Swedish massage versus hip strengthening exercises for pain and function in older adults with knee osteoarthritis: a randomized controlled trial

Abstract

Background

Knee osteoarthritis (KOA) is a progressive disease that impairs mobility and quality of life in older adults.

Aims

To compare the effects of Swedish massage (SM) and hip strengthening exercises (HSE) versus control on pain, range of motion (ROM), and function in older adults with KOA.

Methods

Seventy-five adults > 60 years with symptomatic KOA were randomized to SM, HSE, or control for 8 weeks of home-based intervention (3 sessions/week, 30 min each). Outcomes included pain, function, and ROM at baseline and week 8. Intention-to-treat analysis used ANCOVA adjusted for baseline values, with Bonferroni correction for two pre-specified comparisons (α = 0.025).

Results

Seventy participants completed the study (93.3%); adherence exceeded 85%. No serious adverse events occurred. Both active interventions significantly outperformed control across all outcomes. Compared with control, SM reduced VAS pain by an adjusted mean of 0.81 cm (95% CI 0.38–1.24, p < 0.001, d = 0.69) and HSE by 0.77 cm (95% CI 0.34–1.20, p < 0.001, d = 0.65). KOOS-ADL improved by 3.59 points with SM (95% CI 1.62–5.56, p < 0.001, d = 0.71) and 3.40 points with HSE (95% CI 1.43–5.37, p = 0.001, d = 0.67). Active knee flexion ROM increased by 3.42° (SM, p = 0.001, d = 0.73) and 3.69° (HSE, p < 0.001, d = 0.77) vs. control.

Discussion

This study shows SM and HSE as safe, feasible home-based options for pain relief in frail older adults with KOA.

Conclusions

SM and HSE mitigate KOA pain, with SM uniquely enhancing daily function, supporting integration into clinical practice to promote independence and reduce healthcare burdens in aging populations.

Clinical trial registration

IRCT20150519022320N33 (August 12, 2023).

Supplementary Information

The online version contains supplementary material available at 10.1007/s40520-025-03313-z.

Introduction

Knee osteoarthritis (KOA), the most common form of osteoarthritis, is a leading cause of mobility limitations in older adults [1]. It affects over 650 million individuals worldwide [2]. This condition is more prevalent in older adults and women [3, 4], with approximately 18% of women and 10% of men over 60 years experiencing symptomatic KOA [5]. Women are 1.7 times more likely to develop KOA than men [2].

KOA in older adults not only causes knee pain and stiffness but also compromises independence and quality of life. Primary symptoms include pain, reduced range of motion (ROM), muscle weakness, and functional limitations [6], which increase the risk of falls and dependency [1, 7]. These challenges are exacerbated in older adults by comorbidities such as osteoporosis or balance impairments, which can elevate healthcare costs and the risk of fall-related fractures [8].

The primary goals of KOA treatment include pain relief, preservation of joint function, and improvement of quality of life [9, 10]. Pharmacological treatments, such as non-steroidal anti-inflammatory drugs (NSAIDs) and intra-articular hyaluronic acid injections [1], are less recommended in older adults due to side effects like renal complications [9]. Consequently, clinical guidelines increasingly favor non-pharmacological interventions, such as massage and exercise, which offer longer-term symptom relief [11].

Non-pharmacological approaches are critical for managing KOA in older adults [7, 12]. Swedish massage (SM), a widely used complementary therapy, improves blood flow and reduces muscle tension, thereby alleviating knee pain and stiffness in KOA patients [13, 14]. Recent studies suggest that SM may improve daily functioning in these patients [15, 16].

However, access to trained practitioners may be limited for low-income or rural-dwelling older adults [17], rendering this approach impractical for some.

Exercise offers a practical alternative for KOA management [11]. Hip strengthening exercises (HSE), due to their role in enhancing muscle strength, which is closely linked to KOA progression, are effective in managing this condition [18, 19]. Recent evidence indicates that HSE can reduce pain and improve physical function in patients with KOA [18, 20]. Although exercise programs may be challenging for older adults due to general weakness or balance impairments, they underscore the need for tailored programs [21]. Nevertheless, home-based exercises are particularly appealing for older adults due to their convenience and low cost [22].

