI24,25. Patient satisfaction and self-reported exercise adherence (0–100%) were recorded at 6 weeks. Additional measures included use of analgesics and walking aids, adverse events, and physiotherapist time use across groups. These outcomes were included to capture complementary domains of recovery. All outcomes were documented in medical records and transferred in collaboration with the assessing physiotherapist. Blinding Given the nature of the interventions, participants and care providers could not be blinded to group allocation, but they were blinded to the trial hypothesis to reduce expectation and response bias26,27. Outcome assessors were blinded to allocation at all time points, with baseline assessments conducted before randomization. Patients were instructed not to reveal their allocation, and assessors were trained to avoid cues or questions that could unblind them. The principal investigator, who maintained contact with participants, was not involved in outcome assessments and was aware of allocations. Sample size To compare physical rehabilitation (home-based telerehabilitation and home-based rehabilitation) to no physical rehabilitation, a clinically important difference of 10 points on the HOOS/KOOS subscale: function in daily living (ADL)22,23 was used as the superiority margin. As in the preceding DRAW1 trial, the sample size calculation was based on the primary trial objective in the combined THA/TKA population, reflecting the pre-specified estimand of overall treatment effectiveness in this mixed clinical population rather than separate procedure-specific effects. To ensure methodological comparability with DRAW1, we used the same assumptions: superiority margin of 10 HOOS/KOOS ADL points SD = 20, α = 0.05, and power = 0.80). DRAW2 was therefore independently powered using the same assumptions as DRAW1 and that pooled analyses of DRAW1 and DRAW2 were prespecified for future subgroup analyses. To account for an expected 10% loss to follow-up28, we calculated that 56 patients per group would yield a total of 168 patients (3 × 56 = 168). The trial was not powered to detect separate treatment effects in THA and TKA subgroups. Statistical methods A detailed statistical analysis plan was published as part of the trial protocol for the DRAW1 trial10 and made publicly available prior to DRAW2 data analysis. The primary analysis assessed the superiority of physical rehabilitation (home-based telerehabilitation and home-based rehabilitation combined) compared with no physical rehabilitation on the HOOS/KOOS ADL subscale at 6 weeks. Between-group differences in change from baseline were analyzed using independent two-sample t-tests or Wilcoxon rank-sum tests, depending on distributional assumptions, which were evaluated by visual inspection of quantile–quantile plots and histograms. Comparisons of each rehabilitation intervention individually versus no physical rehabilitation were conducted using the same approach. Analyses were repeated for 3- and 12-month follow-up time points. Results are presented as mean differences in change with two-sided 95% confidence intervals (CI) and corresponding p-values. Nominal outcomes were analyzed using chi-squared tests. All analyses were performed as intention-to-treat (analyzed as randomized). Missing data were imputed using multiple imputations, with imputation models including type of surgery, age, gender and prior values for the specific variable. A per protocol analysis was conducted for the primary outcome, including only participants in the rehabilitation groups who reported ≥ 80% adherence to the prescribed exercise sessions and classified as adherent. Participants with adherence below 80% were excluded from this analysis. The adherent participants were compared to the ‘no physical rehabilitation’ group. Adjustment for multiple testing was applied to secondary outcomes using Bonferroni correction. Sample difference for HOOS/KOOS ADL and proportion of independent walkers at baseline between groups were observed, see results for details. To account for possible bias, an ANCOVA model adjusting the group estimate for these two baseline variables was performed as a sensitivity analysis for the primary analysis. P-values of < 0.05 were considered statically significant. Imputation of missing data was performed in R 4.3.229 and remaining analysis in STATA version 16.130. Results Participant flow: 479 patients (221 THA / 258 TKA) were referred to Bornholm’s outpatient rehabilitation institution between the 29th of March 2021 to the 22nd of March 2024 and assessed for eligibility in the DRAW2 trial. 