Clinically significant effects of gait modification on knee pain: A systematic review and meta-analysis

1.Introduction

Knee osteoarthritis (OA) affects about one in five adults aged over 45 years [1], and it is a leading cause of chronic physical disability. Moreover, this condition is characterized by pain and commonly affects the medial aspect of the knee [2, 3]. In knee OA, excessive mechanical stress in the joint is a major contributor to knee pain [4]. Worsening knee pain can lead to higher medical costs and a lower quality of life [5]. To prevent these issues, early treatment for knee pain relief must be provided.

Land-based exercise is one of the most effective therapies for knee pain relief [6, 7, 8] and walking is one of land-based exercises. As excessive loading on the medial aspect of the knee contributes to knee OA progression [9, 10, 11], performing walking exercises in a manner that reduces the medial knee loading is reasonable. Different types of gait modifications for reducing medial load have been proposed. Mechanical stress on the knee changes with the type of gait [12]. Multiple gait modification methods and their effects on mechanical stress on the knee and knee pain have been investigated [12, 13, 14]. Results showed that gait modification can alleviate knee pain.

Although several types of gait modifications are effective, the level of evidence for their efficacy is unknown. Moreover, even if knee pain relief after gait modification is statistically significant, whether its effect is clinically significant is unknown. Hence, this notion should be investigated to improve the evidence-based practice of walking instructors who use methods for alleviating knee pain.

Therefore, the current study aimed to assess the level of evidence for the efficacy of gait modification against knee pain. Moreover, whether its effect is clinically significant was investigated using the minimum clinical important difference (MCID), which is a criterion to evaluate the significance of an effect from a patient perspective rather than from a statistical perspective. To the best of our knowledge, this is the only study that evaluated the clinical significance of gait modification based on a high level of evidence and the perspective of patients.

2.Materials and methods

A systematic literature review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocol guidelines [15] and registered with UMIN. Because this was a systematic review, approval by the ethics committee was not required.

2.4Data collection process

Figure 1.

Flow diagram of the study selection process. CENTRAL; Cochrane Central Register of Controlled Trials, PEDro; Physiotherapy Evidence Database, CINAHL; Cumulative Index to Nursing and Allied Health Literature.

To extract data about, simple predesigned spreadsheets created using Microsoft Excel 2019 were prepared and used to extract data on participants, interventions, outcome measurements, and results. Two authors (HT and KH) discussed and decided whether gait modification was used and whether the outcome was pain.

