Effects of Osteoarthritis and Fatigue on Proprioception of the Knee Joint

Methods 

The study subjects were 50 outpatients with bilateral OA of the knees, and 30 healthy volunteers. The dominant extremity was the right extremity in all subjects. OA was diagnosed according to the criteria established by the American College of Rheumatology. The severity of OA in each affected knee was radiographically graded using the Kellgren-Lawrence grading system,28 and the Lequesne index of severity for knee osteoarthritis (ISK)29 was used to quantify pain. Patients with very severe pain who reported that they were not independent in activities of daily living because of the pain in their knees and those with grade 4 OA were excluded because they were unable to exercise on a cycle ergometer. Other exclusion criteria included knee OA secondary to another disease, any musculoskeletal disease other than OA, any lower-limb joint replacements, or any neurologic conditions (eg, Parkinson’s disease, stroke).

Ethics approval for the study was obtained from the ethics committee of the university and subjects gave informed consent.

We recorded age, sex, height, weight, body mass index (BMI), and quadriceps and hamstring muscle strength (and ratios comparing these 2 muscle groups) for each of the 66 subjects. For the 50 patients, we also recorded the OA grade for each knee and presence and absence of mediolateral instability with valgus and varus stress tests in the right and left knees. In addition, reposition error was tested bilaterally in all controls and in all but 1 patient before and after a fatiguing exercise. This 1 patient had recently undergone arthroscopic surgery on the right knee and underwent reposition error testing of the left knee only.

We used a computerized isokinetic dynamometera to assess reposition error. For this, the subject was seated on the bench of the dynamometer with hips and knees flexed at 90°. Inflatable pressure boots were worn to immobilize the feet and thus eliminate any sensation cues from the skin or from ankle positioning. Knee range of motion was set at 0° to 90° and the continuous passive-motion mode of the dynamometer extended and flexed the knee at 5°/s. Before starting to take measurements, the physician pushed the “stop” button when the knee was at the middle of the range (target angle for the study, 45°) of the movements from flexion to extension, and from extension to flexion. This was repeated 5 times, and then the patient was instructed to, with eyes closed, push the button when he/she sensed the knee was at 45° of flexion. The reposition angle was recorded 5 times during flexion to extension, and 5 times during extension to flexion. The reposition angles for the first flexion-to-extension movement and first extension-to-flexion movement were omitted, and the mean for the other 4 reposition errors in each movement category was recorded. Reposition error was defined as the difference between the target angle and the reposition angle, and the absolute value of this error was used for statistical analysis. To evaluate the effect of fatigue on reposition error, each subject was asked to push the button when he/she sensed the knee was at 45° of flexion again after moderate exercise (5min on the cycle ergometer). Subjects pedaled at 35 to 45rpm, their maximum heart rate not exceeding 100 beats per minute for the moderate exercise protocol. This protocol was chosen to standardize the exercise between healthy controls and patients who were unable to continue pedaling for more than 5 minutes because of pain.

We assessed the reliability of the measurement of reposition error by the continuous passive-motion mode of the isokinetic dynamometer by repeating the tests on a total of 20 subjects (12 patients, 8 controls). We recorded the above mentioned pre-exercise and postexercise measurements again 2 days after the first testing procedure was carried out.

The isometric strength of the quadriceps and hamstring muscles was measured using the isokinetic dynamometer with the subject’s hips and knees flexed at 90°. The ratio of quadriceps strength to hamstring strength (Q/H ratio) was also calculated for each subject’s left and right sides.

We used SPSS softwareb for Windows for statistical analysis. Two-way analysis of variance (ANOVA) was used to compare numerical values within and between groups. Analysis of reposition error in relation to OA grade was made by using the Student t test for independent samples. Relationships between parametric variables were analyzed using the Pearson correlation coefficient. P values less than .05 were considered statistically significant. Intraclass correlation coefficients (ICCs) as described by Fleiss30 were used to test the reliability of the proprioception measurements. ICC values above .75 represent excellent reliability; values between .40 and .75 represent fair-to-good reliability; and values below .40 indicate that the testing was unreliable.