93 patients were not eligible, and 217 patients declined participation, leaving 169 participants to be included (inclusion rate: 43.8%) (Fig. 1). Baseline assessments were performed over two visits for 76 participants (45%). Lost to follow-up 18 patients were lost before the first follow-up at the end of the 6-week intervention period (five from the home-based telerehabilitation intervention, four from the home-based rehabilitation intervention, and nine from the no physical rehabilitation intervention). Total loss to follow-up during the entire trial period was 44 patients (26%) (12 from the home-based telerehabilitation intervention, 13 from the home-based rehabilitation intervention, and 19 from the no physical rehabilitation intervention, Fig. 1). The groups were similar on most baseline characteristics, except for a higher proportion of independent walkers (e.g., no use of a walking assistant) in the no physical rehabilitation group, 8.5% compared to 1.8% (Table 1). Additionally, some differences were observed in the sample means for baseline HOOS/KOOS ADL between the no physical rehabilitation group (55.2 points) compared to home-based telerehabilitation (52.6 points) and home-based rehabilitation (51.1 points) (Table 2). Table 1Baseline characteristics.Physical rehabilitationa (n = 110)No physical rehabilitation (n = 59)Physical rehabilitation grouped by allocationHome-based telerehabilitation (n = 52)Home-based rehabilitation (n = 58)THA/TKA (count)51 THA /59 TKA25 THA /34 TKA24 THA /28 TKA27 THA /31 TKAOperated knee (left/right, count)53/5729/3025/2528/30Women/men (count)65/4539/2028/2437/21Mean (SD)Mean (SD)Mean (SD)Mean (SD)Days from surgery to baseline assessment8.6 (3.0)8.7 (2.9)8.6 (2.9)8.7 (3.2)Age, yrs68.1 (9.0)67.6 (8.1)69.6 (7.9)66.8 (9.7)Height, cm171.0 (8.8)170.6 (8.4)172.0 (9.3)170.1 (8.2)Weight, kg88.7 (18.2)87.1 (17.5)88.1 (17.6)89.2 (18.8)BMI (kg/m2)30.3 (5.6)29.8 (4.9)29.7 (5.1)30.8 (5.9)Current use of analgesics, percentage of users (%) bParacetamol, %88.3%89.8%91.5%83.8%NSAID, %41.5%45.1%37.7%41.4%Opioids, %47.9%45.1%51.5%47.2%Neuropathic medicine, %3.7%1.7%6.9%2.4%Walking aid, percentage of users (%) bNo walking aids, %1.8%8.5%0%3.5%Use one elbow stick, %22.9%25.4%26.9%19.3%Use two elbow sticks, %64.2%61.0%59.6%68.4%Use a walker, %11.0%5.1%13.5%8.8%Primary outcome – mean (SD)HOOS/KOOS ADL subscale, points51.8 (17.5)55.2 (15.4)52.6 (17.9)51.1 (17.2)Secondary outcomes – mean (SD)HOOS/KOOS-subscalesSymptoms, points54.1 (18.1)56.5 (18.5)55.1 (16.4)53.1 (19.5)Pain, points54.7 (19.0)55.2 (16.5)55.6 (19.0)53.9 (19.0)Quality of life, points39.4 (15.4)43.7 (16.1)39.7 (16.0)39.2 (14.9)Global assessment, points45.8 (16.7)46.8 (18.5)44.2 (15.9)47.2 (17.2)30-s chair stand test (repetitions) c0 (IQR: 0 to 0)0 (IQR: 0 to 3)0 (IQR: 0 to 0)0 (IQR: 0 to 0)4 × 10 m fast-paced walk test (m/s) c0.7 (IQR: 0.5 to 0.9)0.8 (IQR: 0.6 to 1.0)0.7 (IQR: 0.5 to 1.0)0.7 (IQR: 0.5 to 0.8)Physio-time, minutes105.6 (29.1)94.4 (26.5)109.8 (30.2)101.9 (27.6)THA total hip arthroplasty, TKA total knee arthroplasty BMI Body Mass Index. NSAID non-steroidal anti-inflammatory drug, HOOS Hip disability and Osteoarthritis Outcome Score, KOOS Knee injury and Osteoarthritis Outcome Score, ADL activities of daily living.aPhysical rehabilitation = home-based telerehabilitation and home-based rehabilitation combined.bFrequencies and percentages are based on pooled estimates from the multiple imputation data sets (ITT population).cNon-normal distribution and therefore presented as median and interquartile range (IQR). Table 1Baseline characteristics.Physical rehabilitationa (n = 110)No physical rehabilitation (n = 59)Physical rehabilitation grouped by allocationHome-based telerehabilitation (n = 52)Home-based rehabilitation (n = 58)THA/TKA (count)51 THA /59 TKA25 THA /34 TKA24 THA /28 TKA27 THA /31 TKAOperated knee (left/right, count)53/5729/3025/2528/30Women/men (count)65/4539/2028/2437/21Mean (SD)Mean (SD)Mean (SD)Mean (SD)Days from surgery to baseline assessment8.6 (3.0)8.7 (2.9)8.6 (2.9)8.7 (3.2)Age, yrs68.1 (9.0)67.6 (8.1)69.6 (7.9)66.8 (9.7)Height, cm171.0 (8.8)170.6 (8.4)172.0 (9.3)170.1 (8.2)Weight, kg88.7 (18.2)87.1 (17.5)88.1 (17.6)89.2 (18.8)BMI (kg/m2)30.3 (5.6)29.8 (4.9)29.7 (5.1)30.8 (5.9)Current use of analgesics, percentage of users (%) bParacetamol, %88.3%89.8%91.5%83.8%NSAID, %41.5%45.1%37.7%41.4%Opioids, %47.9%45.1%51.5%47.2%Neuropathic medicine, %3.7%1.7%6.9%2.4%Walking aid, percentage of users (%) bNo walking aids, %1.8%8.5%0%3.5%Use one elbow stick, %22.9%25.4%26.9%19.3%Use two elbow sticks, %64.2%61.0%59.6%68.4%Use a walker, %11.0%5.1%13.5%8.8%Primary outcome – mean (SD)HOOS/KOOS ADL subscale, points51.8 (17.5)55.2 (15.4)52.6 (17.9)51.1 (17.2)Secondary outcomes – mean (SD)HOOS/KOOS-subscalesSymptoms, points54.1 (18.1)56.5 (18.5)55.1 (16.4)53.1 (19.5)Pain, points54.7 (19.0)55.2 (16.5)55.6 (19.0)53.9 (19.0)Quality of life, points39.4 (15.4)43.7 (16.1)39.7 (16.0)39.2 (14.9)Global assessment, points45.8 (16.7)46.8 (18.5)44.2 (15.9)47.2 (17.2)30-s chair stand test (repetitions) c0 (IQR: 0 to 0)0 (IQR: 0 to 3)0 (IQR: 0 to 0)0 (IQR: 0 to 0)4 × 10 m fast-paced walk test (m/s) c0.7 (IQR: 0.5 to 0.9)0.8 (IQR: 0.6 to 1.0)0.7 (IQR: 0.5 to 1.0)0.7 (IQR: 0.5 to 0.8)Physio-time, minutes105.6 (29.1)94.4 (26.5)109.8 (30.2)101.9 (27.6)THA total hip arthroplasty, TKA total knee arthroplasty BMI Body Mass Index. NSAID non-steroidal anti-inflammatory drug, HOOS Hip disability and Osteoarthritis Outcome