3.Results

3.2Study characteristics

Table 1

Patient characteristics

Authors	Design	Populations	Size	Age (SD)	Female (%)	Height (SD)	Body mass (SD)	BMI (SD)	K/L 1 (%)	K/L 2 (%)	K/L 3 (%)	K/L 4 (%)
Hunt et al. [19]	RCT	Radiographic	EG: 40	EG: 64.6 (7.6)	EG: 27 (67.5)	EG: 1.64 (0.10)	EG: 73.6 (13.2)	EG: 27.3 (3.5)		EG: 19 (48)	EG: 17 (42)	EG: 4 (10)
		KOA	CG: 39	CG: 65.4 (9.6)	CG: 28 (71.8)	CG: 1.66 (0.09)	CG: 75.6 (13.2)	CG: 27.4 (3.5)		CG: 18 (46)	CG: 14 (36)	CG: 7 (18)
Hunt et al. [20]	Non-RCT	Radiographic	16	64.8 (5.2)	9 (56.2)	Not recorded	Not recorded	29.9 (3.4)		4 (25)	9 (56)	3 (19)
		KOA
Richards et al. [21]	Non-RCT	Radiographic	21	61.3 (5.73)	15 (71.4)	1.72 (0.08)	75.9 (11.0)	25.4 (2.6)	12 (63)	2 (10.5)	4 (21.1)	1 (5.2)
		KOA
Cheung et al. [22]	RCT	Radiographic	EG: 10	EG: 60.8 (6.4)	EG: 5 (50)	EG: 1.62 (0.09)	EG: 64.3 (7.1)	EG: 24.5 (2.4)	EG: 2 (10)	EG: 8 (40)
		KOA	CG: 10	CG: 63.1 (5.9)	CG: 5 (50)	CG: 1.64 (0.10)	CG: 67.4 (6.1)	CG: 25.2 (1.1)	CG: 3 (15)	CG: 7 (35)
Segal et al. [23]	RCT	Symptomatic	EG: 29	EG: 69.1 (3.7)	EG: 22 (75.9)	Not recorded	Not recorded	Not recorded		EG: 5 (18)	EG: 13 (47)	EG: 10 (36)
		KOA	CG: 19	CG: 69.6 (3.2)	CG: 10 (52.6)					CG: 5 (26)	CG: 9 (47)	CG: 5 (26)
Takacs et al. [24]	Non-RCT	Radiographic	12	64.1 (9.1)	7 (58.3)	Not recorded	Not recorded	28.3 (4.9)		5 (42)	5 (41.6)	2 (18)
		medial KOA
Shull et al. [25]	Non-RCT	Radiographic	10	60 (13)	4 (40)	171 (9)	79 (20)	26.6 (4.7)		3 (30)	6 (60)	1 (10)
		medial KOA
Simic et al. [26]	Non-RCT	Radiographic	22	68.4 (5.1)	13 (59)	1.67 (0.05)	78.3 (8.1)	27.9 (2.4)		9 (41)	9 (41)	4 (18)
		medial KOA
Wang et al. [27]	RCT	Radiographic	EG: 31	EG: 59.1 (7.2)	EG: 16 (51.6)	EG: 1.65 (0.09)	EG: 62.0 (11.0)	EG: 1.65 (0.09)	EG: 15 (48)	EG: 16 (51)
		medial KOA	CG: 31	CG: 61.7 (6.8)	CG: 17 (54.8)	CG: 1.65 (0.08)	CG: 64.2 (10.6)	CG: 1.65 (0.08)	CG: 16 (51)	CG: 15 (48)

Table 2

The content of gait modification

Authors	Type of gait modification	Biofeedback	Number of training sessions	Training period
Hunt et al. [19]	Toe-out gait	Mirror-guided biofeedback	8 (1, 2, 4, 6, 8, 10, 12, 15 weeks)	4 months
Hunt et al. [20]	Toe-out gait	Real-time biofeedback of performance	6 (1, 2, 3, 5, 7, 9 weeks)	10 weeks
Richards et al. [21]	Toe-in gait	Visual feedback	6 (1, 2, 3, 4, 5, 6 weeks)	6 weeks
Cheung et al. [22]	Toe-out gait	Visual feedback	6 (1, 2, 3, 4, 5, 6 weeks)	6 weeks
Segal et al. [23]	Gait aimed to move toward symmetrical and typical displacements of the trunk and pelvis about neutral frontal and transverse axes	Visual 3D	24 (biweeks)	12 months
Takacs et al. [24]	Trunk lean gait	Visual feedback	1	Immediate
Shull et al. [25]	Toe-in gait	Visual feedback	6 (1, 2, 3, 4, 5, 6 weeks)	6 weeks
Simic et al. [26]	Trunk lean gait	Real-time movement biofeedback system	1	Immediate
Wang et al. [27]	Toe-in gait	Visual feedback	6 (1, 2, 3, 4, 5, 6 weeks)	6 weeks

Table 2 shows the population characteristics of the studies included in this review. Four RCTs [19, 22, 23, 27] and five non-RCTs [20, 21, 24, 25, 26] were analyzed. The sample sizes ranged from 10 to 79, and the participants were classified under the experimental and control groups. The average age of the participants in the nine studies ranged from 59.1 to 69.6 years.