Direct comparison of these interventions is essential to guide clinical practice [23]. Despite the reported benefits of SM and HSE, no studies have directly compared their effects in older adults with KOA. A systematic search of PubMed and Scopus (up to September 2025) identified no clinical trials comparing these interventions, highlighting a significant research gap. Therefore, this study aimed to compare the effects of Swedish massage and hip strengthening exercises on pain, range of motion, and daily function in older adults with knee osteoarthritis.

Materials and methods

The research design and the protocol used in the research were registered in the Iranian Registry of Clinical Trials (https://trialsearch.who.int/), with the identifier IRCT20150519022320N33.

Participants

A total of 710 older adults with knee osteoarthritis (KOA) referred to an outpatient clinic in Rafsanjan, Iran, were screened for eligibility, of whom 75 were enrolled and randomized. Inclusion criteria included age > 60 years, diagnosis of KOA according to American College of Rheumatology (ACR) criteria [24], Visual Analog Scale (VAS) pain score of 40–70 mm, chronic pain duration of 6 months to 5 years, and absence of autoimmune diseases, skin infections, prior knee surgery, severe frailty (defined as requiring no assistance for basic daily activities), or use of complementary therapies in the prior 6 months. Exclusion criteria applied during screening (prior to enrollment and randomization) included any conditions that would preclude safe participation, such as uncontrolled comorbidities or contraindications to the study interventions.

The high number of screened patients relative to enrollees resulted from stringent inclusion criteria, declining participation due to the time commitment, and preference for other treatments designed to create a homogeneous sample and ensure participant safety.

Post-randomization, some participants experienced symptom exacerbation or voluntary withdrawal. Adherence to the intervention protocol was also monitored. Crucially, for the primary intention-to-treat analysis, all randomized participants (n = 75) were included, regardless of their adherence status or post-randomization discontinuations. This approach preserves the integrity of the randomization and minimizes bias. Additional per-protocol analyses, which included only participants who completed the intervention with adequate adherence, were also conducted as secondary analyses.

The sample size was calculated a priori based on the data reported by Sritoomma et al. [25] for the primary outcome of pain measured using the Visual Analogue Scale (VAS). Using the mean pain scores and standard deviations reported in that study at a comparable follow-up point, we estimated a standard deviation of 0.99 in the intervention (massage) group and 1.15 in the control group. We considered a minimally clinically important difference in pain score of 0.9 units to be detectable (µ1-µ2 = 0.09), with α = 0.05 (two-tailed) and power = 80%. Assuming equal allocation to three groups (two intervention arms and one control arm) and using the larger standard deviation (1.15) for a conservative estimate in a two-group comparison formula (which yields the most conservative sample size when extended to multiple groups), the required sample size was 23 participants per group. Adjusting for an anticipated 10% dropout rate, the final target sample size was increased to 25 participants per group, resulting in a total planned sample size of 75 patients (25 per group).

Experimental design

This randomized clinical trial was registered with the Iranian Registry of Clinical Trials (IRCT20150519022320N33). Ethical approval was obtained from the Rafsanjan University of Medical Sciences Ethics Committee (IR.RUMS.REC.1402.068). Written informed consent was obtained from all participants. Randomization was generated by an independent statistician using Random Allocation Software and implemented with sequentially numbered, opaque, sealed envelopes (SNOSE). We used stratified block randomization with three pre-specified stratification factors (age ≤ 70/>70 years; BMI ≤ 25/>25 kg/m2; baseline VAS ≤ 55/>55), producing 2 × 2 × 2 = 8 strata. Block size was 6 (fixed), chosen to maintain balance within strata given our target sample and to reduce predictability; allocation concealment was maintained by central preparation of envelopes and assignment by study personnel uninvolved in sequence generation. Five participants were lost to follow-up (SM n = 2, HSE n = 2, control n = 1). The distribution of stratification variables remained comparable across arms (see CONSORT Figure and baseline Table 1). All randomized participants were analyzed under the prespecified ITT framework, and sensitivity.

