Introduction

Knee osteoarthritis is characterized by breakdown of the articular cartilage over time1,2. Although cartilage breakdown is the major disease characteristic, osteoarthritis affects all joint tissues, including the synovial membrane, which is usually associated with increased pain and joint dysfunction2,3. Common clinical symptoms include knee pain with gradual onset and that worsens with activity, knee stiffness and swelling, pain after prolonged sitting or resting, and pain that worsens over time4. Some studies have reported that approximately 13% of women and 10% of men aged 60 years or older have symptomatic knee osteoarthritis5,6.

Treatment initially involves nonsurgical modalities and progresses to surgical treatment once nonsurgical methods are no longer effective4. These interventions do not alter the disease process but may substantially reduce pain and disability7,8. Self-management programs, muscle strengthening, low-impact aerobic exercises, neuromuscular therapy, and physical activity are recommended for patients with knee osteoarthritis9,10,11,12. Oral pharmacological agents such as nonsteroidal anti-inflammatory drugs (NSAIDs) and corticosteroids are effective in the treatment of knee osteoarthritis13,14,15. However, oral anti-inflammatory drugs may increase the risks of gastrointestinal, renal, and other systemic toxicities. Topical gels are an alternative treatment option with fewer complications compared with oral anti-inflammatory drugs16,17,18.

Ultrasound, as a treatment modality, has been studied for many decades19. Ultrasound exerts a therapeutic effect through the absorption of mechanical energy and the production of heat in tissues20. Phonophoresis involves the use of ultrasound to deliver therapeutic drugs by absorption and permeation of the skin21. Phonophoresis with an anti-inflammatory gel has been reported to alleviate pain and inflammation in many musculoskeletal conditions22,23,24,25. Despite the wide use of phonophoresis, scientific evidence supporting its use is insufficient, especially with regard to symptomatic knee osteoarthritis. Wu et al. conducted a systematic review and meta-analysis comparing the effects of therapeutic ultrasound with those of sham ultrasound on knee osteoarthritis26. A subgroup analysis indicated that the phonophoresis ultrasound group reported less pain (measured using the visual analog scale [VAS]) than did the conventional nondrug ultrasound group26. No significant differences in functional performance (determined on the basis of Western Ontario and McMaster Universities Arthritis Index [WOMAC] score) were observed between the groups26. However, only three randomized controlled trials (RCTs) were included in the study to compare the effect of phonophoresis with that of nondrug therapeutic ultrasound. Moreover, according to our electronic database search, more RCTs have been published recently. Thus, this study investigated the effect of phonophoresis for various gel types and compared these effects against those of a placebo (nondrug) gel; these gels were used in therapeutic ultrasound for treating knee osteoarthritis.

Method

This review was performed in accordance with the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions27 and is reported following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses28. This systematic review was registered in the International Prospective Register of Systematic Reviews database under the number CRD42021266126 on August 6, 2021.

Eligibility criteria

We included studies that (1) were RCTs; (2) included patients with a diagnosis of knee osteoarthritis; (3) involved treatment with phonophoresis (with an NSAID, corticosteroid, Chinese herbal gel, or other gel) as the intervention; (4) involved therapeutic ultrasound with a placebo (nondrug) gel as the control treatment; and (5) reported clinical outcomes including pain scores (measured using the VAS) and functional performance (assessed on the basis of WOMAC function score, range of motion, and a walk test). We excluded articles that were protocols, non-peer-reviewed articles, conference papers, and letters to the editor. No language restriction was applied in our search strategy.

Literature search

We searched electronic databases, namely Medline (using PubMed), EMBASE, and CENTRAL. In our search strategy, we included terms related to both phonophoresis and knee osteoarthritis and their synonyms (the search strategies are presented in the Supplementary Appendix). If available, RCTs were identified using the refined search function of the databases. Additional articles were identified by manually searching the reference lists of relevant articles. The databases were searched from their inception to June 28, 2021.

Study selection

Only RCTs that compared the effects of phonophoresis with those of nondrug therapeutic ultrasound on pain and physical function in patients with knee osteoarthritis were included. Titles and abstracts were screened to select relevant articles. Two reviewers independently evaluated the eligibility of all titles and abstracts, and disagreements were resolved through discussion. A third reviewer adjudicated any disagreement that could not be resolved through discussion. Subsequently, the full texts of remaining articles were read in detail to determine the eligibility of the articles.