Results 

The patient group comprised 39 (78%) women and 11 (22%) men. The control group consisted of 25 women (83%) and 5 men (17%). The mean age, mean BMI, and sex distribution of the patient and control groups are shown in table 1. There was no statistically significant difference between the groups with respect to age or BMI.

Table 1. Age, Sex, and BMI of the Patient and Control Groups

	Characteristic	Patient Group (n=50)	Control Group (n=30)	P
	Age (y)	60.22±8.43	57.86±9.22	>.05
	Sex (female/male)	39/11	25/5	>.05
	BMI (kg/m2)	30.12±4.92	25.43±2.72	>.05

NOTE. Values are mean ± standard deviation (SD) or n.

Seventeen (34%) patients exhibited instability in both knees, 2 (4%) exhibited instability of the left knee only, and 7 (14%) exhibited instability of the right knee only. The mean ISK score ± standard deviation (SD) for the patient group was 7.45±3.16 (range, 1–14). The radiographic OA grades of the patients are shown in table 2.

Table 2. Radiographic Gradings for the Patient Group

	Grade	Right Knee	Left Knee
	Grade 0	ND	ND
	Grade 1	14	13
	Grade 2	25	24
	Grade 3	11	13
	Grade 4	ND	ND

Abbreviation: ND, no data (ie, no patients had this grade).

The results for isometric quadriceps and hamstring muscle strength revealed that both muscle groups were significantly weaker in the patient group than in the control group. The groups’ muscle strength data are shown in table 3.

Table 3. Mean Isometric Muscle Strengths and Mean Q/H Ratios for the Patient and Control Groups

	Strengths and Ratios	Patient Group (n=50)	Control Group (n=30)	P
	R quadriceps (Nm)	2.44±0.92	3.03±1.18	.015
	L quadriceps (Nm)	2.25±0.86	2.84±1.31	.018
	R hamstring (Nm)	0.63±0.39	0.82±0.44	.055
	L hamstring (Nm)	0.61±0.36	0.74±0.43	.147
	R Q/H	4.89±3.01	4.34±1.78	.371
	L Q/H	4.49±2.33	4.39±2.03	.993

NOTE. Values are mean ± SD. Abbreviations: L, left; R, right.

Testing showed that our process of assessing proprioception using isokinetic dynamometry had fair-to-good reliability. The ICCs for all measurements ranged from 0.5 to 0.7 (table 4).

Table 4. ICCs of the Pre- and Postexercise Measurements Performed in 20 Subjects

	Measurement	ICC
	Right knee flexion to extension, pre-exercise	.673
	Right knee extension to flexion, pre-exercise	.719
	Left knee flexion to extension, pre-exercise	.573
	Left knee extension to flexion, pre-exercise	.681
	Right knee flexion to extension, postexercise	.499
	Right knee extension to flexion, postexercise	.535
	Left knee flexion to extension, postexercise	.526
	Left knee extension to flexion, postexercise	.693

Statistical analysis using 2-way ANOVA did not reveal any difference between group results and pre- versus postexercise results. The data of the 2 groups for reposition errors before and after exercise are shown in table 5.

Table 5. Mean Pre- and Postexercise Reposition Errors⁎ for the 2 Groups and the Tests of Between-Subjects Effects

	Knee	Group	Pre-Exercise	Postexercise
	R flex-ext	Patient	6.29±4.12	6.11±4.81
		Control	5.20±3.80	5.68±2.94
	L flex-ext	Patient	6.45±4.01	6.90±4.59
		Control	5.90±2.84	6.47±2.72
	R ext-flex	Patient	8.95±5.94	6.66±3.49
		Control	7.76±3.91	6.18±3.68
	L ext-flex	Patient	8.11±5.95	7.25±4.62
		Control	8.55±2.65	6.92±3.30