Table 1.

Demographic and clinical characteristics of participants by study group

Variable	SM (n = 25)	HSE (n = 25)	Control (n = 25)	Total (n = 75)
Age (years)	63.52 ± 3.64	64.87 ± 3.46	63.63 ± 2.39	64.00 ± 3.21
Weight (kg)	73.43 ± 11.38	74.43 ± 7.37	73.08 ± 7.47	73.64 ± 8.81
BMI (kg/m²),	27.32 ± 2.09	27.31 ± 2.04	27.30 ± 2.24	27.31 ± 2.10
Disease Duration (months)	30.26 ± 11.36	31.83 ± 15.18	29.50 ± 16.20	30.51 ± 14.25
Pain Duration (months)	21.39 ± 8.83	21.39 ± 10.82	20.50 ± 9.00	21.08 ± 9.45
Fried Frailty Phenotype	1.80 ± 0.70	1.90 ± 0.60	0.80 ± 1.70	1.50 ± 1.17
Charlson Comorbidity Index (CCI),	3.31 ± 0.47	3.28 ± 0.46	3.41 ± 0.50	3.33 ± 0.47
Gender
Male	5 (21.7)	6 (26.1)	6 (25.0)	17 (24.3)
Female	18 (78.3)	17 (73.9)	18 (75.0)	53 (75.7)
Job
Housekeeper	15 (65.2)	16 (69.6)	16 (66.7)	47 (67.7)
Retired	8 (34.8)	7 (30.4)	8 (33.3)	23 (32.9)
Knee
Left	14 (60.9)	12 (52.2)	11 (45.8)	37 (52.8)
Right	9 (39.1)	11 (47.8)	13 (54.2)	33 (47.1)

Table shows baseline characteristics of all 75 randomized participants (intention-to-treat population). Data are mean ± standard deviation or number (%); Between-group differences were not tested statistically, in accordance with CONSORT guidelines; the table is provided for visual inspection of balance across groups.”

SM: Swedish massage; HSE: Hip Strengthening Exercises

To reduce detection bias, the research personnel who collected data, the rheumatologist performing the ROM measurements, and the statisticians conducting the data analyses were blinded to group allocation. Participant blinding was not feasible due to the nature of the interventions, and a sham intervention was not used to avoid complexity in older adults, which is addressed in the Discussion section.

Training protocol

All interventions were designed and reported according to the TIDieR (Template for Intervention Description and Replication) checklist (completed checklist is provided in Supplementary File 1). Detailed protocols, including step-by-step instructions, progress measures, and educational materials, are referenced in the text of the article.

Before beginning the intervention, members of the SM and HSE groups attended two 45-minute face-to-face training sessions and one 30-minute practice session, with initial supervision by a certified researcher (Iranian Physiotherapy Association). Each participant also attended the sessions with their home caregiver, who was also trained to supervise the participant’s implementation of the interventions at home. The interventions protocol was taught by the researcher utilizing instructional booklets, which displayed pictures on one side and detailed explanations on the other, supplemented by educational videos demonstrating correct movement techniques. This methodology aimed to facilitate understanding and ensure proper techniques execution. All participants practiced the exercises during the orientation session and received an individual performance evaluation. At the end of the training sessions, the correct performance of the interventions was assessed individually for each elderly person. Finally, a qualification verification process, conducted by a certified researcher, ensures that participants can perform the techniques safely and effectively at home, thereby ensuring the authenticity of the treatment.

After completing the training and practice sessions, participants in the SM and HSE intervention groups performed their respective techniques for three 30-minute sessions per week, for 8 weeks, at home, following the protocol.

The 30-minute SM protocol is divided into three parts: pre-SM care (5-minute), SM Techniques and Application (20-minute), and post-SM care (5-minute).

