Non-Pharmacological Complementary and Alternative Therapies for Knee Osteoarthritis: A Scoping Review of Randomized Controlled Trials

Background

Knee osteoarthritis (KOA) is a chronic degenerative joint disease characterized by progressive cartilage loss, osteophyte formation, synovial inflammation, pain, stiffness, and functional limitation. It is one of the leading causes of disability worldwide, and its burden is increasing with population ageing and rising obesity prevalence.^1,2 Current clinical guidelines recommend education, exercise, weight management, and other non-pharmacological strategies as core components of KOA management, while pharmacological and surgical options are selected according to symptoms, risk profile, and disease severity.^3,4 Although analgesics and non-steroidal anti-inflammatory drugs may relieve symptoms, long-term or inappropriate use can be limited by gastrointestinal, cardiovascular, and renal adverse events.⁵

In this review, complementary and alternative medicine (CAM) is used as an umbrella term for health practices, products, and therapeutic systems that are not usually regarded as part of conventional biomedicine when used alone, or that are used alongside conventional care when applied as complementary or integrative approaches. For KOA, CAM-related non-pharmacological interventions include traditional medicine techniques such as acupuncture, moxibustion, Tuina, Gua Sha, and mind-body exercise, as well as manual therapy, yoga, massage, behavioral self-management, physical agent modalities, and other non-drug supportive approaches when evaluated as alternatives or adjuncts to usual care.^3,4,6

Several systematic reviews, meta-analyses, umbrella reviews, and network meta-analyses have already examined specific non-pharmacological modalities or compared selected intervention classes for KOA.^6–12 These studies are useful for estimating comparative effects when interventions, comparators, and outcomes are sufficiently similar. However, the present literature also contains substantial diversity in intervention concepts, treatment dose, provider background, control design, follow-up schedule, and outcome domains. A scoping review is therefore appropriate when the objective is not to pool effects, but to map the range of available RCT evidence, clarify how interventions have been conceptualized and studied, and identify evidence gaps that can guide future systematic reviews and pragmatic trials.¹³

The objective of this scoping review was to map RCT evidence on non-pharmacological CAM and related integrative interventions for adults with KOA. Specifically, we aimed to describe the distribution of studies, classify intervention categories, summarize comparator and outcome patterns, identify key methodological and reporting features, and clarify evidence gaps relevant to future research and clinical implementation.

Methods

This scoping review followed the methodological framework proposed by Arksey and O’Malley and was reported in accordance with the PRISMA-ScR guideline.¹³ A formal effect-size synthesis or risk-of-bias-based certainty assessment was not planned because the purpose was evidence mapping rather than comparative efficacy estimation.

Research Questions Based on PCC

The review question was structured using the population-concept-context framework. The population was adults with clinically or imaging-confirmed KOA. The concept was non-pharmacological CAM, complementary, alternative, or integrative interventions evaluated in RCTs. The context included hospital, outpatient, community, home-based, online, and other clinical or research settings. The review addressed four questions: (1) Which categories of non-pharmacological CAM or related integrative interventions have been evaluated in RCTs for KOA? (2) What intervention components, dose parameters, provider backgrounds, and delivery settings have been reported? (3) What comparator types and outcome domains have been used? (4) What evidence gaps and methodological patterns can be identified from the included RCTs?

Eligibility Criteria

Inclusion criteria were as follows: (1) study design: RCTs; (2) participants: adults aged 18 years or older with clinically or imaging-confirmed KOA; (3) interventions: non-pharmacological, non-surgical CAM, complementary, alternative, or integrative interventions, including traditional medicine techniques, mind-body exercise, manual or body-based therapies, physical agent modalities, behavioral/self-management programs, and other non-drug supportive approaches; and (4) outcomes: at least one pain, physical function, quality-of-life, psychological, biomechanical, physiological, safety, adherence, or implementation-related outcome.

Trials were eligible when the CAM or non-pharmacological component could be separated from the pharmacological effect by study design. Therefore, trials comparing a non-pharmacological intervention plus the same background pharmacological treatment versus that pharmacological treatment alone were included, because the incremental contribution of the non-pharmacological component could be mapped. Trials using pharmacological comparators were also included when the experimental arm was a non-pharmacological intervention and the comparator was used as an active control. Trials were excluded when the effect of the non-pharmacological component could not be separated from a co-intervention, when participants had inflammatory arthritis or non-KOA pain conditions without separable KOA data, when critical data were missing, or when the publication was a conference abstract, protocol, review, commentary, or retracted article.

