606 The Journal of Arthroplasty Vol. 18 No. 5 August 2003 The goals of this study are to measure and com- patient performed 3 seconds of maximal knee ex-pare knee strength in control subjects (no TKA) and tension (concentric quadriceps muscle contraction)in subjects with a clinically well-functioning TKA immediately followed by 3 seconds of maximaland correlate those measurements to categorical knee ﬂexion (concentric hamstrings muscle con-patient variables and clinical outcomes. traction). There was a 30-second rest period be- tween testing at each position. During testing, a computer monitor displayed a real-time column Materials and Methods graph of the generated torque. The test subjects were allowed to observe this graph as feedback inSubjects an attempt to enhance effort.After obtaining Institutional Review Board ap- At each position, peak torque values (foot-proval and informed consent, 52 control knees (no pounds) of ﬂexion (hamstrings) and extensionTKA) in 31 volunteer subjects (16 women and 15 (quadriceps) were recorded and then used to calcu-men) were evaluated. All control knees were clin- late the hamstring to quadriceps (H/Q) ratios. Theically normal: no pain or other limitation. For this ratio of knee ﬂexion strength to knee extensionreason, not all knees in control subjects were in- strength, the so-called H/Q ratio (hamstrings/quad-cluded. Demographics of control subjects are pro- riceps), is an established method to assess relativevided in Table 1. strength of the muscle groups . Nineteen patient volunteers with a total of 32knee arthroplasties were recruited because the ar- Statistical Analysisthroplasties were clinically well-functioning, andthe patient had no physical or mental condition that The statistical analysis was performed using thewould prohibit or inhibit participation. The out- Stata 5.0 software (Stata, College Station, TX). Dif-come of the TKA was evaluated using the Knee ferences between groups were compared using aSociety Clinical Rating System . All TKAs were 2-sample Student’s t-test. The outcome measurescemented and posterior-stabilized, with a cemented (isometric ﬂexion and extension torques and H/Qall-polyethylene patellar component. All patients ratios) were adjusted for patient characteristicswere at least 2 years after surgery (average, 2.8 (age, gender, weight, height, and BMI) using ayears; maximum, 6 years). Thirteen subjects had step-wise multivariate regression analysis. The con-bilateral TKAs. Demographics of TKA subjects are trol subjects were younger (P .0001), tallerprovided in Table 1. (P .09), lighter (P .1), and had lower BMI (P .008) than the subjects with a TKA.Test Protocol In addition to the step-wise multivariate analysis, we also compared subsets of matched patients. TenUsing a LIDO Active Dynamometer (LIDO 2.1 control subjects (7 women, 3 men) and 16 subjectsmodel 200 300 A; Loredan Biomedical, Davis, CA), with TKAs (12 women, 4 men) were selected basedisometric peak extension and ﬂexion torques were on similarities in age, height, weight, and BMI. Formeasured from 0° to 90° of knee ﬂexion. the 10 control subjects (15 knees), the average age To warm-up for testing, subjects walked on a was of 62.0 years (range, 51.4 –72.2 years; SD, 7.3treadmill at a moderately vigorous rate (2.5 to 3.5 years), the average height was 168.8 cm (range,miles per hour) for 5 minutes. Subjects were then 