The pre-SM care was a five-minute warm-up and consisting of gentle shoulder stretches (with rotational movements), arm stretches (by bending and extending the elbows), and wrist stretches (by rotating the wrists in a circular motion).

The main SM protocol was to perform three Swedish massage techniques that each participant applied to themselves, which included effleurage (Long, gentle strokes to the muscles around the knee joint that run towards the heart along the abdominal muscles, using light to medium pressure for 5 to 7 min), petrissage (Kneading and squeezing the muscle tissue around the knee joint between the fingers and thumb, focusing on areas of tension for 5 to 7 min), and tapotement (a percussion-based massage where hands repeatedly, rhythmically, and lightly strike the muscle tissue around the knee joint, such as hacking or pounding movements) [16, 26, 27]. These techniques were performed around the knee joint, with special emphasis on the quadriceps, hamstrings, and calf muscles.

The post-SM care was a five-minute cool-down and consisting of rest and drink Enough of fluids.

Participants were instructed to sit upright in a firm, straight-backed chair with both feet flat on the floor and knees flexed at approximately 90 degrees. Clothing was adjusted to fully expose the knees and the surrounding area. Prior to massage, 3–5 mL of a hypoallergenic oil (Firooz Baby Moisturizing Oil) was warmed between the palms to enhance absorption and reduce skin friction.

The 30-minute HSE protocol is divided into three parts: pre-HSE care (5-minute), HSE Techniques and Application (20-minute), and post-HSE care (5-minute).

The pre-HSE care was a five-minute warm-up and consisting of slow walking in place and static stretches like gentle leg swings (forward-backward and side-to-side) while holding onto the chair.

The main HSE protocol was to perform three parts of hip strengthening exercises that each participant applied to themselves, which included side-lying leg lift (the participant lies on their side and uses a resistance band around the distal thighs to raise the top leg to a 30-degree angle, holds for 2 to 5 s, and then lower), standing single-leg stabilization (the participant stands and uses a resistance band around the ankles to hold one leg stationary and abduct the other leg 30 degrees, holds for 2 to 5 s, and then return to normal), single-leg standing off the side of a 10-cm step (The participant places one foot on a 10-centimeter step and the other foot on the ground. Then, by contracting the hip abductor muscle in the leg on the step, the participant raises the other leg to the step level, holds for 2 to 5 s, and then returns to normal. The three types of exercises above were performed by the participants in 3 sets of 10–15 repetitions, with 30 s of rest between sets. Participants also used graded Thera-Band bands (red → green → blue) based on progression. Progression to greater resistance levels occurred when participants could perform the exercise without fatigue for 20 repetitions [28].

The post-HSE care was a five-minute cool-down and consisting of static stretches such as quadriceps and hamstring stretch.

Participants were instructed to perform exercises in a clear, safe space near a sturdy chair or countertop for balance support. Comfortable, non-restrictive clothing and supportive footwear were recommended.

Also, a reminder software (an alert to the participant at a specific time pre-set in the software) was also used to remind the participant of the intervention start time to ensure proper adherence and timely completion of the exercises at home.

The control group received no active exercise or massage intervention beyond their routine clinical care, which included analgesics as prescribed and general physician advice to maintain daily activities as tolerated.

To minimize attrition rates and facilitate safety monitoring while avoiding any attention bias, participants attended bi-weekly assessment-only visits lasting 15–20 min. These visits involved solely the collection of outcome measures (e.g., VAS pain score, WOMAC index) and a brief general health check; no therapeutic education, counseling, motivational support, or placebo-like activities were provided. This design ensured ethical equivalence in follow-up without confounding the intervention effects.

Adherence was tracked using daily self-report logs (completed paper form), supplemented by weekly telephone follow-ups (5–10 min) and the bi-weekly in-person visits for log review and troubleshooting. Also, a reminder software (an alert to the participant at a specific time pre-set in the software) was also used to remind the participant of the intervention start time to ensure proper adherence and timely completion of the exercises at home. Compliance exceeded 85% across both intervention groups, with non-adherent sessions addressed through motivational reminders without altering the protocol.