Information Sources and Search Strategy

PubMed, Embase, the Cochrane Library, and Web of Science were searched from 1 January 2016 to 10 January 2026. Search strategies combined terms for KOA, complementary or alternative therapies, and randomized trials. Full database-specific strategies were added to Supplementary File 1. The PubMed strategy is shown below as an example.

#1 “Osteoarthritis, Knee”[Mesh] OR (“Knee Osteoarthritides”[tiab] OR “Knee Osteoarthritis”[tiab] OR “Osteoarthritis of Knee”[tiab] OR “Osteoarthritis of the Knee”[tiab])

#2 “Complementary Therapies”[Mesh] OR (“Complementary Therapy”[tiab] OR “Alternative Medicine”[tiab] OR “Complementary Medicine”[tiab] OR “Alternative Therapies”[tiab] OR acupuncture[tiab] OR moxibustion[tiab] OR Tuina[tiab] OR massage[tiab] OR yoga[tiab] OR Tai Chi[tiab] OR Baduanjin[tiab] OR “manual therapy”[tiab] OR “mind-body”[tiab] OR “transcutaneous electrical nerve stimulation”[tiab] OR TENS[tiab])

#3 “randomized controlled trial”[Publication Type] OR randomized[Title/Abstract] OR randomised[Title/Abstract] OR placebo[Title/Abstract] OR sham[Title/Abstract]

#4 #1 AND #2 AND #3

The Embase strategy used Emtree terms and title/abstract keywords for knee osteoarthritis, complementary therapy, acupuncture, moxibustion, Tuina, massage, yoga, Tai Chi, Baduanjin, manual therapy, mind-body therapy, TENS, and randomized trial design. The Cochrane Library and Web of Science strategies used equivalent controlled vocabulary where available and free-text terms in title, abstract, and keyword fields. Complete strategies are provided in Supplementary File 1 to enable replication.

Study Selection and Data Charting

Two reviewers independently screened titles/abstracts and full texts according to the eligibility criteria. Disagreements were resolved by discussion or consultation with a third reviewer. A predesigned charting form was used and is provided as Supplementary File 2. Extracted items included bibliographic details, country, study setting, participant characteristics, sample size, intervention category, intervention components, dose and duration, provider background, comparator type, follow-up schedule, outcome domains, safety reporting, and key design features.

Synthesis of Results

Evidence was summarized descriptively and narratively. We mapped the included trials by intervention category, comparator type, follow-up duration, geographic distribution, outcome domain, and design characteristics. Because intervention protocols, comparators, and outcome measures were heterogeneous, and because the aim was to characterize the evidence base rather than estimate comparative effectiveness, no meta-analysis or network meta-analysis was conducted.

Critical Appraisal and Design-Characteristic Summary

A formal risk-of-bias assessment was not undertaken because it is not mandatory for scoping reviews and was not aligned with the evidence-mapping objective. However, to improve interpretability, we added a design-characteristic summary documenting sample size, blinding or sham-control status, comparator type, and follow-up duration for each included RCT (Supplementary Table S1). These data provide a minimal quality signal and help readers judge the maturity and limitations of the evidence base.

Results

Study Selection

The database searches identified 1141 records: Cochrane Library (n=361), PubMed (n=444), Web of Science (n=67), and Embase (n=269). After removal of 609 duplicates, 532 records were screened by title and abstract. Of these, 446 were excluded because they were clearly irrelevant (n=299), had an ineligible study type (n=133), or were not written in English (n=14). Eighty-six full-text articles were assessed for eligibility. Thirty-three were excluded because the outcome was not eligible or insufficiently reported (n=14), the intervention did not meet the review concept (n=8), the population was not KOA or KOA data were not separable (n=6), or critical information/full text was unavailable (n=5). Finally, 53 RCTs were included in the scoping review. The selection process is shown in Figure 1.

Figure 1 PRISMA flow diagram of study selection. The flow diagram reports 1141 records identified, 609 duplicates removed, 532 records screened, 446 title/abstract exclusions, 86 full-text articles assessed, 33 full-text exclusions, and 53 RCTs included in the scoping review.

Characteristics of Included Studies

The 53 included RCTs were published between 2016 and 2026 and were conducted across multiple countries, with a high concentration in China, the United States, Turkey, and several other regions. Table 1 summarizes the characteristics of included trials. The evidence base included hospital-based, outpatient, community, home-based, and online delivery settings. Sample sizes ranged from small pilot trials to larger multicenter or multi-arm studies. Intervention duration ranged from a single session to 32 weeks, while most studies assessed outcomes at baseline and immediately after treatment; longer follow-up beyond 6 months was less common.