Score, KOOS Knee injury and Osteoarthritis Outcome Score, ADL activities of daily living.aPhysical rehabilitation = home-based telerehabilitation and home-based rehabilitation combined.bFrequencies and percentages are based on pooled estimates from the multiple imputation data sets (ITT population).cNon-normal distribution and therefore presented as median and interquartile range (IQR). Baseline characteristics. THA total hip arthroplasty, TKA total knee arthroplasty BMI Body Mass Index. NSAID non-steroidal anti-inflammatory drug, HOOS Hip disability and Osteoarthritis Outcome Score, KOOS Knee injury and Osteoarthritis Outcome Score, ADL activities of daily living. aPhysical rehabilitation = home-based telerehabilitation and home-based rehabilitation combined. bFrequencies and percentages are based on pooled estimates from the multiple imputation data sets (ITT population). cNon-normal distribution and therefore presented as median and interquartile range (IQR). Table 2Changes from baseline to follow ups between physical rehabilitation and no physical rehabilitation.Follow upPhysical rehabilitationa Mean change (95%CI)No physical rehabilitation Mean change (95%CI)Mean difference between groups (95%CI)P-valuePrimary outcomeHOOS/KOOS ADL subscale, points6-week follow-up22.8 (21.3 to 24.3)23.3 (21.2 to 25.5)-0.5 (-3.1 to 2.1)0.70Secondary outcomeHOOS/KOOS: ADL, points3-months follow-up29.5 (28.0 to 31.1)28.7 (26.7 to 30.8)0.8 (-1.7 to 3.4)0.5212-months follow-up34.9 (33.3 to 36.5)34.5 (32.8 to 36.3)0.4 (-2.2 to 2.9)0.78HOOS/KOOS: Symptoms, points6-week follow-up12.8 (11.1 to 14.5)14.9 (12.7 to 17.0)-2.1 (-4.9 to 0.7)0.153-months follow-up19.8 (18.3 to 21.3)18.3 (16.0 to 20.7)1.4 (-1.3 to 4.1)0.2912 months follow-up29.7 (28.1 to 31.4)28.2 (26.1 to 30.4)1.5 (-1.2 to 4.2)0.28HOOS/KOOS: Pain, points6-week follow-up17.5 (16.0 to 19.0)19.2 (17.6 to 20.8)-1.7 (-4.1 to 0.6)0.153-months follow-up25.1 (23.5 to 26.7)26.2 (24.2 to 28.2)-1.2 (-3.8 to 1.5)0.3912-months follow-up32.9 (31.2 to 34.5)35.0 (33.2 to 36.8)-2.1 (-4.7 to 0.4)0.10HOOS/KOOS: Quality of life, points6-week follow-up19.9 (18.2 to 21.5)19.0 (17.0 to 20.9)0.0 (-1.7 to 3.5)0.503-months follow-up28.4 (26.7 to 30.1)27.0 (24.9 to 29.1)1.44 (-1.3 to 4.2)0.3012-months follow-up36.9 (35.1 to 38.8)36.8 (34.7 to 39.0)0.08 (-2.9 to 3.0)0.96Global assessment, points6-week follow-up22.3 (20.6 to 23.9)24.2 (21.9 to 26.4)-1.9 (-4.7 to 0.9)0.173-months follow-up33.2 (31.6 to 34.8)33.4 (30.9 to 35.9)-0.23 (-3.1 to 2.6)0.8712-months follow-up38.1 (36.3 to 39.8)37.3 (34.8 to 39.8)0.76 (-2.2 to 3.8)0.62Satisfaction, percentage6-week follow-up81.0 (79.6 to 82.4)80.2 (78.5 to 81.8)0.83 (-1.4 to 3.1)0.47Exercise adherence, percentage6-week follow-up80.7 (78.8 to 82.6)30-s chair stand test (repetitions) b6-week follow-up7 (IQR: 0 to 11)6 (IQR: 0 to 11)0.953-months follow-up11 (IQR: 4 to 13)9 (IQR: 5 to 12)0.0612-months follow-up12 (IQR: 8 to 14)10 (IQR: 7 to 16)0.234 × 10 m. fast paced walk test (m/s) b6-week follow-up0.5 (IQR: 0.3 to 0.6)0.4 (IQR: 0.3 to 0.6)0.023-months follow-up0.5 (IQR: 0.4 to 0.7)0.5 (IQR: 0.3 to 0.7)0.7512-months follow-up0.6 (IQR: 0.5 to 0.8)0.6 (IQR: 0.4 to 0.9)0.41Consultation time, (minutes)6-week follow-up74.1 (72.8 to 75.4)71.5 (69.5 to 73.6)2.6 (0.2 to 4.9)0.033-months follow-up61.4 (60.0 to 62.9)65.8 (64.2 to 67.4)− 4.4 (-6.6 to -2.1)0.0012-months follow-up62.7 (61.3 to 64.1)58.6 (56.8 to 60.5)4.1 (1.8 to 6.4)0.00Percentage of users of analgesics c, dParacetamol, %6-week follow-up43.3%52.2%0.013-months follow-up32.2%25.1%0.0312-months follow-up20.7%18.3%0.40NSAID, %6-week follow-up17.8%14.2%0.183-months follow-up11.5%17.3%0.0212-months follow-up5.8%5.8%0.97Opioids, %6-week follow-up7.3%3.7%0.043-months follow-up5.8%2.0%0.0112-months follow-up5.5%4.8%0.66Neuropathic medicine, %6-week follow-up2%0%0.013-months follow-up3.8%0%0.0012-months follow-up5.1%2.0%0.03Percentage of independent walkers c, d6-week follow-up70.2%75.3%0.003-months follow-up86.7%92.5%0.0012-months follow-up93.1%93.2%0.10Between-group differences in change from baseline were analyzed using independent two-sample t-tests.a Physical rehabilitation = home-based telerehabilitation and home-based rehabilitation combined.b None-normal distributed, and presented with median and interquartile range (IQR), and compared using the wilcoxon rank sum test.c The use of analgesics and numbers of independent walkers were analyzed using the Chi Squared test.d Frequencies and percentages are based on pooled estimates from the multiply imputation data sets. NSAID: non-steroidal anti-inflammatory drug. HOOS: Hip disability and Osteoarthritis Outcome Score. KOOS: Knee injury and Osteoarthritis Outcome Score. ADL: activities of daily living.Follow ups. A1: after the 6-week intervention. A2: 3 months postoperatively. A3: 12 months postoperatively. Table 2Changes from baseline to follow ups between physical rehabilitation and no physical rehabilitation.Follow upPhysical rehabilitationa Mean change (95%CI)No physical rehabilitation Mean change (95%CI)Mean difference