NOTE. Values are mean degrees ± SD. The results of 2-way ANOVA. The statistical significance of the tests for group (patient vs control), pre- vs postexercise and right vs left knee is underlined in the data below: Abbreviations: ext, extension; flex, flexion. Source Type III Sum of Squares df Mean Square F Signif Observed Power⁎ Corrected model 683.687† 15 45.579 2.374 .002 0.986 Intercept 28440.214 1 28440.214 1481.467 .000 1.000 Group (patient vs control) 53.402 1 53.402 2.782 .096 0.384 Pre- vs postexercise 47.743 1 47.743 2.487 .115 0.350 Right vs left knee 45.581 1 45.581 2.374 .124 0.337 Error 11979.132 624 19.197 Total 43659.875 640 Corrected total 12662.819 639 Abbreviation: Signif, significance. ⁎ Tests of Between-Subject Effects Dependent Variable: Reposition Error ⁎ Computed using α=.05. † R2=.054 (adjusted R2=.031).

Analysis of reposition error in relation to OA grade revealed no significant differences in reposition error either before or after exercise between the subgroups of patients with grade 1 and grade 2 OA, or between the subgroups with grade 2 and grade 3 OA. The grade 1 subgroup, however, showed significantly smaller reposition error than the grade 3 subgroup at both time points (from flexion to extension, P=.008; from extension to flexion, P=.005) for the right knee. For the left knee, reposition errors did not differ between the grade 1 subgroup and grade 3 subgroups of patients. The reposition errors for the subgroups with grade 1 and grade 3 OA for the right and left knees are shown in table 6.

Table 6. Mean Reposition Errors of the Right and Left Knees of the Patients

	Radiographic Grade	Right Knee			Left Knee
		No. of Patients	Reposition Error (ext-flex)	Reposition Error (flex-ext)	No. of Patients	Reposition Error (ext-flex)	Reposition Error (flex-ext)
	1	14	4.03±3.06	3.95±2.89	13	5.38±3.10	4.88±3.15
	2	24	5.89±3.17	5.12±3.67	24	7.92±4.02	5.12±3.23
	3	11	8.93±5.36	8.01±4.83	13	8.83±5.78	6.76±3.91
	P⁎		.008	.005		.128	.189

NOTE. Values are mean degrees ± SD.

⁎ Significant between grade 1 and 3 patients.

Correlation analysis revealed that reposition error was not associated with age, sex, BMI, left or right quadriceps strengths, left or right hamstring strengths, or left or right Q/H ratios in either the patient or control group. As well, in the patient group, neither ISK score nor knee instability correlated with reposition error (P>.005).

Discussion 

The results of this study indicate that reposition error in patients with OA of the knees is similar to that in controls with similar age. We did find, however, that patients with more severe OA based on radiographic grading had poorer proprioception than those with less severe radiographic signs. Some of the previous studies that have investigated the relationship between knee OA and proprioception have shown a direct relationship between knee OA and reduced proprioception,3, 7, 31, 32 and some have examined how the combination of age and OA of the knees affects proprioception.1 In line with our findings, a previous study33 documented lower proprioceptive acuity in patients with severe OA than in patients with mild OA.

Some research has shown that proprioceptive deficits are correlated with Western Ontario and McMaster Universities Osteoarthritis Index scores.1 Sharma et al34 found that patients with unilateral knee OA had poorer proprioception than controls in both knees with no clinical or radiographic evidence of OA in either knee. Based on these findings, they speculated that impaired proprioception does not exclusively result from local disease in knee OA.