Data extraction

Two authors extracted data from each study by using a structured form, and the characteristics of all eligible studies are summarized in a table. The following data were extracted: (1) basic information of qualifying studies (first author and publication date); (2) demographic, clinical, and treatment characteristics (e.g., number and mean age of patients in the control and treatment groups); (3) therapeutic ultrasound parameters (mode, frequency, intensity, and duration); (4) content of the gel used in phonophoresis; (5) follow-up period; and (6) outcome measures. Moreover, the means and standard deviations (SDs) of outcome measurements before and after treatment for the experimental and control groups were extracted. If crucial data could not be extracted from an article, we sent an email to the corresponding author, requesting the data.

Outcome measurements

The outcome measurements of this study were pain score, WOMAC function score, range of motion, walk test score, and adverse events. The pain scores were VAS scores29. Higher VAS scores indicate a higher intensity of pain. The WOMAC function score is obtained from a self-administered questionnaire widely used for evaluating hip and knee osteoarthritis30. Higher WOMAC function scores indicate higher pain intensity, greater stiffness, and poorer physical function. Range of motion is the range through which a joint can be moved31. The walk tests included in this study were the 6-min walk test, timed up and go test, 15-min walk test, and 20-min walk test32,33.

Risk of bias assessment

Risk of bias was examined using the RoB 2 tool, a revision of the Cochrane risk-of-bias tool for RCTs, which is widely used for assessing the quality of RCTs34. The following domains were considered: (1) the randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) outcome measurement, (5) the selection of reported results, and (6) overall bias34. Following the Cochrane Handbook for Systematic Reviews of Interventions, the risk of bias was assessed by two independent reviewers27. Disagreements between the reviewers were resolved through discussion and consultation with a third reviewer.

Statistical analysis

Statistical analyses were performed using RevMan 5.3 software, which is provided by the Cochrane Collaboration (https://training.cochrane.org/online-learning/core-software-cochrane-reviews/revman/revman-5-download). Continuous data were extracted as changes from baseline measurements. For studies not reporting SDs, the authors were contacted for raw data or, if unavailable, the data were estimated by calculating correlation coefficients in accordance with the instructions provided in the Cochrane Handbook for Systematic Reviews of Interventions27. Results with P < 0.05 were considered statistically significant. We used the I2 test to objectively measure statistical heterogeneity, with I2 ≥ 75% indicating considerable heterogeneity35. A random effects model was used in this meta-analysis due to clinical and methodological heterogeneity. Continuous variables are presented as standardized mean differences (SMDs) with 95% confidence intervals (CIs). The analysis was performed on the basis of different gel contents (corticosteroid, NSAID, herbal gel, or other gel). The meta-analysis was conducted only when at least two RCTs assessed the same gel type. SMDs were used to examine the strength of the relationships between variables in a population; an SMD of < 0.2 was considered to indicate a trivial effect with no clinical meaningfulness; 0.2–0.5 indicated a small effect; 0.5–0.8 indicated a moderate effect; and > 0.8 indicated a large effect36.

Sensitivity analysis was performed by excluding one or two studies at a time to examine the stability and reliability of the meta-analysis. We performed this analysis to evaluate the effect of outliers27. Articles with a high risk of bias or those reporting dubious results were regarded as outliers. Outliers were identified by the extent to which their removal reduced overall heterogeneity. Furthermore, we identified the probable causes of outliers after performing the sensitivity analysis.

A funnel plot was constructed to examine publication bias if the number of studies included in each analysis was more than 10.

Results

Search results

By using the search terms mentioned in the supplementary appendix, we initially retrieved 2176 studies. Of these, 633 duplicates were excluded using EndNote X937. Furthermore, 1491 studies that did not meet the inclusion criteria were excluded upon screening of their titles and abstracts. We screened the full texts of the remaining 52 papers and determined that 4 studies included duplicate study populations, 9 were not yet published, 2 did not compare the intervention with placebo gel, 21 did not examine phonophoresis, 1 had an additional intervention to the experimental intervention, 2 were not peer-reviewed articles, 2 compared phonophoresis with iontophoresis, and 2 did not mention SDs. Finally, nine articles were included in the meta-analysis38,39,40,41,42,43,44,45,46. A PRISMA flowchart illustrates the selection process and the number of articles, with reasons included for why studies were excluded at each step of the meta-analysis47 (Fig. 1).