Table 1 Characteristics of Included Randomized Controlled Trials (n=53)

Intervention Categories

The included studies were reorganized into six evidence-mapping categories according to intervention concept, delivery mechanism, and clinical tradition. Acupuncture and related traditional needling therapies included manual acupuncture, electroacupuncture, superficial needling, Fu’s subcutaneous needling, acupotomy, thread-embedding acupuncture, warm needling/moxibustion, and intradermal needling. Manual and body-based therapies included Tuina, Swedish massage, Persian massage, joint mobilization, dry needling, reflexology, and Gua Sha. Exercise and mind-body programs included Tai Chi, yoga, chair yoga, Baduanjin, structured strengthening, treadmill-based rehabilitation, and isokinetic training. Physical agent modalities included TENS, NMES, interferential current, periosteal stimulation therapy, and related device-based stimulation. Psychological and behavioral interventions included positive psychological activities, guided imagery, mindfulness meditation, self-management support, transtheoretical model-based exercise adherence support, gamification, and graded motor imagery. The remaining interventions were classified as other integrative or complementary approaches, including aromatherapy, black cumin seed oil massage, non-contact spiritual therapy/Reiki, rTMS, and multimodal integrative programs. Because this last group was clinically heterogeneous, it was interpreted as a residual evidence-mapping category rather than a single coherent therapeutic class.

Outcome Domains Mapped Across Studies

Patient-reported pain and function were the dominant outcome domains. Pain was commonly assessed using VAS, NRS, WOMAC pain, KOOS pain, SF-MPQ, and related measures. Function and symptoms were commonly assessed using WOMAC, KOOS, Lequesne index, Oxford Knee Score, KOOS JR, patient acceptable symptom state, global assessment, or responder definitions. Performance-based tests included Timed Up and Go, 6-minute walk test, walking speed, 30-second chair stand, repeated chair rise, stair climb, single-leg stance, balance tests, and simulated activities of daily living.

Quality-of-life and psychological outcomes were reported less consistently. These included SF-36/SF-12, EQ-5D, WHOQOL-BREF, HADS, BDI, PHQ-4/PHQ-9, DASS-21, pain catastrophizing, kinesiophobia, self-efficacy, sleep quality, fatigue, and treatment satisfaction. Mechanistic and objective outcomes were present in a smaller subset of trials and included range of motion, quadriceps strength, pressure pain threshold, conditioned pain modulation, inflammatory markers, gut microbiota profiles, ultrasound-measured cartilage thickness, step counts, adherence, and neuroimaging indices such as resting-state functional connectivity and ALFF.

Comparator Types and Design Features

Comparator designs varied substantially. Sham or placebo controls were common in acupuncture, electrical stimulation, massage, physical therapy, and energy-therapy trials. Other trials used usual care, wait-list or no-intervention controls, education, home exercise, active non-pharmacological comparators, pharmacological comparators, or combination designs in which the non-pharmacological component was added to background pharmacological or rehabilitation care. This diversity is important for evidence mapping, but it limits any direct interpretation of comparative treatment effects across categories.

Evidence Mapping Patterns and Gaps

Table 2 summarizes the main evidence-mapping patterns by intervention category. Acupuncture-related and exercise/mind-body trials formed the largest clusters, whereas psychological/behavioral interventions, rTMS, energy therapies, aromatherapy, and some multimodal integrative programs were represented by fewer trials. Most trials emphasized short-term changes in pain and function; fewer studies examined long-term durability, adverse events in detail, implementation fidelity, adherence maintenance, costs, or healthcare utilization. Direct head-to-head comparisons between major CAM categories were uncommon. There were also gaps in population reporting, including KOA severity grades, comorbidity burden, older adults with frailty, and participants outside TCM-centered healthcare systems.

Table 2 Evidence-Mapping Summary by Intervention Category

Discussion

Principal Evidence-Mapping Findings

This scoping review mapped the distribution and design characteristics of RCT evidence on non-pharmacological CAM and related integrative interventions for KOA. The main finding is not that one intervention category is superior to another, but that the RCT evidence base is broad, heterogeneous, and unevenly developed. Acupuncture-related therapies, exercise and mind-body programs, and manual/body-based approaches account for much of the available trial literature, whereas several emerging or less conventional approaches are represented by small numbers of studies. Across categories, the most frequently measured domains were pain and physical function, while long-term durability, implementation outcomes, cost-related outcomes, safety detail, and mechanistic measures were less consistently reported.

The choice of a scoping review is supported by this heterogeneity. Existing systematic reviews and meta-analyses have synthesized selected modalities or intervention classes.^6–12 In contrast, the present review maps how the broader RCT landscape is organized, including intervention taxonomies, comparator selection, design features, outcome domains, and evidence gaps. These mapping results can help determine where future systematic reviews are feasible and where primary research remains too sparse or heterogeneous for formal comparative synthesis.