153.7–188.0 cm; SD, 11.6 cm), the average weightseated on the LIDO test apparatus and stabilized was 82.4 kg (range, 56.4 –106.4 kg; SD, 18.3 kg),around the pelvis and mid-thigh (Fig. 1). With the and the average BMI was 28.9 (range, 21.9 –38.2;knee ﬂexed to 90°, the center of rotation of the SD, 5,9). For the 16 subjects with TKAs (25 knees),LIDO lever arm was aligned in parallel with the the average age was of 65.1 years (range, 50.4 –78.9femoral condyles. The lower extremity was at- years; SD, 8.1 years), the average height was 168.0tached to the LIDO lever arm by way of a padded cm (range, 147.3–198.1 cm; SD, 12.6 cm), the av-cuff with a fastener just above the ankle. Subjects erage weight was 87.6 kg (range, 55.9 –101.8 kg;were instructed on how to perform the tests, em- SD, 12.9 kg), and the average BMI was 31.1 (range,phasizing the importance of maximum effort dur- 23.4 –36.9; SD, 4.4). There were no signiﬁcant dif-ing the test and encouraged during the test to push ferences, in age, height, weight, or BMI betweenas hard as they could. these 2 subgroups. Isometric testing was performed at 7 positions, Correlations between patient characteristics andbeginning with 90° of ﬂexion and moving to full outcome measures were obtained using univariateextension in 15° increments. At each position, the and multivariate regression analyses. A Pearson
Knee Strength After Total Knee Arthroplasty • Silva et al. 607 product-moment coefﬁcient of correlation (r) 181.0 (167.6–198.1) [8.3] 166.5 (147.3–198.1) [12.6] 161.3 (147.3–170.2) [7.0] 186.1 (177.8–198.1) [9.1] 93.5 (74.1–100.0) [13.0] 67.1 (50.4–78.9) [12.0] 27.0 (23.4–30.7) [3.2] greater than 0.75 indicated a very good to excellent 4) correlation; 0.51 to 0.75 indicated a moderate to good correlation; 0.25 to 0.50 indicated a fair degree Men (n of correlation; and equal or less than 0.25 was considered as little or no correlation. A P value of .05 was considered statistically signiﬁcant. 19) 85.9 (55.9–101.8) [12.3] Results Patients Undergoing TKA (n 67.3 (53.0–83.2) [8.6] 33.1 (25.8–45.9) [5.0] 15) Isometric Extension Torque Women (n Isometric extension peak torque values decreased as the knee came into extension (Table 2). There was a high degree of variability in isometric exten- sion peak toque at all positions tested. On average, women control subjects generated 40.4% lower 87.5 (55.9–101.8) [12.5] isometric extension peak torque values than men 67.3 (50.4–83.2) [9.1] 31.8 (23.4–45.9) [5.3] controls (P .0001). Regression analysis indicates a 19) correlation between average isometric extension peak torque values and height (r 0.67, P .0001) All (n and age (r 0.82; P .0001) in control subjects.Table 1. Subject Demographics On average, women TKA patients generated 52.4% lower isometric extension peak torque values than men TKA patients (P .0001). Height and weight were positively correlated to isometric extension 85.9 (66.4–106.4) [13.1] 38.1 (20.1–72.2) [17.3] 26.2 (21.6–34.0) [3.3] peak torque values in subjects with a TKA (r 0.82; 15) P .0001 and r 0.47; P .007, respectively). In ter- minal extension (30°, 15°, and 0° of ﬂexion), older Men (n TKA patients ( 70 years) generated lower isomet- ric extension peak torque values than younger TKA patients (24.2%, P .05; 26.5%, P .05; 29.0%, P .05, respectively). After adjustments in patient characteristics, iso- 164.4 (152.4–177.8) [18.2] Abbreviations: BMI, body mass index; TKA, total knee arthroplasty. 