Figure 1
figure 1

Flowchart of article selection.

Full size image

Study characteristics

The selected studies included 222 and 201 patients in the intervention and control groups, respectively. All the selected RCTs were randomized, placebo-controlled trials38,39,40,41,42,43,44,45,46. Two studies used a corticosteroid gel38,41, six used an NSAID gel40,42,43,44,45,46, and one used a herbal gel39. One study had a 1-month follow-up period40, and one study had a 3-month follow-up period;42 other studies obtained follow-up data within 1 week after the intervention38,39,41,43,44,45,46. Table 1 lists the main characteristics of the nine RCTs.

Table 1 Characteristics of selected randomized controlled trials.
Full size table

Risk‑of‑bias assessment

Figure 2 illustrates the risk of bias for each study. Nine studies had low risk associated with the randomization process38,39,40,41,42,43,44,45,46. Six studies exhibited some concerns regarding the risk associated with deviations from the intended intervention40,41,42,44,45,46, whereas three studies exhibited low risk38,39,43. All nine studies had low risk related to missing outcome data38,39,40,41,42,43,44,45,46. Regarding outcome measurements, five studies exhibited low risk38,39,41,43,44, and four exhibited some concerns40,42,45,46. Regarding the selection of reported results, all nine studies had low risk38,39,40,41,42,43,44,45,46. The overall risk of bias was low for three studies38,39,43 and uncertain for six studies40,41,42,44,45,46.

Figure 2
figure 2

Study quality assessment.

Full size image

Pain scores

Pain scores were assessed in five studies where patients were treated with NSAID gel40,42,43,44,46. These studies included 145 patients in the experimental group and 125 patients in the control group. In the analysis, SMD = − 0.53, 95% CI [− 1.02, − 0.05], and I2 = 73%, which indicated a significant difference and favored the intervention group (Fig. 3).

Figure 3
figure 3

Forest plot for changes from baseline determined using the pain scores. SD, standard deviation; CI, confidence interval.

Full size image

Additionally, pain scores were assessed in two studies where patients were treated with corticosteroid gel38,41. These studies included 42 patients in the experimental group and 41 patients in the control group. In the analysis, SMD = − 0.53, 95% CI [− 1.13, 0.07], and I2 = 46%, which indicated no significant difference between groups (Fig. 3).

WOMAC function scores

WOMAC function scores were assessed in five studies where patients were treated with NSAID gel40,42,44,45,46. These studies included 137 patients in the experimental group and 117 patients in the control group. In the analysis, SMD = − 0.75, 95% CI [− 1.63, 0.13], I2 = 90%, which indicated no significant difference between groups (Fig. 4).

Figure 4
figure 4

Forest plot for changes from baseline determined using the Western Ontario and McMaster Universities Arthritis Index (WOMAC) function scores. SD, standard deviation; CI, confidence interval.

Full size image

WOMAC function scores were assessed in two studies where patients were treated with corticosteroid gel38,41. These studies included 42 patients in the experimental group and 41 patients in the control group. In the analysis, SMD = − 0.96, 95% CI [− 1.47, − 0.44], and I2 = 20%, which indicated a significant difference and favored the intervention group (Fig. 4).

Range of motion

The range of motion was assessed in three studies where patients were treated with NSAID gel40,45,46. These studies included 84 patients in the experimental group and 64 patients in the control group. In the analysis, SMD = 1.07, 95% CI [− 0.09, 2.00], and I2 = 90%, which indicated no significant difference between groups (Fig. 5). No study assessed the range of motion when corticosteroid gel was applied.

Figure 5
figure 5

Forest plot for changes from baseline based on range of motion. SD, standard deviation; CI, confidence interval.

Full size image

Walk tests

Walk tests were adopted in three studies where patients were treated with NSAID gel44,45,46. These studies included 78 patients in the experimental group and 78 patients in the control group. In the analysis, SMD = − 0.57, 95% CI [− 1.27, 0.12], and I2 = 76%, which indicated no significant difference between groups (Fig. 6).

Figure 6
figure 6

Forest plot for improvement in walk test scores. SD, standard deviation; CI, confidence interval.