between groups (95%CI)P-valuePrimary outcomeHOOS/KOOS ADL subscale, points6-week follow-up22.8 (21.3 to 24.3)23.3 (21.2 to 25.5)-0.5 (-3.1 to 2.1)0.70Secondary outcomeHOOS/KOOS: ADL, points3-months follow-up29.5 (28.0 to 31.1)28.7 (26.7 to 30.8)0.8 (-1.7 to 3.4)0.5212-months follow-up34.9 (33.3 to 36.5)34.5 (32.8 to 36.3)0.4 (-2.2 to 2.9)0.78HOOS/KOOS: Symptoms, points6-week follow-up12.8 (11.1 to 14.5)14.9 (12.7 to 17.0)-2.1 (-4.9 to 0.7)0.153-months follow-up19.8 (18.3 to 21.3)18.3 (16.0 to 20.7)1.4 (-1.3 to 4.1)0.2912 months follow-up29.7 (28.1 to 31.4)28.2 (26.1 to 30.4)1.5 (-1.2 to 4.2)0.28HOOS/KOOS: Pain, points6-week follow-up17.5 (16.0 to 19.0)19.2 (17.6 to 20.8)-1.7 (-4.1 to 0.6)0.153-months follow-up25.1 (23.5 to 26.7)26.2 (24.2 to 28.2)-1.2 (-3.8 to 1.5)0.3912-months follow-up32.9 (31.2 to 34.5)35.0 (33.2 to 36.8)-2.1 (-4.7 to 0.4)0.10HOOS/KOOS: Quality of life, points6-week follow-up19.9 (18.2 to 21.5)19.0 (17.0 to 20.9)0.0 (-1.7 to 3.5)0.503-months follow-up28.4 (26.7 to 30.1)27.0 (24.9 to 29.1)1.44 (-1.3 to 4.2)0.3012-months follow-up36.9 (35.1 to 38.8)36.8 (34.7 to 39.0)0.08 (-2.9 to 3.0)0.96Global assessment, points6-week follow-up22.3 (20.6 to 23.9)24.2 (21.9 to 26.4)-1.9 (-4.7 to 0.9)0.173-months follow-up33.2 (31.6 to 34.8)33.4 (30.9 to 35.9)-0.23 (-3.1 to 2.6)0.8712-months follow-up38.1 (36.3 to 39.8)37.3 (34.8 to 39.8)0.76 (-2.2 to 3.8)0.62Satisfaction, percentage6-week follow-up81.0 (79.6 to 82.4)80.2 (78.5 to 81.8)0.83 (-1.4 to 3.1)0.47Exercise adherence, percentage6-week follow-up80.7 (78.8 to 82.6)30-s chair stand test (repetitions) b6-week follow-up7 (IQR: 0 to 11)6 (IQR: 0 to 11)0.953-months follow-up11 (IQR: 4 to 13)9 (IQR: 5 to 12)0.0612-months follow-up12 (IQR: 8 to 14)10 (IQR: 7 to 16)0.234 × 10 m. fast paced walk test (m/s) b6-week follow-up0.5 (IQR: 0.3 to 0.6)0.4 (IQR: 0.3 to 0.6)0.023-months follow-up0.5 (IQR: 0.4 to 0.7)0.5 (IQR: 0.3 to 0.7)0.7512-months follow-up0.6 (IQR: 0.5 to 0.8)0.6 (IQR: 0.4 to 0.9)0.41Consultation time, (minutes)6-week follow-up74.1 (72.8 to 75.4)71.5 (69.5 to 73.6)2.6 (0.2 to 4.9)0.033-months follow-up61.4 (60.0 to 62.9)65.8 (64.2 to 67.4)− 4.4 (-6.6 to -2.1)0.0012-months follow-up62.7 (61.3 to 64.1)58.6 (56.8 to 60.5)4.1 (1.8 to 6.4)0.00Percentage of users of analgesics c, dParacetamol, %6-week follow-up43.3%52.2%0.013-months follow-up32.2%25.1%0.0312-months follow-up20.7%18.3%0.40NSAID, %6-week follow-up17.8%14.2%0.183-months follow-up11.5%17.3%0.0212-months follow-up5.8%5.8%0.97Opioids, %6-week follow-up7.3%3.7%0.043-months follow-up5.8%2.0%0.0112-months follow-up5.5%4.8%0.66Neuropathic medicine, %6-week follow-up2%0%0.013-months follow-up3.8%0%0.0012-months follow-up5.1%2.0%0.03Percentage of independent walkers c, d6-week follow-up70.2%75.3%0.003-months follow-up86.7%92.5%0.0012-months follow-up93.1%93.2%0.10Between-group differences in change from baseline were analyzed using independent two-sample t-tests.a Physical rehabilitation = home-based telerehabilitation and home-based rehabilitation combined.b None-normal distributed, and presented with median and interquartile range (IQR), and compared using the wilcoxon rank sum test.c The use of analgesics and numbers of independent walkers were analyzed using the Chi Squared test.d Frequencies and percentages are based on pooled estimates from the multiply imputation data sets. NSAID: non-steroidal anti-inflammatory drug. HOOS: Hip disability and Osteoarthritis Outcome Score. KOOS: Knee injury and Osteoarthritis Outcome Score. ADL: activities of daily living.Follow ups. A1: after the 6-week intervention. A2: 3 months postoperatively. A3: 12 months postoperatively. Changes from baseline to follow ups between physical rehabilitation and no physical rehabilitation. Between-group differences in change from baseline were analyzed using independent two-sample t-tests. a Physical rehabilitation = home-based telerehabilitation and home-based rehabilitation combined. b None-normal distributed, and presented with median and interquartile range (IQR), and compared using the wilcoxon rank sum test. c The use of analgesics and numbers of independent walkers were analyzed using the Chi Squared test. d Frequencies and percentages are based on pooled estimates from the multiply imputation data sets. NSAID: non-steroidal anti-inflammatory drug. HOOS: Hip disability and Osteoarthritis Outcome Score. KOOS: Knee injury and Osteoarthritis Outcome Score. ADL: activities of daily living. Follow ups. A1: after the 6-week intervention. A2: 3 months postoperatively. A3: 12 months postoperatively. Primary outcome: self-reported function (HOOS/KOOS ADL) Analysis: Physical rehabilitation versus no physical rehabilitation In the primary intention-to-treat analysis for DRAW2, self-reported function measured by the HOOS/KOOS improved from baseline to the 6-week (A1) assessment by 22.8 (95%CI: 21.3 to 24.3) points after physical rehabilitation and by 23.3 (95%CI: 21.2 to 25.5) points after no physical rehabilitation. The difference between groups at the 6-week follow-up was not statistically