Adverse events were prospectively and systematically assessed at every bi-weekly visit using a standardized questionnaire (template reproduced in Supplementary File 2), which probed for specific symptoms such as muscle soreness, joint pain, skin irritation, fatigue, falls, or any new/worsening complaints. No serious or moderate adverse events (e.g., joint injury, allergic reactions, or falls requiring medical attention) were reported in any group. Mild, transient muscle soreness (resolving within 48 h) occurred in three HSE participants and was managed with rest and hydration.

Outcome measures

Demographic and clinical characteristics

Self-reported data included age, sex, BMI, disease duration, frailty, and comorbidities. Frailty was assessed using the Fried Frailty Phenotype, which evaluates five criteria (unintentional weight loss, exhaustion, weakness, slow walking speed, low physical activity), scored 0–5 to classify participants as robust, pre-frail, or frail [29–31]. Comorbidities were measured using the Charlson Comorbidity Index, which assigns weighted scores to 17 chronic conditions based on severity and mortality risk, with higher scores indicating greater disease burden [32, 33].

Pain

Pain intensity was measured using the Visual Analogue Scale (VAS, 0–10 cm) at baseline and week 8 [16, 18, 33, 34]. We acknowledge that the VAS is a self-reported outcome and participants were not blinded to their group allocation, which is a potential source of bias due to expectations (e.g., placebo effects). To minimize other sources of bias, a blinded assessor was responsible for recording the VAS scores provided by the participants.

Daily function

Daily function was assessed using the Knee Injury and Osteoarthritis Outcome Score (KOOS) subscale (17 Likert items, scored 0–100) at weeks 0 and 8 [9, 35, 36].

Range of motion

Knee flexion range of motion was measured using a SAEHAN goniometer at baseline and week 8. Measurements were performed by a blinded rheumatologist who was unaware of the participants’ group allocation. Three repetitions were conducted and averaged for each measurement (intraclass correlation coefficient > 0.9). In cases of bilateral involvement, the dominant leg was used for assessment.

Statistical analysis

Data were analyzed using IBM SPSS Statistics version 24.0 (IBM Corp., Armonk, NY, USA). Continuous variables are expressed as mean ± standard deviation and categorical variables as frequency (percentage). All efficacy analyses were conducted according to the intention-to-treat (ITT) principle, including all 75 randomized participants. Missing data at the 8-week endpoint (n = 5) were handled using multiple imputation (fully conditional specification method with 10 imputed datasets).

Normality of the data and residuals was evaluated using the Shapiro–Wilk test and visual inspection of Q-Q plots. In accordance with CONSORT guidelines, baseline characteristics were presented for visual inspection only; no statistical tests were performed on baseline variables.

Between-group differences in change from baseline to week 8 were analyzed using analysis of covariance (ANCOVA), with the baseline value of the respective outcome entered as covariate. The two pre-specified primary comparisons (Swedish massage vs. control and hip strengthening exercises vs. control) were performed; p-values were adjusted using the Bonferroni correction (α = 0.05 / 2 = 0.025). Within-group pre–post changes were assessed using paired-samples t-tests. Categorical variables were compared using the χ² test or Fisher’s exact test as appropriate.

Standardized effect sizes (Cohen’s d) were calculated by dividing the adjusted mean difference by the pooled baseline standard deviation. All tests were two-sided, and statistical significance was declared at p < 0.05 (p < 0.025 after Bonferroni adjustment for the two primary comparisons). Confidence intervals reported are 95%.

Results

Participant flow from screening to analysis is presented in the CONSORT diagram (Fig. 1), with detailed baseline and outcome data in Tables 1 and 2. Outcome measurements included patient-reported outcomes for pain and function, and a blinded objective assessment for range of motion. Also, the analyses were performed with blinded statisticians using the intention-to-treat (ITT) approach with last observation carried forward (LOCF) for missing data.

