Electromyographic analysis of exercises proposed for differential activation of medial and lateral quadriceps femoris muscle components.

Weakness and atrophy of the quadriceps femoris muscle (QF) are commonly associated with a wide variety of traumatic and pathological conditions involving the knee, and active exercises aimed at strengthening the QF have long played a prominent role in the rehabilitation of patients with knee disorders. A wide variety of static and dynamic exercise regimens have been shown to be effective for increasing the extensor torque-producing capabilities of the QF group as a whole,[1-3] and a number of studies have compared the relative effectiveness of various exercises in terms of the extent and rapidity of those increases.[2,4-6] Though the rate and magnitude of gains in QF torque production are undeniably important considerations in the choice of an appropriate strengthening regimen, it has become apparent that other considerations should enter into that choice. For example, modifications of QF strengthening protocols have been proposed for reducing stress to specific joint structures, such as the inferior surface of the patella in patients with chondromalacia patellae.[7-9]

Biomechanical analyses suggest that normal alignment and function of the patellofemoral joint appear to depend on an appropriate balance of medial and lateral forces exerted on the patella by passive structures (eg, the patellar retinacula) and by active muscular forces.[9] Imbalance of the medial and lateral forces acting on the patella is thought to be an important etiological factor in a number of patellofemoral disorders. Most commonly, the net force on the patella is directed too laterally, resulting in predisposition to lateral subluxation or dislocation of the patella and to degenerative disorders such as chondromalacia patellae.[9-11] Treatment for such problems is generally aimed at increasing the medial force relative to the lateral force acting on the patella.[8,9,11-14]

Approaches to increasing the ratio of medial to lateral forces acting on the patella have included both conservative techniques, such as biofeedback,[15,16] electrical stimulation,[7] taping,[14] and active exercise,[8,9,11,14] and surgical techniques,[17] such as lateral retinacular release or plication of the vastus medialis obliquus component (VMO).

Variations in muscle fiber orientation and attachment to the patella result in marked differences in the direction of the force vectors exerted by various QF components. Studies of muscle fiber orientation[13,18,19] suggest that the vastus lateralis muscle (VL) force component is directed 12 to 15 degrees laterally with respect to the femoral shaft, whereas the VMO force is directed 40 to 55 degrees medially and that of the vastus medialis longus component (VML) is directed 15 to 18 degrees medially. These studies supply the rationale for prescribing active strengthening exercises that preferentially activate the medial QF components, especially the VMO, to a greater degree than the lateral components. Evidence for the efficacy of these particular exercise regirnens, however, is lacking. The focus of this study is on specific active exercises that have been proposed for selective strengthening of the QF components that contribute significantly to the medially directed forces on the patella.

In the early 1950s, Wheatley and Jahnke[20] suggested that the VMO was responsible for the terminal phase of knee extension, and thus selective VMO activation could be achieved by limiting isotonic exercises to the last few degrees of knee extension. This selective action theory has since been refuted by a number of electromyographic (EMG) studies demonstrating that all QF components are active throughout the range of knee extension. (For a discussion of this controversy, readers are referred to a 1977 review by Speakman and Weisberg.[21])

More recently, exercises combining activity of the hip adductors in conjunction with knee extensor strengthening activities have been advocated as a means of preferentially activating and exercising the VMO.[14,22-24] Specific exercises proposed include straight leg raising performed with the hip in lateral rotation[22] (SLR/LR) as well as isometric hip adduction exercises in conjunction with conventional knee extensor strengthening exercises[24] such as straight leg raising combined with isometric hip adduction (SLR/ADD). The most frequently cited rationale for such exercises is an anatomical linkage between the VMO and the hip adductor muscles.[14,22-24] Based on anatomical studies indicating that many of the VMO fibers originate from the tendon of the adductor magnus muscle,[18,19] it has been suggested that concurrent activation of the hip adductors with the QF group would provide the VMO with a more stable proximal attachment, and thus facilitate preferential activation of that component of the QF group.

Though this proposed rationale seems plausible, convincing evidence that the specific exercises that have been proposed for concurrent activation of knee extensors and hip adductors (eg, SLR/LR or SLR/ADD) actually result in measurable increases in the VMO to VL activation ratio is not found in the literature. Hanten and Schulthies[23] have reported that an increased VMO to VL activation ratio results from performing isolated hip adduction exercises, suggesting a functional link between the VMO and the hip adductor muscles. Their study, however, did not directly test the SLR/LR or SLR/ADD exercises used clinically, because subjects were positioned with the knee in 60 degrees of flexion and were only instructed to perform isometric hip adduction rather than concurrent hip adduction and knee extension. Basmajian[25] and Wild and associates,[26] however, provide anecdotal evidence suggesting that the SLR/LR does not change the relative activity pattern of the QF components, though neither of those reports provide data directly supporting that claim.