Table 2 shows the content of gait modification performed in the studies. The gait modification methods were toe-out gait [19, 20, 22], toe-in gait [21, 25, 27], trunk lean gait [24, 26], and move toward symmetrical and typical displacements of the trunk and pelvis about neutral frontal and transverse axes [23]. Chueng et al. [22], Shull et al. [25], and Wang et al. [27] did not indicate a specific gait modification method, but all subjects walked with toe-out gait or toe-in gait. All gait modification methods were visual feedback using mirrors and biofeedback.

Table 3

Outcome data

Authors	Outcome scale	Evaluate point	Baseline (SD), [IQR]		Post-intervention (SD), [IQR]		Mean difference (SD)
			EG	CG	EG	CG	EG	CG
Hunt et al. [19]	WOMAC P, WOMAC F, NRS	4, 5 months	WOMAC P: 7.6 (0.5) WOMAC F: 28.1 (1.9) NRS: 4.7 (0.4)	WOMAC P: 6.4 (0.4) WOMAC F: 21.4 (1.5) NRS: 3.7 (0.3)	WOMAC P: 4.2 (0.5) WOMAC F: 13.0 (1.6) NRS: 2.0 (0.3) 4 months	WOMAC P: 5.4 (0.5) WOMAC F: 16.7 (1.6) NRS: 2.3 (0.3) 4 months	WOMAC P: -2.7 (3.0)§ WOMAC F: -11.4 (9.7)§ NRS: -2.1 (1.8)§ 4 months	WOMAC P: -1.5 (3.0)§ WOMAC F: -7.7 (9.5)§ NRS: -1.8 (1.8)§ 4 months
Hunt et al. [20]	WOMAC P, WOMAC T, NRS	11 weeks	WOMAC P: 7.4 (3.4) WOMAC T: 36.9 (14.8) NRS: 4.5 (1.7)	Not application	WOMAC P: 5.3 (2.9) WOMAC T: 26.4 (13.5) NRS: 2.6 (1.8)	Not application	WOMAC P: -2.1 (3.3)§ WOMAC T: -10.5 (15.2)§ NRS: -1.9 (1.6)§	Not application
Richards et al. [21]	WOMAC P, WOMAC F, NRS	6 weeks, 3, 6 months	WOMAC P: 10.5 [9] WOMAC F: 14.5 [16] NRS: 1.5 [2]	Not application	WOMAC P: 7 [6] WOMAC F: 6.5 [10] NRS: 1 [2] 6 weeks	Not application	Not recorded	Not application
Cheung et al. [22]	WOMAC P, WOMAC F	6 weeks 6 months	WOMAC P: 30.0 (12.5) (%) WOMAC F: 29.3 (7.2) (%)	WOMAC P: 22.5 (12.3) (%) WOMAC F: 24.7 (7.2) (%)	Not recorded	Not recorded	Not recorded	Not recorded
Segal et al. [23]	KOOS	3, 6, 12 months	62.7 (10.8)	59.8 (13.1)	Not recorded	Not recorded	8.2 (14.0)§ 3 months	1.1 (13.7)§ 3 months
Takacs et al. [24]	NRS	Immediate	1.8 (1.8)	Not application	2.2 (2.1)	Not application	0.3 (0.9)§	Not application
Shull et al. [25]	WOMAC P, WOMAC F, VAS	6, 10 weeks	Not recorded	Not application	Not recorded	Not application	Not recorded	Not application
Simic et al. [26]	WOMAC P, WOMAC F, NRS	Immediate	WOMAC P: 5 (3) WOMAC F: 19 (11) NRS: 2.2(2.0)§	Not application	NRS: 2.3 (2.9)§ 6∘ leam 2.2 (2.3)§ 9∘ leam 2.1 (2.0)§ 12∘ leam	Not application	Not recorded	Not application
Wang et al. [27]	KOOS, VAS	6 weeks	KOOS: 73.1 (7.9) VAS: 4.3 (2.2)	KOOS: 72.5 (12.1) VAS: 4.5 (2.0)	Not recorded	Not recorded	KOOS: 6.4 (8.8)§ VAS: Not recorded	KOOS: Not recorded VAS: Not recorded

VAS: Visual Analog Scale, WOMAC P: Western Ontario and McMaster Universities Osteoarthritis pain, WOMAC F: Western Ontario and McMaster Universities Osteoarthritis function, WOMAC T: Western Ontario and McMaster Universities Osteoarthritis total, IQR: Interquartile range, KOOS: Knee injury and Osteoarthritis Outcome Score, NRS: Numerical Rating Scale, EG: experimental group, CG: control group.