Fig. 1.

CONSORT diagram

Table 2.

Between-group comparisons of changes in clinical outcomes from baseline to end of intervention (intention-to-treat analysis, N = 75)

Outcome	SM (n = 25) Mean change (95% CI)	HSE (n = 25) Mean change (95% CI)	Control (n = 25) Mean change (95% CI)	SM vs. Control Adjusted mean difference (95% CI) Cohen’s d	HSE vs. Control Adjusted mean difference (95% CI) Cohen’s d	Overall p-value (ANCOVA)
VAS pain (cm)	–0.56 (− 0.90 to − 0.22)	–0.52 (− 0.86 to − 0.18)	+ 0.25 (− 0.09 to + 0.59)	−0.81 (− 1.24 to − 0.38) p < 0.001 d = 0.69	−0.77 (− 1.20 to − 0.34) p < 0.001 d = 0.65	< 0.001
KOOS-ADL score	+ 2.30 (+ 0.85 to + 3.75)	+ 2.11 (+ 0.66 to + 3.56)	−1.29 (− 2.74 to + 0.16)	+ 3.59 (+ 1.62 to + 5.56) p < 0.001 d = 0.71	+ 3.40 (+ 1.43 to + 5.37) p = 0.001 d = 0.67	0.001
Knee flexion ROM (degrees)	+ 2.17 (+ 0.69 to + 3.65)	+ 2.44 (+ 0.96 to + 3.92)	−1.25 (− 2.73 to + 0.23)	+ 3.42 (+ 1.40 to + 5.44) p = 0.001 d = 0.73	+ 3.69 (+ 1.67 to + 5.71) p < 0.001 d = 0.77	< 0.001

Adjusted for baseline value of the respective outcome (ANCOVA) with Bonferroni correction for the two pre-specified primary comparisons. Cohen’s d = adjusted mean difference divided by pooled baseline standard deviation of the total sample (n = 75). Positive values for VAS indicate worsening, negative values indicate improvement. Positive values for KOOS-ADL and ROM indicate improvement. Missing post-intervention data (n = 5) were handled by multiple imputation. VAS: Visual Analogue Scale (0–10 cm); KOOS-ADL: Knee Injury and Osteoarthritis Outcome Score – Activities of Daily Living subscale; ROM: active knee flexion range of motion; SM: Swedish massage; HSE: Hip strengthening exercises

Of 75 initially enrolled participants (age > 60 years, KOA per ACR criteria, VAS 40–70), 70 completed the 8-week intervention. Five participants were lost to follow-up: two from the Swedish massage (SM) group (one due to low adherence [< 70% sessions], one withdrew), two from the hip strengthening exercises (HSE) group (one due to low adherence, one withdrew), and one from the control group (low adherence).

All analyses were performed according to the intention-to-treat (ITT) principle using the full randomized sample of 75 participants. Missing endpoint data for the five participants were handled by multiple imputation.

Baseline demographic and clinical characteristics were well balanced across the three groups (Table 1). Table 1 presents the intention-to-treat population (n = 75). In accordance with CONSORT guidelines, no statistical testing was performed on baseline variables; the table is provided solely for visual inspection of between-group balance.

Table 2 presents the primary efficacy results: between-group comparisons of changes from baseline to week 8 (ANCOVA adjusted for baseline values).

Both active interventions resulted in statistically significant and clinically meaningful greater improvements than the control group across all outcomes (Table 2). The reduction in VAS pain was significantly greater in the SM group (adjusted mean difference vs. control: −0.81 cm, 95% CI − 1.24 to − 0.38; p < 0.001) and the HSE group (− 0.77 cm, 95% CI − 1.20 to − 0.34; p < 0.001). KOOS-ADL scores improved significantly more in the SM group (+ 3.59 points, 95% CI + 1.62 to + 5.56; p < 0.001) and HSE group (+ 3.40 points, 95% CI + 1.43 to + 5.37; p = 0.001) compared with control. Active knee flexion range of motion also increased significantly more in both intervention groups (SM vs. control: +3.42°, p = 0.001; HSE vs. control: +3.69°, p < 0.001).