Methodological Patterns and Implications for Interpretation

Several methodological patterns limit direct interpretation. First, many trials used short intervention periods and short follow-up, making it difficult to map durability beyond treatment completion. Second, comparator designs differed substantially, ranging from sham controls to active pharmacological comparators, usual care, education, home exercise, and complex combination controls. Third, provider training, fidelity monitoring, intervention dose, and maintenance strategies were often incompletely reported. Fourth, a subset of studies used small sample sizes, single-session designs, or limited blinding. These features do not invalidate the scoping map, but they indicate that readers should avoid interpreting the included trials as a uniform evidence base for comparative efficacy.

The evidence map also highlights reporting priorities. Future RCTs should report intervention components, provider qualifications, dose, frequency, duration, adherence, co-interventions, comparator rationale, blinding feasibility, allocation concealment, attrition, adverse events, and follow-up time points in sufficient detail. Such information is necessary for later systematic reviews, implementation studies, and guideline translation.

Intervention Categorization and Heterogeneous Residual Groups

Intervention classification required careful handling because many CAM and integrative approaches overlap conceptually. For example, graded motor imagery was classified under psychological and behavioral interventions rather than exercise/mind-body therapies because its primary mechanism is neurocognitive imagery and pain-related sensorimotor representation rather than physical training. Similarly, the category of other integrative or complementary approaches was retained only as a residual mapping category. It includes aromatherapy, spiritual or energy-based interventions, rTMS, and multimodal integrative programs, which should not be interpreted as clinically equivalent. This distinction was added to avoid overgeneralization from heterogeneous interventions.

Generalizability and Geographic Concentration

A large proportion of included RCTs originated from China, particularly trials evaluating TCM-centered interventions such as acupuncture, Tuina, Baduanjin, moxibustion, Gua Sha, and acupotomy. This concentration is important for interpretation. Trials conducted in Chinese TCM hospitals or TCM-integrated health systems may involve provider training, patient expectations, treatment credibility, cultural familiarity, and care pathways that differ from those in non-TCM health systems. Therefore, generalizability to regions where these therapies are less integrated into routine care should be considered cautiously. Future trials should include more diverse healthcare contexts and report setting-specific implementation factors.

Specific Evidence Gaps Identified from Included Trials

The included studies suggest several specific gaps. First, direct comparisons between major CAM categories, such as acupuncture versus manual therapy, acupuncture versus structured exercise, and mind-body exercise versus conventional strengthening, remain limited. Second, long-term follow-up beyond 6 to 12 months is uncommon, limiting evidence on durability and maintenance. Third, implementation outcomes such as adherence, fidelity, provider training, patient preference, treatment credibility, cost, and scalability are under-reported. Fourth, mechanistic outcomes are concentrated in a small subset of studies and are rarely linked to pragmatic clinical outcomes. Fifth, many trials lack detailed reporting of KOA severity, comorbidities, medication use, and population characteristics needed to evaluate applicability to older adults, frail patients, or patients with advanced structural disease.

Strengths and Limitations

This review has several strengths. It focused on RCTs, covered multiple databases, used duplicate screening and data charting, and mapped intervention categories, outcome domains, comparator types, follow-up schedules, and key design characteristics. The revision also included complete database-specific search strategies, clarified eligibility for pharmacological comparators and combination designs, followedthe PRISMA flow, and included a design-characteristic summary.

This review also has limitations. First, no formal risk-of-bias assessment or certainty grading was performed, consistent with the scoping objective, so the results should not be interpreted as comparative efficacy conclusions. Second, because only English-language publications were included, relevant trials in other languages may have been missed. Third, the broad review concept produced substantial heterogeneity, especially in the residual category of other integrative or complementary approaches. Fourth, some methodological details were incompletely reported in the included trials, limiting the granularity of the evidence map. Fifth, the geographic concentration of TCM-centered trials may limit generalizability.

Conclusions

This scoping review maps a broad and heterogeneous body of RCT evidence on non-pharmacological CAM and related integrative interventions for KOA. The available trials are concentrated in several intervention clusters, especially acupuncture-related therapies, exercise and mind-body programs, manual and body-based therapies, and physical agent modalities. The evidence base is characterized by diverse comparators, short-term follow-up, variable reporting of intervention dose and provider features, and uneven reporting of design quality signals. Rather than supporting direct comparative effectiveness claims, the review identifies where evidence is concentrated and where research remains insufficient. Future work should prioritize transparent reporting, longer follow-up, better characterization of intervention fidelity and implementation context, more diverse populations and settings, and targeted systematic reviews or pragmatic trials where the evidence base is sufficiently mature.