31) 74.3 (53.6–133.6) [22.6] metric extension peak torque values in control sub- 41.7 (15.9–71.0) [18.2] 27.6 (20.4–52.2) [8.7] 16) jects were, on average, 9.7 ft-lb (95% CI, 0.7 to NOTE: Values are given as Mean (range) [standard deviation]. Control Subjects (n 19.4; P .05) higher than those in TKA patients. A Women (n difference in adjusted isometric extension peak torque values between control subjects and TKA patients was evident at all positions tested (Table 2). Isometric Flexion Torque 172.4 (152.4–198.1) [11.4] 79.8 (53.6–133.6) [19.1] Isometric ﬂexion peak torque values increased with 40.0 (15.9–72.2) [17.6] 26.9 (20.4–52.2) [6.6] knee extension (Table 2). There was a high degree 31) of variability in isometric ﬂexion peak torque at all positions tested. On average, women control sub- All (n jects generated 43.6% lower isometric ﬂexion peak torque values than men controls (P .0001). Iso- metric ﬂexion peak torques were correlated to height (r 0.71, P .0001), age (r 0.51, P .0001) and weight (r 0.38, P .005). On average, women Weight (kg) Height (cm) TKA patients generated 44% lower isometric ﬂex- Age (y) ion peak torque values than men (P .0001). In BMI TKA patients, age was not correlated to the average
608 The Journal of Arthroplasty Vol. 18 No. 5 August 2003 Fig. 1. Subjects were seated on the LIDO test apparatus and stabilized around the pelvis and mid-thigh. Table 2. Isometric Extension Torque, Isometric Flexion Torque, and Hamstring to Quadriceps Ratio Control Raw Difference 95% CI for the P value for the All Knees Knees TKAs Between Difference Between Adjusted Adjusted (n 84)* (n 52)* (n 32)* Groups Groups† Difference DifferenceIsometric extension torque (ft-lb)90° 109.3 (59.5) 135.2 (59.0) 67.2 (28.6) 68.0 67.9 45.7 to 90.1 .000175° 115.2 (57.2) 142.8 (51.3) 70.5 (33.2) 72.3 23.7 8.0 to 39.4 .00460° 106.6 (50.2) 129.9 (44.6) 68.7 (32.9) 61.2 18.5 5.4 to 31.6 .00645° 89.8 (38.1) 105.9 (35.2) 63.6 (26.7) 42.3 13.4 2.3 to 24.5 .0230° 69.8 (29.6) 81.5 (27.7) 50.8 (22.0) 30.7 30.7 19.2 to 42.1 .000115° 59.2 (23.2) 59.3 (22.1) 37.9 (18.5) 21.4 21.3 12.0 to 30.7 .0001 0° 35.1 (18.7) 41.1 (18.8) 25.5 (14.3) 15.6 15.6 7.9 to 23.3 .0001Isometric ﬂexion torque (ft-lb)90° 46.5 (29.0) 61.1 (27.0) 22.1 (8.6) 39.0 11.6 3.4 to 19.3 .00375° 54.8 (31.2) 70.8 (28.7) 28.8 (11.4) 42.0 15.0 7.0 to 22.9 .000160° 59.7 (32.5) 75.6 (31.0) 33.9 (12.0) 41.7 12.1 3.5 to 20.7 .00645° 63.9 (32.8) 79.0 (30.8) 39.2 (15.3) 39.8 12.2 3.0 to 21.4 .0130° 68.5 (33.2) 83.9 (31.2) 43.6 (17.5) 40.3 13.1 3.5 to 22.6 .00815° 72.4 (36.9) 88.6 (35.4) 46.0 (20.6) 42.6 9.6 0.5 to 19.7 .06 0° 69.2 (34.0) 84.2 (32.3) 44.8 (19.7) 39.4 9.1 1.7 to 19.9 .09H/Q ratio90° 0.42 (0.12) 0.46 (0.99) 0.35 (0.12) 0.11 0.11 0.06 to 0.16 .000175° 0.47 (0.12) 0.49 (0.11) 0.43 (0.13) 0.06 0.06 0.01 to 0.11 .0360° 0.56 (0.15) 0.57 (0.10) 0.54 (0.22) 0.03 0.03 0.04 to 0.10 .4445° 0.70 (0.17) 0.74 (0.13) 0.65 (0.20) 0.09 0.09 0.02 to 0.16 .0230° 1.01 (0.42) 1.08 (0.46) 0.92 (0.32) 0.16 0.16 0.03 to 0.34 .115° 1.42 (0.39) 1.49 (0.29) 1.32 (0.50) 0.17 0.24 0.06 to 0.43 .01 0° 2.20 (0.97) 2.18 (0.64) 2.22 (1.36) 0.04 0.04 0.48 to 0.40 .86 *Mean (SD). †Adjusted by patient characteristics. ‡Degrees of ﬂexion. Abbreviations: TKAs, total knee arthroplasties; CI; conﬁdence interval; H/Q, hamstring to quadriceps ratio; SD, standard deviation.