Standardized effect sizes (Cohen’s d) for the primary comparisons (active arms vs. control) ranged from 0.65 to 0.77, indicating moderate treatment effects.

Discussion

This randomized controlled trial directly compared the effectiveness of Swedish massage (SM) and hip abductor strengthening exercises (HSE) in adults aged over 60 years with knee osteoarthritis (KOA). The primary aim was to identify the optimal non-pharmacological approach for reducing pain, improving range of motion, and enhancing daily function to promote healthy aging. Results indicated that both interventions, relative to a no-intervention control group, produced statistically significant and clinically meaningful improvements across all primary outcomes (Cohen’s d ranging from 0.65 to 0.77; moderate effect size).

The head-to-head comparison between the two active interventions, which was the stated primary objective, represents a central finding: no statistically significant differences were observed between SM and HSE on any primary outcome (all p > 0.70). The adjusted mean difference between groups was 0.04 cm for VAS pain reduction, 0.19 points for KOOS-ADL improvement, and 0.27° for active knee flexion range of motion increase (all 95% confidence intervals included zero). This numerical equivalence aligns with the clinically meaningful benefits of both approaches in a frail, multimorbid older adult population.

These results are consistent with recent evidence: for SM, the RCT by Perlman et al. (2019) [16] reported VAS reductions of approximately 0.7–1.0 cm (d = 0.6–0.8) with weekly 60-min sessions over 8 weeks, congruent with our SM group (VAS − 0.56 cm), though without specific focus on older adults; Bervoets et al. [37]deemed massage superior to no treatment for short-term pain and function in musculoskeletal disorders (SMD = − 0.65), but noted limited long-term durability, unlike our short-term stability. For HSE, Qiu et al. [38] showed KOOS improvements of approximately 3 points (d = 0.7) with hip exercises added to quadriceps strengthening, aligning with our HSE results, yet Hislop et al. [39] reported no significant ROM effects a novelty in our study with + 2.44° ROM gain (d = 0.77), likely attributable to the home-based protocol. Network meta-analysis by Lan et al. [12] ranked physical modalities like strengthening exercises as equivalent to manual therapies for pain (SMD = − 0.72), but lacked direct head-to-head trials; our study addresses this gap in older adults, where Yan et al. [40] positioned HSE as among the top exercise modalities for pain and function. A key divergence: pre-2020 studies often yielded smaller effects (d < 0.5), whereas our d = 0.65–0.77 highlights enhanced home-based efficacy, contrasting non-congruent findings in Sabet et al. [15] no long-term gait improvements with SM. While direct comparisons exist in broader OA literature (e.g., Abbott et al. [41], comparing manual therapy to multi-modal exercise with equivalent pain reductions), the specific SM-HSE pairing for older KOA adults is among the few available reports, adding to targeted head-to-head evidence.

Quantitative equivalence of home-delivered SM and HSE for pain, ROM and function in frail older adults with KOA extends previous network-meta-analytic inferences with direct patient-level data.

This clinical equivalence despite divergent mechanisms is scientifically compelling. Hip abductor strengthening primarily engages biomechanical pathways, reducing knee adduction moment (KAM) and optimizing lower limb dynamic alignment to lessen medial compartment loading [28, 39]. Conversely, Swedish massage targets neurophysiological routes: mechanical receptor stimulation in skin and soft tissue → activation of gate control theory and descending inhibitory pathways → diminished central pain perception, tissue stiffness, and pain catastrophizing/fear of movement [37, 42]. Thus, both disrupt the maladaptive pain-fear-avoidance-weakness cycle via distinct entry points, converging on equivalent outcomes a finding resonant with the biopsychosocial model of chronic pain in older adults [43, 44].