§Using the literature made by the statistical methods group of the Cochrane Collaboration (https://training.cochrane.org/handbook/statistical-methods-revman5), the authors calculated these values using the statistics described in the original articles.

Table 4

GRADE summary of findings

Certainty assessment								No. of patients		Certainty	Importance
Outcome	No. of studies	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	EG	CG
Pain	2	RCT	Very seriousa	Not serious	Not serious	Seriousb	None	65	22	⊕○○○	Important
										Very low

GRADE; Grading of Recommendations Assessment, Development and Evaluation, RCT; randomized controlled trial, EG; experimental group, CG; control group. Explanations: aIndicating one study with high risk of bias; bWide 95% confidence interval.

Figure 2.

Forest plot of the standardized mean differences (SMD) between gait modification and non-gait modification.

Table 3 shows the outcome data of the studies. Six studies used the WOMAC [19, 20, 21, 22, 25, 26]; two, the KOOS [23, 27]; two, the VAS [25, 27]; and five, the NRS [19, 20, 21, 24, 26]. Hunt et al. [19] revealed that the mean deference in the WOMAC pain subscale score between baseline and post-intervention was 2.7 in the experimental group. Cheung et al. [22] found that the rate of change in knee pain in the experimental group was 49% based on the WOMAC. Wang et al. [27] showed that the percentage of change in the VAS score was 35.5%. Three non-RCTs used toe-in or toe-out gait. In the research of Hunt et al. [20], the WOMAC score changed by 2.1 and the NRS score by 1.9. Meanwhile, Richards et al. [21] revealed that the WOMAC score changed from 10.5 at baseline to 7 at post-intervention. The percentage of change in the WOMAC score was 29% and the decrease in VAS was 2 points on the study of Shull et al. [25].

4.Discussion

Nine interventional studies were included in the systematic review and two RCTs in the meta-analysis. Moreover, the GRADE system was used to determine the level of evidence. Results showed that the quality of evidence for the efficacy of gait modification against knee pain was very low. However, several studies have revealed that gait modification effects were greater than the MCID. That is, some types of gait modifications can alleviate knee pain, thereby providing patient satisfaction.

The level of evidence supporting the effect of gait modification on knee pain is a novel finding. The conventional practical guidelines for knee OA recommends land-based exercises [6, 7]. However, to the best of our knowledge, there are no guidelines on the characteristics of walking (e.g., foot direction and stride length). Our review, which evaluated the level of evidence for gait modification methods, can contribute in updating guidelines that recommend walking.

Our study was conducted specifically on individuals with knee OA. In a previous systematic review, the target population comprised healthy individuals and those with knee OA, and it focused mainly on knee adduction moment [12, 13, 14]. In contrast, our study performed a systematic review and meta-analysis focusing on pain in people with knee OA alone. Our findings are valuable for clinicians who want to provide satisfactory care to people with knee OA.

Gait modification, which can significantly alleviate knee pain, focuses on the feet. The results of the GRADE system did not provide a high level of evidence for the efficacy of gait modification against knee pain. This indicates that the effect size is likely to change in future studies. However, changing foot orientation is a common denominator among gait modifications with statistical significance in RCTs [19, 22] with a low risk of bias (PEDro score of ⩾ 6). In the study by Hunt et al. [19], the experimental group participants were instructed to perform toe-out gait. Cheung et al. referred to previous study [25] and conducted an similar intervention study [22] in which participants freely selected between three different gait modifications, including changes in foot progression angle, to reduce knee adduction to below 80%. These two studies included foot-focused gait modification in their interventions.