In this study, quantitative EMG techniques were used to assess the relative activity levels of various QF components during conventional quadriceps femoris setting (QS) and straight leg raising (SLR) exercises and during two variations of SLR that require concurrent hip adductor activation (SLR/LR and SLR/ADD). By studying these four exercises, we were able to address the following questions:

1. Do the relative activity levels of the

various QF components remain

fixed during volitional QF strengthening

exercises, or can specific

exercises vary the relative contribution

of those components?

2. Do specific exercises that have

been advocated for preferential

strengthening of the VMO on the

basis of a proposed functional link

between activation of the hip adductor

muscles and the VMO actually

result in a greater VMO to VL

activity ratio than conventional QS

or SLR exercises?

Though the clinical efficacy of these exercise techniques must ultimately be assessed by studying a patient population, we chose a population without knee pathology for our initial study of these exercises for two main reasons. First, the anatomical linkage between the VMO and the hip adductors, which is the purported mechanism by which these exercises might enhance VMO activity, should be equally present in both healthy subjects and patients with patellofemoral dysfunction. As such, any effects on VMO activity attributable to concomitant hip adductor activation should be observable in both healthy and patient populations. Furthermore, a previous investigation of QF activity patterns during QS and SLR exercises[27] did not reveal significant differences between healthy subjects and those with knee pathology, suggesting that knee pathology does not necessarily alter the neural activity patterns of the QF group. Given these considerations, quantifying the effects of these exercises in a healthy population seems a logical first step in determining clinical efficacy.

Method

Subjects

Twelve subjects (6 women, 6 men) with no current knee pathology volunteered to participate in this study. Descriptive statistics pertaining to height, weight, and age are presented in Table 1. Each subject provided informed consent prior to participating in the study.

[TABULAR DATA OMITTED]

Electromyographic Procedures

Electromyographic activity was recorded using a bipolar configuration and silver-silver chloride surface electrodes. Following skin preparation by removal of excess hair and cleansing with isopropyl alcohol, pairs of electrodes were applied to the skin over the VMO, VML, VL, and rectus femoris muscle (RF). A common reference electrode was placed on the patella, and the thigh was wrapped in an elastic bandage to prevent movementrelated artifacts. Electrode placement was standardized based on a modification of the technique proposed by Zipp[28] for the vastus medialis muscle. Figure 1 illustrates the typical location and orientation of these electrode pairs.

Electromyographic signals were differentially amplified using high-performance bioamplifiers (Model 575-03)(*) having a bandwidth of 0.1 to 20,000 Hz and a maximum gain of 100,000. After conventional band-pass filtering (100-1,000 Hz), the EMG signals were full-wave rectified and smoothed using a contour-following integrator (Model S76-01)(*) (time constant 15 milliseconds) to obtain the linear envelope of the EMG signal. The rectified and smoothed signals were then digitized[(dagger)] on-line at a sampling rate of 500 Hz, and the EMG data were stored on magnetic media for later analysis.

Exercise Procedure

Subjects were positioned supine on a standard treatment table. At the foot of the table, a cardboard background with two foot silhouettes placed 25 cm above the table surface served as a target during the SLR exercises. One silhouette was vertical so that the subject would maintain the hip in 0 degrees of rotation for the SIR and SLR/ADD exercises, and the second outline was rotated 45 degrees laterally to serve as the target for the SLR/LR exercise (Fig. 2). To allow for visual alignment of the foot with the target silhouette while maintaining constant head and body position, the subject's head was supported with the neck in slight flexion.

All exercises were performed with the right lower extremity. The contralateral lower extremity was placed in 45 degrees of hip and knee flexion, and the pelvis was stabilized with a restraining strap across the iliac crests. During all variations of the SLR exercise (SLR, SLR/ADD, and SLR/ILR) resistance was added by applying ankle weights equivalent to 5% of body weight. For the SLR/ADD exercise, an external torque tending to abduct the hip was applied via a rope and pulley apparatus attached just distal to the malleoli (Fig. 2). The apparatus was adjusted for each subject so that it provided a lateral force (equivalent to 5% of body weight) acting at a right angle to the tibia during the isometric hold phase of the exercise.