Full size image

Only one study evaluated the effect of herbal gel. The study was not included in the meta-analysis because only one RCT was conducted39. This study included 20 patients in the experimental group and 20 patients in the control group and focused on pain scores and walk test results. The herbal gel had a larger effect in the intervention group than in the control group in both tests.

Sensitivity analysis

In the sensitivity analysis, we observed that pain scores with NSAID gel were affected by the removal of studies. However, the effect of NSAID gel on range of motion became significant and favored the intervention group after one article was removed. All results indicated lower heterogeneity after articles were removed (I2 < 75%; Table 2).

Table 2 Sensitivity analysis.
Full size table

Adverse events

Of the nine selected RCTs, four reported adverse effects39,40,43,46. No adverse events were observed in these studies, indicating that participants tolerated the interventions well.

Discussion

Knee osteoarthritis is a degenerative joint cartilage condition1,2. The common clinical symptoms of knee osteoarthritis include knee pain that is gradual in onset and worsens with activity, knee stiffness and swelling, pain after prolonged sitting or resting, and pain that worsens over time4. Topical anti-inflammatory drugs are an alternative treatment choice with fewer gastrointestinal complications relative to oral drugs16,17,18.

In phonophoresis, ultrasound is used to deliver therapeutic drugs through absorption and permeation of the skin21. Despite the wide usage of this treatment, supporting scientific evidence is insufficient, especially with regard to symptomatic knee osteoarthritis. Thus, we conducted this study to investigate the effect of phonophoresis on knee osteoarthritis symptoms. Our analysis revealed significant differences in pain scores that indicated phonophoresis with NSAID gel and significant differences in WOMAC function score that indicated phonophoresis with corticosteroid gel.

We examined the strength of the relationships between variables in a population by determining effect sizes. The results revealed that NSAID gel moderately affected pain scores and that corticosteroid gel greatly affected WOMAC function scores. The findings indicated that phonophoresis could clinically and meaningfully improve the patients’ pain and functional performance. In addition, no study has compared the effects of phonophoresis with those of physical therapy or other types of medical treatments according to our electronic database research. Future studies can fill this research gap.

Therapeutic ultrasound is a deep-heating modality used in physical therapy42. According to Rao et al., in therapeutic ultrasound, a transducer converts electrical energy into ultrasound through the piezoelectric principle48. Although the exact mechanism underlying its effect remains known, the effect may be composed of two components, namely thermal and nonthermal49. In terms of the thermal effect, therapeutic ultrasound induces muscle relaxation and increases connective tissue extensibility and local blood flow, all of which lead to tissue regeneration and reduce inflammation42,49. The nonthermal ultrasound effect is related to acoustic cavitation with resultant increases in cell permeability, which is a potential pain relief mechanism49. Zhang et al. suggested that therapeutic ultrasound is beneficial for reducing knee pain and improving physical function in patients with knee osteoarthritis and can be a safe treatment option50. Phonophoresis refers to the use of ultrasound to deliver therapeutic drugs by absorption and permeation through the skin21. The advantage of therapeutic ultrasound is that it may promote the transdermal penetration of therapeutic drugs43,46. Moreover, this method is noninvasive and has minimal risk of the adverse effects associated with the systemic administration of anti-inflammatory drugs; it also combines the therapeutic effects of ultrasound and topical drugs43. Phonophoresis accounts for up to 30% of physiotherapy visits in some medical centers45.

Recently, gels with different contents have been developed for phonophoresis. Corticosteroid and NSAID gels are commonly used. Among the included RCTs, two focused on corticosteroid gels38,41, six focused on NSAID gels40,42,43,44,45,46, and one focused on herbal gels39. The two studies that used corticosteroid gels used dexamethasone gels38,41. Among the six RCTs that focused on NSAID gels, three used diclofenac gels40,42,45, one used the ibuprofen gel46, one used the ketoprofen gel44, and one used the piroxicam gel43. The herbal gel used was the Phyllanthus amarus gel39. Although each type of gel exerts anti-inflammatory effects, their chemical properties (e.g., their tissue permeability through ultrasound waves) differ, as reported by Akinbo et al.45 In their literature review, Srbely et al. indicated that the depth of penetration of a drug depends on its mass (which is inversely proportional to its molecular weight)51. Molecular weight is different from the contents of gels discussed in the selected RCTs. Dexamethasone has high molecular weight; thus, it has a low drug mass and high permeability when applied through ultrasound. The aforementioned reasons may explain why patients in the corticosteroid gel subgroup exhibited greater improvements in some outcomes than did those in the NSAID gel subgroup45. Thus, the selection of drug for phonophoresis appears to be as crucial to treatment success as the selection of ultrasound parameters38.