significant (mean difference of self-reported function: -0.5 points (95% CI: -3.1 to 2.1; p = 0.70), and well below the predefined 10 points superiority margin23,24. Corresponding changes in self-reported function from baseline to the later follow-ups were: 29.5 (95%CI: 28.0 to 31.1) points for the physical rehabilitation interventions and 28.7 (95%CI: 26.7 to 30.8) points for the no physical rehabilitation comparator at 3 months (A2) and 34.9 (95%CI: 33.3 to 36.5) points for the physical rehabilitation interventions and 34.5 (95%CI: 32.8 to 36.3) points for the no physical rehabilitation comparator at 12 months (A3). The between-group differences were not statistically significant at 3 months (0.8 points: 95%CI: -1.7 to 3.4, p = 0.52) or 12 months (0.4 points: 95%CI: -2.2 to 2.9, p = 0.78). Analysis: home-based telerehabilitation and home-based rehabilitation versus no physical rehabilitation The two physical rehabilitation interventions (home-based telerehabilitation and home-based rehabilitation) were compared individually to no physical rehabilitation. The self-reported function improved from baseline to 6-week follow-up (A1) by 24.1 (95%CI: 21.9 to 26.3) points for the home-based telerehabilitation intervention and 21.7 (95%CI: 19.7 to 23.7) points for the home-based rehabilitation intervention. Comparing home-based telerehabilitation to no physical rehabilitation, the between-group difference at 6 weeks (A1) was 0.7 (95%CI: -2.4 to 3.8; p = 0.65) points, while the between-group difference between the home-based rehabilitation intervention and the no physical rehabilitation comparator was − 0.1 (95%CI: -3.3 to 3.0; p = 0.94) points at the same time point. Corresponding changes in self-reported function from baseline to the later follow-ups were: 30.2 (95%CI: 27.9 to 32.4) points for the home-based telerehabilitation intervention and 29.0 (95%CI: 26.9 to 31.1) points for the home-based rehabilitation intervention at 3 months (A2). At 12 months (A3) the improvement was 34.1 (95%CI: 31.5 to 36.7) points for home-based telerehabilitation intervention and 35.6 (95%CI: 3.7 to 37.6) points for the home-based rehabilitation intervention. Comparing home-based telerehabilitation and no physical rehabilitation, the between-group difference was 1.5 (95%CI: -1.6 to 4.5; p = 0.35) points at 3 months (A2) and − 0.4 (95%CI: -3.6 to 2.7; p = 0.78) points at 12 months (A3). The between-group difference between home-based rehabilitation intervention and the no physical rehabilitation comparator was 0.3 (95%CI: -2.6 to 3.2; p = 0.85) points at 3 months (A2) and 1.1 (95%CI: -1.6 to 3.7; p = 0.42) points at 12 months (A3). Sensitivity analysis adjusting for baseline variables did lead to change in estimates and widening of CI, however, not to an extent that changed the interpretation of the result (supplement Table 4). Ancillary analyses The per-protocol analysis, restricted to patients with ≥ 80% adherence, yielded results consistent with the ITT analysis, confirming no clinically relevant between-group differences at any time point (see supplements, Table 3). Table 3Between-group differences at follow-ups by the individual physical rehabilitations (home-based telerehabilitation and home-based rehabilitation) compared to No physical rehabilitation.OutcomesMean difference (95%CI)P-valueMean difference (95%CI)P-valuePrimary outcomeHome-based telerehabilitation vs. No physical rehabilitationHome-based rehabilitation vs. No physical rehabilitationHOOS/KOOS-ADL, point difference6-week follow-up0.7 (-2.4 to 3.8)0.65-0.1 (-3.3 to 3.0)0.94Secondary outcomeHOOS/KOOS-ADL, point difference3-months follow-up1.5 (-1.6 to 4.5)0.350.3 (-2.6 to 3.2)0.8512-months follow-up-0.4 (-3.6 to 2.7)0.781.1 (-1.6 to 3.7)0.42Symptoms, point difference6-week follow-up-0.5 (-3.7 to 2.8)0.77-3.5 (-6.7 to -0.3)0.033-months follow-up1.4 (-1.9 to 4.7)0.401.5 (-1.6 to 4.6)0.3512-months follow-up1.9 (-1.2 to 5.0)0.221.1 (-2.1 to 4.3)0.49Pain, point difference6-week follow-up0.6 (-2.0 to 3.2)0.66-3.8 (-6.4 to -1.2)0.003-months follow-up-0.2 (-3.3 to 2.8)0.89-2.0 (-5.0 to 1.0)0.1912-months follow-up-1.2 (-4.1 to 1.8)0.43-3.0 (-5.8 to -0.2)0.04Quality of life, point difference6-week follow-up2.0 (-0.8 to 4.9)0.16-0.1 (-3.3 to 3.0)0.943-months follow-up2.1 (-1.1 to 5.2)0.200.9 (-2.2 to 4.0)0.5812-months follow-up3.1 (0.1 to 6.4)0.06-2.7 (-6.0 to 0.7)0.12Global assessment, point difference6-week follow-up0.4 (-2.7 to 3.5)0.82-3.9 (-7.2 to 0.7)0.023-months follow-up2.5 (0.9 to 5.8)0.15-2.7 (-6.1 to 0.8)0.1312-months follow-up1.9 (-1.8 to 5.5)0.31-0.2 (-3.6 to 3.2)0.90Satisfaction, point difference6-week follow-up4.3 (2.0 to 6.6)0.00-2.3 (5.0 to 0.4)0.1030-s chair stand test (repetitions) a6-week follow-up8 (IQR: 0 to 11) vs. 6 (IQR: 0 to 11)0.105 (IQR: 0 to 11) vs. 6 (IQR: 0 to 11)0.153-months follow-up11 (IQR: 5 to 13) vs. 9 (IQR: 5 to 12)0.0110 (IQR: 1 to 13) vs. 9 (IQR: 5 to 12)0.4212-months follow-up12 (IQR: 7 to 14) vs. 10 (IQR: 7 to 15)0.4012 (IQR: 8 to 14) vs. 10 (IQR: 7 to 15)0.224 × 10 m fast-paced walk test (m/s) a6-week follow-up0.4 (IQR: 0.3 to 0.6) vs. 0.4 (IQR: 0.3 to 0.6)0.060.4 (IQR: 0.3 to 0.6) vs. 0.4 (IQR: 0.3 to 0.6)0.023-months follow-up0.5 (IQR: 0.4 to 0.7) vs. 0.5 (IQR: 0.3 to 0.7)0.890.5 (IQR: 0.3 to 0.7) vs. 0.5 (IQR: 0.3 to 0.7)0.6812-months follow-up0.6 (IQR: 0.5 to 0.8) vs. 0.6 (IQR: 0.4 to 0.9)0.930.6 (IQR: 0.4 to 0.9) vs. 0.6 (IQR: 0.4 to 0.9)0.20Consultation time difference (minutes)6-week follow-up1.4 (1.2 to 1.6)0.233.3 (0.6 to 6.0)0.023-months follow-up-6.6 (-9.1 to -4.1)0.00-2.4 (-4.9 to 0.2)0.0712-months follow-up4.2 (2.0 to 7.4)0.003.5 (0.9 to 6.2)0.01Percentage of users of analgesics b, cParacetamol, %6-week follow-up39.6% vs. 52.2%0.0046.6% vs. 52.2%0.173-months follow-up26.2% vs. 25.1%0.7737.6% vs. 25.1%0.0012-months follow-up20.4% vs. 18.3%0.5421.0% vs. 18.3%0.41NSAID, %6-week follow-up8.1% vs. 14.2%0.0226.6% vs. 14.2%0.003-months follow-up5.8% vs. 1.3%0.0016.6% vs. 17.3%0.8112-months follow-up5.8% vs. 5.8%1.005.9% vs. 5.8%1.00Opioids, %6-week follow-up8.9% vs. 3.7%0.015.9% vs. 3.7%0.233-months follow-up4.6% vs. 2.0%0.096.9% vs. 2.0%0.0012-months follow-up4.2% vs. 4.8%0.776.6% to 4.8%0.34Neuropathic medicine, %6-week follow-up4.2% vs. 0.0%0.000.0% vs. 0.0%!3-months follow-up8.1% vs. 0.0%0.000.0% vs. 0.0%!12-months follow-up6.2% vs. 2.0%0.014.1% vs. 2.0%0.14Percentage of independent walkers6-week follow-up72.7% vs. 75,3%0.0067.9% vs. 75.3%0.013-months follow-up86.5% vs. 92.5%0.0086.9% vs. 92.5%0.0012-months follow-up95.8% vs. 93.2%0.3890.7% vs. 93.2%0.00Between-group differences in change from baseline were analyzed using independent two-sample t-tests.a30s chair stand test and 4 × 10 m fast-paced walk test were analysed with Wilcoxon sum rank-test.bThe use of analgesics and use of walking aid were analysed using the Chi Squared test.cFrequencies and percentages are based on pooled estimates from the multiply imputation data sets. ! = can not be calculated due to no events. Table 3Between-group differences at follow-ups by the individual physical rehabilitations (home-based telerehabilitation and home-based rehabilitation) compared to No physical rehabilitation.OutcomesMean difference (95%CI)P-valueMean difference (95%CI)P-valuePrimary outcomeHome-based telerehabilitation vs. No physical rehabilitationHome-based rehabilitation vs. No physical rehabilitationHOOS/KOOS-ADL, point difference6-week follow-up0.7 (-2.4 to 3.8)0.65-0.1 (-3.3 to 3.0)0.94Secondary outcomeHOOS/KOOS-ADL, point difference3-months follow-up1.5 (-1.6 to 4.5)0.350.3 (-2.6 to 3.2)0.8512-months follow-up-0.4 (-3.6 to 2.7)0.781.1 (-1.6 to 3.7)0.42Symptoms, point difference6-week follow-up-0.5 (-3.7 to 2.8)0.77-3.5 (-6.7 to -0.3)0.033-months follow-up1.4 (-1.9 to 4.7)0.401.5 (-1.6 to 4.6)0.3512-months follow-up1.9 (-1.2 to 5.0)0.221.1 (-2.1 to 4.3)0.49Pain, point difference6-week follow-up0.6 (-2.0 to 3.2)0.66-3.8 (-6.4 to -1.2)0.003-months follow-up-0.2 (-3.3 to 2.8)0.89-2.0 (-5.0 to 1.0)0.1912-months follow-up-1.2 (-4.1 to 1.8)0.43-3.0 (-5.8 to -0.2)0.04Quality of life, point difference6-week follow-up2.0 (-0.8 to 4.9)0.16-0.1 (-3.3 to 3.0)0.943-months follow-up2.1 (-1.1 to 5.2)0.200.9 (-2.2 to 4.0)0.5812-months follow-up3.1 (0.1 to 6.4)0.06-2.7 (-6.0 to 0.7)0.12Global assessment, point difference6-week follow-up0.4 (-2.7 to 3.5)0.82-3.9 (-7.2 to 0.7)0.023-months follow-up2.5 (0.9 to 5.8)0.15-2.7 (-6.1 to 0.8)0.1312-months follow-up1.9 (-1.8 to 5.5)0.31-0.2 (-3.6 to 3.2)0.90Satisfaction, point difference6-week follow-up4.3 (2.0 to 6.6)0.00-2.3 (5.0 to 0.4)0.1030-s chair stand test (repetitions) a6-week follow-up8 (IQR: 0 to 11) vs. 6 (IQR: 0 to 11)0.105 (IQR: 0 to 11) vs. 6 (IQR: 0 to 11)0.153-months follow-up11 (IQR: 5 to 13) vs. 9 (IQR: 5 to 12)0.0110 (IQR: 1 to 13) vs. 9 (IQR: 5 to 12)0.4212-months follow-up12 (IQR: 7 to 14) vs. 10 (IQR: 7 to 15)0.4012 (IQR: 8 to 14) vs. 10 (IQR: 7 to 15)0.224 × 10 m fast-paced walk test (m/s) a6-week follow-up0.4 (IQR: 0.3 to 0.6) vs. 0.4 (IQR: 0.3 to 0.6)0.060.4 (IQR: 0.3 to 0.6) vs. 0.4 (IQR: 0.3 to 0.6)0.023-months follow-up0.5 (IQR: 0.4 to 0.7) vs. 0.5 (IQR: 0.3 to 0.7)0.890.5 (IQR: 0.3 to 0.7) vs. 0.5 (IQR: 0.3 to 0.7)0.6812-months follow-up0.6 (IQR: 0.5 to 0.8) vs. 0.6 (IQR: 0.4 to 0.9)0.930.6 (IQR: 0.4 to 0.9) vs. 0.6 (IQR: 0.4 to 0.9)0.20Consultation time difference (minutes)6-week follow-up1.4 (1.2 to 1.6)0.233.3 (0.6 to 6.0)0.023-months follow-up-6.6 (-9.1 to -4.1)0.00-2.4 (-4.9 to 0.2)0.0712-months follow-up4.2 (2.0 to 7.4)0.003.5 (0.9 to 6.2)0.01Percentage of users of analgesics b, cParacetamol, %6-week follow-up39.6% vs. 52.2%0.0046.6% vs. 52.2%0.173-months follow-up26.2% vs. 25.1%0.7737.6% vs. 25.1%0.0012-months follow-up20.4% vs. 18.3%0.5421.0% vs. 18.3%0.41NSAID, %6-week follow-up8.1% vs. 14.2%0.0226.6% vs. 14.2%0.003-months follow-up5.8% vs. 1.3%0.0016.6% vs. 17.3%0.8112-months follow-up5.8% vs. 5.8%1.005.9% vs. 5.8%1.00Opioids, %6-week follow-up8.9% vs. 3.7%0.015.9% vs. 3.7%0.233-months follow-up4.6% vs. 2.0%0.096.9% vs. 2.0%0.0012-months follow-up4.2% vs. 4.8%0.776.6% to 4.8%0.34Neuropathic medicine, %6-week follow-up4.2% vs. 