Knee Strength After Total Knee Arthroplasty • Silva et al. 609 Table 3. Knee Strength Data Summary by Matched Subgroup 90° 75° 60° 45° 30° 15° 0°Isometric extension torque (ft-lb)Control knees (n 15) 83.6 (30.5) 100.8 (36.7) 92.6 (32.4) 81.1 (29.3) 59.7 (24.0) 44.6 (16.4) 30.1 (13.6)TKAs (n 25) 67.9 (32.2) 69.8 (37.1) 68.9 (36.9) 63.9 (30.0) 51.6 (24.3) 39.2 (20.2) 26.4 (15.5)Isometric ﬂexion torque (ft-lb)Control knees (n 15) 37.1 (16.0) 47.4 (21.4) 50.3 (21.6) 56.6 (23.2) 62.1 (22.5) 64.4 (27.8) 62.2 (24.7)TKAs (n 25) 22.6 (8.8) 28.4 (12.2) 33.3 (12.6) 38.8 (16.4) 44.0 (18.7) 46.5 (22.5) 44.4 (21.2)H/Q RatioControl knees (n 15) 0.45 (0.11) 0.47 (0.11) 0.54 (0.09) 0.69 (0.11) 1.21 (0.81) 1.42 (0.34) 2.25 (0.94)TKAs (n 25) 0.35 (0.12) 0.44 (0.14) 0.55 (0.23) 0.64 (0.20) 0.92 (0.30) 1.28 (0.46) 2.17 (1.42) NOTE: Values are given as mean (standard deviation). All groups are matched subgroups. ° Degrees of ﬂexion. Abbreviation: H/Q, hamstring to quadriceps.isometric ﬂexion peak torque (r 0.16, P .4) but and BMI was found (r 0.44, P .007); more obeseheight (r 0.62, P 0.0001) and weight (r 0.44, patients have relatively lower quadriceps strength.P .01) were. Multivariate regression analysis indi- After adjustments in patient characteristics, H/Qcates that the average isometric ﬂexion peak torque ratios in control subjects were, on average, 0.8is strongly correlated to height (r 0.72, P .009). (95% CI, 0.03 to 0.2; P .2) higher than those inIsometric knee ﬂexion and extension strength were TKA patients. A difference in adjusted H/Q ratioshighly correlated in all subjects (r 0.95, P .0001). between control subjects and TKA patients was After adjustments in patient’s characteristics, iso- evident at all but 2 of the position tested (60° andmetric ﬂexion peak torque values in control sub- 0°) (Table 2).jects were, on average, 12.1 ft-lb (95% CI, 4.2 to20.0; P .003) higher than those in TKA patients. A Matched Subgroupsdifference in adjusted isometric ﬂexion peak torquevalues between control subjects and TKA patients Isometric extension peak torque values in TKA pa-was evident at all positions tested (Table 2). tients were highly variable and, on average, 21.2% lower than those from control subjects, throughout the motion arc (P .09) (Table 3). A reduction inH/Q Ratios average isometric extension peak torque of 18.8% (P .1), 30.7% (P .01), 25.6% (P .05), and 21.2%For all subjects, isometric H/Q ratios increased with (P .08) was observed at 90°, 75°, 60°, and 45° ofknee extension (Table 2). There was a high degree ﬂexion, respectively, in the TKA group (Fig. 2).of variability in isometric H/Q ratios at all positionstested. Univariate and multivariate regression anal-ysis showed no correlation between average iso-metric H/Q ratios and other variables such as age,gender, weight, height, or BMI. No signiﬁcant dif-ferences in isometric H/Q ratios were found be-tween men and women or between younger andolder subjects. There was a trend for the isometric H/Q ratio toincrease near terminal extension as patient ageincreased. Older TKA subjects ( 70 years old) hadisometric H/Q ratios that were 18.3% (P .15),22.9% (P .1), and 46.3% (P .07) higher thanyounger TKA subjects at 30°, 15°, and 0° of ﬂexion,respectively. Univariate regression analysis indi-cates that BMI and height are correlated to isomet-ric H/Q ratios in TKA patients (r 0.35, P .05, and Fig. 2. Isometric extension. Knee extension strength wasr 0.42, P .02, respectively). At 90° of ﬂexion, a generally lower in subjects with a TKA. Error bars indi-stronger correlation between isometric H/Q ratio cate standard deviation.