In most of the above-mentioned previous studies, the severity of knee OA was radiographic grade 2 or higher. In our study, we had to exclude patients with grade 4 OA because their severe pain precluded exercise on the cycle ergometer. In contrast with most other studies, our investigation included many knees with grade 1 disease. Thus, our study compared the reposition errors of patients with mild-to-moderate bilateral OA of the knees with the reposition errors of normative controls. Had we included patients with more severe OA, we might have observed significantly greater reposition error in the group with OA. In fact, we found that the grade 3 OA patients had greater right-knee reposition error than the patients with grade 1 OA at both time points. The data do not provide any explanation for why this would apply only to the right knee, however. This could speculatively be related to dominant side.

Research has shown that periarticular muscle strengthening exercise has a clear positive effect on proprioception.35 In our study, we found that the muscles supporting the knee joint were weaker in the patient group than in the control group. Speculatively, reflex inhibition because of pain might be responsible for this weakness, or this group might not be exercising as much as a non-OA subject population, thus leading to weaker muscle strength. Analysis showed, however, that neither muscle strength nor ISK score nor mediolateral knee instability was correlated with reposition error in the OA patients. This can be explained by the presence of factors such as proper muscle and ligament balance, pain, and proprioception, all of which play individual roles in the pathogenesis of OA but also influence each another. Taking all these factors into consideration, reposition error, as we assessed in our study, may not be different between groups, but still there might be proprioceptive deficits in the OA group, because we cannot assume that only reposition error will reflect any deficit in proprioception. As well, none of our patients had grade 4 OA, and mild-to-moderate disease is another likely reason why we found no relationship between muscle strength and reposition error.

There have also been conflicting findings with respect to the effects of periarticular muscular fatigue on proprioception. Marks and Quinney36 found that proprioception was not affected by a fatigue protocol involving concentric and eccentric quadriceps contractions. In another study where shoulder proprioception was measured in healthy subjects after light and hard exercise, the acuity of movement sense was reported to be significantly reduced in the setting of localized muscle fatigue, and that the effect was more pronounced with hard exercise, and in women.16 In our study, we observed no reduction of reposition error in either the patient group or control group after fatiguing exercise. Indeed, there was a trend toward smaller reposition error after exercise, but the difference was not statistically significant. This partial but not significant increase in proprioception could have been a result of learning. As well, the exercise in our study was only of moderate intensity, and this could also explain this result. Our OA patients could not tolerate intense exercise for more than 5 minutes. To standardize the conditions in both groups, we had all subjects perform 5 minutes of mild-to-moderate intensity exercise on the cycle ergometer. A limitation of the study here is that we did not use the Borg rating of perceived exertion or any percentage of maximum voluntary contraction of muscle groups. We preferred to produce a standardized fatigue based on time, pedaling rate, and heart rate. In this procedure, there may have been patients and/or subjects who performed the exercise, but yet were not truly fatigued. The ways in which fatigue affects proprioception could be more clearly evaluated in a group of healthy volunteers who can tolerate harder and longer exercise, and by using a more definite way of producing fatigue.

Different studies have assessed proprioceptive acuity using different methods. Threshold of detection of passive movement has been used34 for assessment of joint position sense in the normal and pathologic knee joint. The method of passive joint position setting and analog reproduction using a goniometer has been used, with reproducible results, in other studies.37, 38, 39 In our study, we did not test the threshold to detection of passive knee motion, nor did we ask subjects to reproduce a remembered angle of flexion. Our method of measuring proprioception, as described before, was in between the 2 other methods used in previous studies, and the reliability analysis revealed that the reliability was not excellent, but was fair to good.

Conclusions 

Our data revealed no difference in reposition error between patients with bilateral knee OA and healthy controls. The results, however, do suggest that patients with more advanced (grade 3) knee OA have greater reposition errors of the right knees than those with less severe (grade 1) disease. Investigations that include patients with more severe (grade 4) OA might show reduced proprioceptive acuity compared with normative controls. Mild-to-moderate intensity exercise does not affect reposition error in patients with knee OA or in healthy subjects. Indeed, although not significant, there might be a trend toward smaller error (which would indicate better proprioception, not worse) after exercise, because of learning.