Byl et al. reported that diffusion of topically applied drugs through the skin can be enhanced by preheating the skin to increase the kinetic energy52. Among our included RCTs, three studies followed this approach41,45,46. The application of heat before treatment may have affected the results of these studies. Our analysis indicated that the outcomes reported by the studies that applied preheating were inconsistent. Some studies reported improved outcomes, whereas others indicated no differences when compared with outcomes without preheating application. Therefore, the exact effects of preheating the skin should be investigated in the future.

We conducted a sensitivity analysis by excluding one or two studies at a time to examine the stability and reliability of the meta-analysis. According to the Cochrane Handbook for Systematic Reviews of Interventions, heterogeneity may arise due to the presence of one or two outlying studies with results that are in conflict with those of the remaining studies27. If an obvious reason for the outlying result is apparent, the study might be removed with confidence27. Both Akinbo et al. and Kozanoglu et al. applied heat to the treatment site before treatment45,46. They followed the principle indicated by Byl et al.52 However, the application of heat may affect treatment outcomes. The findings of the three studies applying preheating might differ from those of the other included studies39,45,46. This difference might explain the change in results in the sensitivity analysis.

In their systematic review and meta-analysis, Wu et al. examined the effectiveness and safety of various therapeutic ultrasound methods26. In the subanalysis of phonophoresis, three RCTs were examined43,44,46. The results revealed that the phonophoresis ultrasound group had lower pain scores (SMD = − 0.41, 95% CI [− 0.71, − 0.10]) but demonstrated no significant difference in functional performance (assessed on the basis of WOMAC score; SMD = − 0.16, 95% CI [− 0.46, 0.14]). In recent years, more studies on phonophoresis have been conducted38,39,40,41,42,43,44,45,46. We examined the effect of phonophoresis on patients with knee osteoarthritis. We focused on the outcomes of pain and functional performance and determined that phonophoresis effectively improved such outcome measures.

This systematic review and meta-analysis has several strengths. First, this is the first meta-analysis of RCTs investigating the effects of phonophoresis for different types of gels on patients with knee osteoarthritis. Second, our electronic database search indicates that several studies are ongoing in this field. Thus, the results of this study can serve as a reference for future studies. Third, multiple major databases were searched, without language restrictions, prior to the selection of RCTs. Fourth, the data and quality of the selected studies were examined by at least two reviewers through a group consensus approach.

Our study has several limitations that may limit the generalizability of our results. First, heterogeneity was moderate to high for some outcomes, possibly because of varying disease severity, symptom duration, patient characteristics, and treatment protocol. Thus, additional studies are required to establish a standardized treatment protocol. Second, different gel contents, such as lidocaine or capsaicin, that could be applied in the experimental group were not investigated. Furthermore, only one study focused on herbal gels and was not included in our meta-analysis, although it revealed a positive effect. Future studies should examine the effects of different gel contents. Third, some studies did not mention blinding to therapeutics and the blinding of patients or assessors. Hence, some concerns regarding the risk of bias may persist. Fourth, we observed that the pain scores with NSAID gel were affected by the removal of articles when conducting the sensitivity analysis. This might compromise the stability and reliability of the meta-analysis. Fifth, the included studies had short follow-up durations. Therefore, we could not analyze long-term outcomes. Fifth, the sample sizes in each study and the number of studies included for each outcome were different. Thus, the outcomes should be interpreted with caution. Additional high-quality large-scale RCTs with long-term follow-up periods are required to overcome these limitations.

Conclusion

This is the first meta-analysis of RCTs to investigate and provide adequate evidence for the effect of phonophoresis for different gel types on patients with knee osteoarthritis. Our findings indicated that phonophoresis improves pain and functional performance with a moderate to large effect size over a short-term follow-up with either corticosteroid or NSAID gel. Furthermore, no adverse events were reported in the selected studies. Phonophoresis can be an effective treatment option for patients with knee osteoarthritis. However, because of the limitations of this study, additional high-quality, large-scale RCTs with long follow-up periods are required to confirm the benefit and long-term effects of this intervention.