Clinically, this parity empowers patient-centered selection based on movement fear, physical capacity, and preferences. Swedish massage may be preferable for those with severe pain or kinesiophobia avoiding resistance training, while hip strengthening suits individuals tolerating moderate loading and seeking sustained biomechanical correction [45].

The 8-week implementation in a frail cohort (mean Fried Frailty ≈ 1.8–1.9) with high comorbidity burden (Charlson Index ≈ 3.3) achieved excellent adherence (> 92%) and no serious adverse events, affirming safety and feasibility of both home protocols in complex cases. Post-hoc inspection showed that participants with Fried ≥ 2 achieved the same relative benefit as those with Fried = 1, but absolute residual pain remained higher, suggesting that eight weeks may be insufficient for full symptomatic control in the frailest stratum; longer or progressive “dose-extension” phases should therefore be tested. The high screening-to-enrollment ratio (> 10:1) underscores well-known recruitment barriers in geriatric trials; nonetheless, post-enrollment, home delivery yielded superior retention, positioning these as pragmatic strategies to broaden non-pharmacological access in resource-limited settings [12]. Only 75 of 710 screened patients (10.6%) entered the trial; while home delivery retained volunteers, it did not solve upstream reluctance to enroll, indicating that proactive outreach (tele-screening, caregiver engagement) remains essential for implementation.

The high refusal rate, a common barrier in geriatric and self-management trials [46], indicates that while home-based programs are viable and engaging for those who accept them, significant upstream barriers exist. These may include time commitment, skepticism about non-pharmacological treatments, or a preference for passive care. Therefore, while our home-based model successfully supported adherence post-enrollment, its success in “enhancing participation” is primarily relevant to those already willing to engage. Broader implementation will require complementary strategies to improve initial uptake, such as targeted education, involvement of primary care providers in referral, and potentially tele-health support to overcome initial hurdles [47].

Conclusions

This randomized controlled trial demonstrates that both Swedish massage and hip-strengthening exercises represent distinct yet equally effective non-pharmacological approaches for managing knee osteoarthritis in frail older adults. SM offers a viable option for patients seeking passive pain management that simultaneously improves functional capacity, while HSE provides an active approach that yields comparable benefits across pain, function, and range of motion domains. The complementary nature of these interventions suggests that clinical implementation could be tailored to individual patient preferences, capabilities, and clinical characteristics. For older adults with mobility limitations and limited healthcare access, both modalities represent practical, safe, and effective strategies that can be integrated into existing care frameworks to address the multifaceted challenges of knee osteoarthritis.

Limitations

This study has several limitations. First, the lack of participant blinding and the absence of a sham control group increase the risk of performance bias and Hawthorne effects, although blinding of the research personnel who collected data, the rheumatologist performing the ROM measurements, and the statisticians conducting the data analyses blinding helped minimize detection bias. Second, the 8-week intervention period and lack of long-term follow-up prevent assessment of the durability of observed effects. Third, the small sample size (n = 70 after attrition) and relatively high number of excluded participants may also affect the generalizability of the findings. The high number of screened patients relative to enrollees resulted from stringent inclusion criteria designed to create a homogeneous sample and ensure participant safety. This enhances the study’s internal validity by reducing confounding variables but means our results are most directly generalizable to older adults similar to our cohort. Finally, the study’s focus on Iranian older adults with mild-to-moderate frailty, combined with the high rate of refusal to participate, limits the generalizability of results to other cultural contexts, healthcare systems, and population subgroups.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

A.A. and A.R. wrote the main manuscript text, and M.A. helped in gathering data, and Z.K. analyzed the data, and T.M. revised the manuscript.

Data availability

Data sets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

The study protocol was approved by the Ethics Committee of Rafsanjan University of Medical Sciences, Rafsanjan, Iran with the code of IR.RUMS.REC.1402.068.

Human and animal rights

Approvals and permissions were obtained from research ethic committees of Iran, Rafsanjan University of Medical Sciences.

Informed consent

Written informed consent was obtained from all participants.