610 The Journal of Arthroplasty Vol. 18 No. 5 August 2003 a function of gender, age, height, and degree of obesity. Although knee strength can be restored to normal levels after a TKA, it is uncommon. In the present study, average isometric knee extension and ﬂexion strength of TKA subjects was more than 30% lower than matched control subjects (P .01). Regardless of statistical analyses, such reductions in strength have practical signiﬁcance . The reduc- tion in muscle strength seen in TKA subjects is probably the result of muscle atrophy caused by disuse before the TKA that has not been recovered after the TKA .Fig. 3. Isometric ﬂexion. Knee ﬂexion strength was con- Knee strength is an important factor in the clin-sistently lower in subjects with a TKA. Error bars indicate ical outcome after TKA. In the current study, westandard deviation. found that isometric extension peak torque and the H/Q ratio had a strong correlation with the Knee Society Functional Score (r 0.57, P .004 and r 0.78, P .0001, respectively). The need for ad- equate extensor mechanism function is a prerequi- Isometric ﬂexion peak torque values in patients site for common activities of daily living such aswith a TKA were highly variable and, on average, climbing stairs, so it is logical that quadriceps32.2% lower than those from control subjects strength is associated with the functional score.throughout the motion arc (P .004) (Table 3). Re- Caution should be taken in assigning any cause andduction of 39.5% (P .001), 40.0% (P .001), 33.9% effect relationship. It could be argued that better(P .003), 31.4% (P .007), 29.2% (P .009), functioning knees allow more vigorous activity, and27.8% (P .03), and 28.6 (P .02) was found at 90°, greater quadriceps strength is a result of higher75°, 60°, 45°, 30°, 15°, and 0°, respectively, in the activity.TKA group (Fig. 3). Isometric H/Q ratios in subjects Compared with normal controls, a signiﬁcant re-with TKA were, on average, 9.5% lower than those duction in ﬂexion strength was observed at everyfrom control subjects, throughout the motion arc point on the arc of motion tested. This may be the(P .3). result of surgical technique, the design and result- ant biomechanics of total knee prostheses, theKnee Society Scores quadriceps-focused rehabilitation of our TKA pa-The average Knee Society (KS) Clinical Score was tients, the postoperative activities of the patients, or92 (range, 76 –100) and the average KS Functional a combination of these or other factors.Score was 92 (range, 70 –100). Average isometric As detected by the KSS, relative hamstring weak-extension or ﬂexion strength did not show a corre- ness had a lower level of functional signiﬁcancelation with the clinical score (r 0.09, P .66 and (r 0.33, P .1). The absence of a stronger correla-r 0.15, P .46, respectively). The functional tion between hamstring weakness after TKA andscores were, however, positively correlated to the the KSS is a reﬂection of the relatively low-levelaverage isometric extension peak torque (r 0.57, activities assessed by the KSS. Hamstring weaknessP .004) and to the average isometric ﬂexion peak would become apparent in more vigorous activitiestorque (r 0.33, P .1). The clinical score was not such as fast walking, uphill walking, and running.correlated to the average isometric H/Q ratio In a study of patients with a torn anterior cruciate(r 0.2, P .3). Functional scores were negatively ligament, it was found that subjects whose ham-correlated to the average isometric H/Q ratio string strength was equal to or greater than the(r .78, P .0001); in other words, relatively quadriceps strength in the involved limb returnedgreater quadriceps strength was associated with a to higher levels of participation in sports than didbetter functional score. subjects whose hamstring strength was less than their quadriceps strength . In the present study, nearly 70% of the patients Discussion were women. Although this is biased toward women, the female to male ratio for TKA is approx-As would be expected in a study of human perfor- imately 3 to 1 . Because each subject is theirmance, there is great variability in knee strength as own control, H/Q ratios are less affected by patient