0.0%0.000.0% vs. 0.0%!3-months follow-up8.1% vs. 0.0%0.000.0% vs. 0.0%!12-months follow-up6.2% vs. 2.0%0.014.1% vs. 2.0%0.14Percentage of independent walkers6-week follow-up72.7% vs. 75,3%0.0067.9% vs. 75.3%0.013-months follow-up86.5% vs. 92.5%0.0086.9% vs. 92.5%0.0012-months follow-up95.8% vs. 93.2%0.3890.7% vs. 93.2%0.00Between-group differences in change from baseline were analyzed using independent two-sample t-tests.a30s chair stand test and 4 × 10 m fast-paced walk test were analysed with Wilcoxon sum rank-test.bThe use of analgesics and use of walking aid were analysed using the Chi Squared test.cFrequencies and percentages are based on pooled estimates from the multiply imputation data sets. ! = can not be calculated due to no events. Between-group differences at follow-ups by the individual physical rehabilitations (home-based telerehabilitation and home-based rehabilitation) compared to No physical rehabilitation. Between-group differences in change from baseline were analyzed using independent two-sample t-tests. a30s chair stand test and 4 × 10 m fast-paced walk test were analysed with Wilcoxon sum rank-test. bThe use of analgesics and use of walking aid were analysed using the Chi Squared test. cFrequencies and percentages are based on pooled estimates from the multiply imputation data sets. ! = can not be calculated due to no events. Harms During the trial, 13 adverse events occurred in 13 participants (Table 4); three in the home-based telerehabilitation group, six in the home-based rehabilitation group and four in the no physical rehabilitation group. Table 4Adverse events.Adverse eventsHome-based telerehabilitationHome-based rehabilitationNo physical rehabilitationBefore first follow up(6 weeks)N = 1One had PJIN = 4One had PJIOne had a dislocated hipOne had a general nerve inflammationOne was hospitalized due to bacteremiaN = 2One was a deep vein thrombosisOne was hospitalized due to suspected DVTBefore second follow up(3 months)N = 1One had MUAN = 2Two had MUAN = 1One had MUABefore third follow up(12 months)N = 0N = 1One had a general streptococcus infectionN = 1One had a cerebral hemorrhage Total N = 2N = 7N = 4PJI: periprosthetic joint infection. MUA: manipulation under anesthesia. Table 4Adverse events.Adverse eventsHome-based telerehabilitationHome-based rehabilitationNo physical rehabilitationBefore first follow up(6 weeks)N = 1One had PJIN = 4One had PJIOne had a dislocated hipOne had a general nerve inflammationOne was hospitalized due to bacteremiaN = 2One was a deep vein thrombosisOne was hospitalized due to suspected DVTBefore second follow up(3 months)N = 1One had MUAN = 2Two had MUAN = 1One had MUABefore third follow up(12 months)N = 0N = 1One had a general streptococcus infectionN = 1One had a cerebral hemorrhage Total N = 2N = 7N = 4PJI: periprosthetic joint infection. MUA: manipulation under anesthesia. Adverse events. Before first follow up (6 weeks) N = 1 One had PJI N = 4 One had PJI One had a dislocated hip One had a general nerve inflammation One was hospitalized due to bacteremia N = 2 One was a deep vein thrombosis One was hospitalized due to suspected DVT Before second follow up (3 months) N = 1 One had MUA N = 2 Two had MUA N = 1 One had MUA Before third follow up (12 months) N = 1 One had a general streptococcus infection N = 1 One had a cerebral hemorrhage PJI: periprosthetic joint infection. MUA: manipulation under anesthesia. Discussion This replication trial confirmed the main finding from the DRAW1 trial: that no superiority of home-based physical rehabilitation was observed compared to no physical rehabilitation in terms of self-reported function at 6 weeks (primary endpoint), and at 3 and 12 months (secondary endpoints). Importantly, the confidence intervals for both the intention-to-treat and per-protocol analyses excluded the predefined minimal clinically important difference of 10 points (95%CI: -3.1 to 3.4 and − 4.0 to 1.6, respectively) at any time point, supporting the robustness of this finding. Thus, the fundamental clinical effectiveness of physical rehabilitation, after THA and TKA, could not be established. Taken together, the DRAW1 and DRAW2 trials provide remarkably consistent results. In DRAW1, the between-group difference in self-reported function (HOOS/KOOS ADL) at 6 weeks was 3.3 points (95% CI: − 1.9 to 8.6; p = 0.10), and in DRAW2, the difference was 0.5 (95%CI: -3.1 to 2.1; p = 0.70). Both trials found confidence intervals that excluded the predefined minimal clinically important difference of 10 points, and both showed a lack of statistically or clinically meaningful superiority of rehabilitation over no physical rehabilitation. This replication thus strongly reinforces the robustness of the original finding. Reproducibility is foundational to trustworthy science, particularly for trials with practice-changing implications. As such, the replication of DRAW1 in DRAW2 strengthens the internal validity and credibility of both findings. Replication studies are rare in clinical rehabilitation