Although the effect size cannot be guaranteed, foot-focused interventions can have effects that are beyond patient satisfaction. According to Pham et al. [28], the MCID of the WOMAC improved by pain > 20% and > 10 points on a scale of 0–100 and by function > 20% and > 10 points on a scale of 0–100. In the study of Salaffi et al. [29], the MCID of the NRS improved by > 15% or > 1 point on a scale of 0–10. Wandel et al. [30] used the criteria that the MCID of the VAS is > 0.90 cm on a scale of 0–10 cm. The WOMAC or NRS score was the outcome used in Hunt et al. [19] and Cheung et al. [22], and the VAS score was the outcome used in Wang et al. [27], and the change in pain was beyond the MCID. When focusing on toe-in or toe-out gait, two studies [20, 25], other than the RCT, showed changes caused by gait modification that were greater than the MCID. These findings could be used as evidence supporting the notion that foot-focused gait modification can significantly relieve knee pain.

Pain relief through gait modification that focuses on the foot can be effective for physical function. According to Newberry et al. [31], the MCID of the WOMAC function is 9.1 on a scale of 0 to 100 [32]. Of the studies included in our review, there were five studies [19, 21, 22, 25, 26] that used the scale of the WOMAC function. Four of them [19, 21, 22, 25] had a gait modification focused on the foot, all of which significantly improved the scores and three of them [19, 22, 25] exceeded the MCID. One study [21] did not report the amount of change, so it could not be determined if the change exceeded the MCID. Simic et al. [26], who focused on trunk lean, did not report the amount of change, so it could not be determined whether the effect exceeded the MCID. Thus, gait modification focused on the foot may be clinically effective not only in reducing pain but also in improving physical function.

This study showed the characteristics of individuals with knee OA who can benefit from the procedure beyond the MCID. Five studies, which showed that the effect of toe-in or toe-out gait was beyond the MCID, included overweight women aged over 60 years. Based on the OARSI guideline [6, 7], a common core intervention is recommended for all patients. Additionally, patients are sub-grouped, and different interventions are recommended for each subgroup. The two criteria for subgrouping are OA other than in the knee and presence of complications. Nevertheless, age, sex, and body composition are not considered. To make walking more effective than gait modification, a subgroup of older and overweight women should be targeted.

The current study had several limitations. First, there were only four RCTs in the systemic review, and only two RCTs specified the extent of change between the experimental and control groups. Therefore, no reliable conclusions about the effect size could be drawn. Nevertheless, more high-quality studies should be performed to provide stronger evidence about the effect size of gait modification. Second, only few studies used similar gait modification methods. The data of multiple RCTs could not be integrated using similar gait modification methods, and a reliable effect size was not achieved. Since effect sizes between methods could not be compared, we cannot confirm which gait modification method is the most effective. Because the effect size of gait modification focused on the toe-in and/or toe-out exceeded the MCID, more RCTs confirming this effect size are warranted. Despite these limitations, to the best of our knowledge, there are currently no systematic reviews that determined the level of evidence for the efficacy of gait modification. Our review can help clinicians in the decision-making process regarding the use of gait modification and can facilitate designing future studies.

5.Conclusion

The level of evidence for the efficacy of gait modification against knee pain in people with knee OA is very low. However, toe-in or toe-out gait is effective in relieving knee pain in RCTs, and several studies revealed that it is more effective than MCID. These reports included overweight women aged over 60 years old.

Ethical approval

This study has been registered with UMIN (registration number: 000045647).

Informed consent

Not applicable.

Funding

This work was supported by the Grants-in-Aid for Scientific Research of Japan Society for the Promotion of Science (Grant Number JP20K11230).

Author contributions

Protocol stage: All authors drafted the protocol.

Review stage: KH and HT selected which reviews to include.

Review stage: HT, KH and KH extracted data from reviews.

Review stage: HT entered data into RevMan.

Review stage: HT carried out the analysis.

Review stage: All authors interpreted the analysis.

Review stage: All authors drafted the final review.

Update stage: All authors updated the review.