Knee Strength After Total Knee Arthroplasty • Silva et al. 611characteristics than the absolute values of extension 2. Huang CH, Cheng CK, Lee YT, Lee KS: Muscleor ﬂexion strength. In general, age, gender, weight, strength after successful total knee replacement: a 6-height, and BMI did not affect the H/Q ratio. How- to 13-year followup. Clin Orthop 328:147, 1996ever, within the TKA group, women, older subjects, 3. Berman AT, Bosacco SJ, Israelite C: Evaluation ofand relatively obese subjects tended to have higher total knee arthroplasty using isokinetic testing. Clin Orthop 271:106, 1991isometric H/Q ratios (relatively lower quadriceps 4. Aagaard P, Simonsen EB, Trolle M, et al: Isokineticstrength) than other subjects, with greater variabil- hamstring/quadriceps strength ratio: inﬂuence fromity in terminal extension. Having shown a positive joint angular velocity, gravity correction and contrac-correlation between extension strength and func- tion mode. Acta Physiol Scand 154:421, 1995tional outcome, these data indicate a need for more 5. Bolanos AA, Colizza WA, McCann PD, et al: A com-aggressive rehabilitation, especially in these sub- parison of isokinetic strength testing and gait analysisgroups. in patients with posterior cruciate-retaining and sub- Compared with rehabilitation protocols after ath- stituting knee arthroplasties. J Arthroplasty 13:906,letic injuries of the knee, structured rehabilitation 1998after TKA is inferior in both intensity and duration. 6. Huang CH, Lee YM, Liau JJ, Cheng CK: ComparisonAfter anterior cruciate reconstruction, 52 weeks of of muscle strength of posterior cruciate-retained ver-structured rehabilitation has been recommended to sus cruciate-sacriﬁced total knee arthroplasty. J Ar-reliably return the patient to a preinjury level of throplasty 13:779, 1998function . Because TKA is being performed on 7. Kannus P, Jarvinen M: Knee ﬂexor/extensoryounger and more active patients who desire a strength ratio in follow-up of acute knee distortion injuries. Arch Phys Med Rehabil 71:38, 1990higher level of function, the demands and expecta- 8. Murray MP, Gardner GM, Mollinger LA, Sepic SB:tions of the arthroplasty are increasing. Rehabilita- Strength of isometric and isokinetic contractions:tion after TKA needs to evolve to meet these rising knee muscles of men aged 20 to 86. Phys Therdemands and expectations. The aggregate data in- 60:412, 1980dicate that knee strength is an important element in 9. Seto JL, Oroﬁno AS, Morrissey MC, et al: Assessmenthigher function. Similar to in other patients with of quadriceps/hamstring strength, knee ligament sta-anterior cruciate ligament– deﬁcient knees, greater bility, functional and sports activity levels ﬁve yearsemphasis is needed on hamstring strengthening. after anterior cruciate ligament reconstruction. Am J Knee strength can be restored to normal levels Sports Med 16:170, 1988after TKA, but there is great variability. These data 10. Zakas A, Mandroukas K, Vamvakoudis E, et al: Peaksuggest a need for more aggressive rehabilitation torque of quadriceps and hamstring muscles in bas-after TKA, especially in women, older patients, and ketball and soccer players of different divisions.more obese patients. J Sports Med Phys Fitness 35:199, 1995 11. Insall JN, Dorr LD, Scott RD, Scott WN: Rationale of the Knee Society clinical rating system. Clin Orthop Acknowledgment 248:13, 1989 12. Giove TP, Miller SJ, Kent BE, et al: Non-operative The authors thank Mylene A. de la Rosa, BS, for treatment of the torn anterior cruciate ligament.her assistance in the preparation of this manuscript J Bone Joint Surg Am 65:184, 1983and Frederick J. Dorey, PhD for his assistance with 13. Kouyoumdjian JA: Neuromuscular abnormalities inthe statistical analyses of the data. disuse, ageing and cachexia. Arq Neuropsiquiatr 51: 299, 1993 14. Knutson K, Lewold S, Robertsson O, Lidgren L: The References Swedish knee arthroplasty register: a nation-wide study of 30,003 knees 1976-1992. Acta Orthop Scand 1. Healy WL, Wasilewski SA, Takei R, Oberlander M: 65:375, 1994 Patellofemoral complications following total knee ar- 15. Podesta L, Sherman MF, Bonamo JR, Reiter I: Ratio- throplasty: correlation with implant design and pa- nale and protocol for postoperative anterior cruciate tient risk factors. J Arthroplasty 10:197, 1995 ligament rehabilitation. Clin Orthop 257:262, 1990