research, and this trial contributes a rare and methodologically rigorous example—preregistered, adequately powered, and conducted at the same site using the same protocol and population. While small deviations occurred (e.g., split baseline assessments), these are unlikely to have biased the results. Some may argue that the lack of superiority observed in DRAW1 and DRAW2 is due to suboptimal intervention content—that more intensive or differently structured exercise programs might have produced different results. As researchers with backgrounds in physiotherapy, we acknowledge this perspective and have ourselves previously advocated for more intensive rehabilitation based on earlier assumptions about its efficacy31. However, both trials were deliberately designed as pragmatic and clinically grounded evaluations, using exercise protocols that align closely with current practice and national guidelines for postoperative rehabilitation following THA and TKA from Denmark and England32–34, which consists of home-based exercise following initial face-to-face instruction by a physiotherapist. Thus, the results do not reflect a complete absence of professional involvement, but rather a model in which patients continued their rehabilitation independently after initial guidance. Accordingly, the findings should be interpreted in the context of the pragmatic rehabilitation models evaluated in the DRAW trials. Importantly, no specific rehabilitation protocol (e.g. supervised vs. home-based or telerehabilitation) has consistently demonstrated superiority over others in systematic reviews and large RCTs1,5,7,35–37– including enriched trials in patients at risk for poor post-operative outcome after TKA38. Recent trials, such as Groot et al. (2025), which found that personalized treat-to-target physical therapy had no additional effect compared to usual care following THA or TKA, further support the absence of meaningful differences between rehabilitation modalities39. Similarly, recent reviews of telerehabilitation and telemedicine in fracture and arthroplasty care have generally reported non-inferiority rather than superiority compared with conventional rehabilitation approaches, further supporting the limited differential effect between rehabilitation modalities40. This lack of differential effect across intervention types reinforces the notion that the fundamental clinical effectiveness of physical rehabilitation in this population may be limited—at least within the timeframe and outcomes studied. Two systematic reviews that specifically compared postoperative rehabilitation with no physical rehabilitation following THA and TKA identified only 5 trials (n = 314 participants)8,9. Substantial heterogeneity precluded any firm conclusions regarding the effectiveness of physical rehabilitation following THA and TKA. In total, 13 adverse events occurred during the trial (three in home-based telerehabilitation group, six in the home-based rehabilitation group, and four in the no physical rehabilitation group) (Table 4). Although the number of adverse events differed slightly across groups, there was no apparent clustering, and events were relatively evenly distributed. This suggests that no single rehabilitation approach was associated with a markedly higher risk of adverse events. Although speculative, the findings from DRAW1 and DRAW2 suggest that natural recovery may account for much of the functional improvement after THA or TKA. This challenges the prevailing assumption that all patients should routinely receive structured rehabilitation. It also resonates strongly with the Choosing Wisely campaign41, which advocates avoiding low-value interventions that offer little benefit relative to their burden on patients and health systems. Applying this principle to postoperative rehabilitation suggests that structured exercise therapy should be offered selectively and focus on those with the greatest potential to benefit42. The trial also has limitations. Investigating a mixed cohort of both THA and TKA patients does not allow stratification of the results to either type of surgery alone, while the unequal group sizes may have had an influence on variance estimates. Although minor baseline differences were observed (e.g., walking aid use and slightly higher baseline scores in the no physical rehabilitation group), these were small in magnitude and unlikely to have influenced the overall findings (supplement Table 4). In conclusion, the main finding from the DRAW1 trial was replicated and no superiority of physical rehabilitation following THA or TKA was observed compared no physical rehabilitation in terms of self-reported function in the first year after surgery and rehabilitation. These findings question the default use of postoperative physical rehabilitation following THA or TKA and instead points towards involving patients in shared-decision making around their needs, options and preferences in relation to post-surgical rehabilitation. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1