Standardized verbal instructions were given prior to the first repetition of each exercise. These instructions were provided by the same investigator for all 12 subjects. The four exercises performed in this study were:

1. Quadriceps setting (QS): Subjects

were instructed to maximally activate

their thigh muscles in order to

straighten their knee and were

provided with the prompt "tighten"

and a two count, by which

time they were to attain and hold a

maximal contraction. At the completion

of the 5-second data-collection

period, subjects were

directed to relax.

2. Straight leg raising (SLR): Subjects

were instructed to perform a maximal

QS exercise prior to the lifting

phase of this exercise. Subjects

were provided with the prompts

"tighten ... lift" and an additional

two count, by which time they were

to align their foot with a target foot

silhouette. Subjects were instructed

to maintain that position until the

investigator indicated the completion

of the trial, 5 seconds after

reaching the designated position.

3. Straight leg raising with lateral rotation

of the hip (SLR/LR): Subjects

began in and maintained 45 degrees

of lateral rotation at the hip

for this exercise. Instructions were

identical to those for the SLR exercise

except that subjects were directed

to align their foot with a

background target rotated 45 degrees

clockwise from vertical.

4. Straight leg raising with isometric

hip adduction (SLR/ADD): This

exercise was identical to the SLR

exercise except that an external

abduction torque acted on the hip

via a rope and pulley apparatus, as

illustrated in Figure 2.

The exercise session was initiated with five repetitions of the QS exercise. The subject then sequentially performed one repetition each of the SLR, SLR/LR, and SLR/ADD exercises until a total of five repetitions of each SLR variation was completed. The session was concluded with an additional five QS repetitions. Subjects were allowed to perform a practice trial of each exercise before data collection was begun. A rest period of approximately 30 seconds was given between repetitions in order to minimize fatigue, and subjects were instructed to notify the investigators of any fatigue or discomfort associated with the exercises.

Data Analysis

The magnitude of EMG activity for each repetition of each exercise was quantified by digital integration of the smoothed and rectified signals over the 5-second isometric phase of the exercise, as illustrated in Figure 3. Within-subject reliability of the EMG measures was established by calculating intraclass correlation coefficients (ICC[1,1]) using a one-way analysis of variance (ANOVA) model, as described by Baumgartner.[29] Two aspects of the experimental protocol were designed to exclude the possibility of fatigue, practice, or learning effects. First, we attempted to eliminate any differential effects of fatigue by having subjects perform single repetitions of the three SLR variations sequentially rather than in a single block of five repetitions and by allowing for rest between repetitions. Second, two sets of the QS exercise were performed, one at the start of the session and one at the end.

Linear regression analysis[(doubler dagger)] of mean EMG activity for the initial and final sets of the QS exercise was used to examine the possibility that the results might be affected by factors such as fatigue, learning, and changes in electrode impedance. In order to allow for comparisons across subjects, mean EMG values for each QF component were then normalized relative to the maximum value obtained for that muscle during any of the exercises performed by that subject. Based on the normalized mean EMG activity, medial to lateral activity ratios (VMO to VL and VML to VL) were calculated for each exercise. A repeated-measures one-way ANOVA was used to test for between-group (ie, between-exercise) effects, and post hoc multiple comparisons were made using paired-samples Bonferroni confidence intervals[30] with the alpha level established at .05.

Results

Intraclass correlation coefficients of within-subject reliability for each of the four QF components for five sets of exercises are shown in Table 2. The ICCs for all comparisons were greater than .92, with a mean reliability coefficient of .979 over all 20 comparisons. Because this model treats all sources of variation other than differences among subjects as a lack of reliability, the high correlation coefficients indicate a high degree of intrasubject reliability and we believe justify our use of the mean value of the five repetitions of each exercise for each subject in subsequent analyses. Linear regression of initial QS EMG values on final QS EMG values also demonstrated a high degree of repeatability, as indicated bv the high correlation coefficient (R = .97) and a slope of very nearly 1.00 (y = 0.998 x -0.0004), indicating that effects attributable to fatigue, learning, or changes in electrode impedance were minimal.

Table2. Intraclass Correlation Coefficients for Each of the
Quadriceps Femoris Muscle Components Over Each of the Five Sets
of Exercises
               Muscle Component(b)
Exercise(a)    VMO    VML    VL     RF
QS (initial)   .992   .996   .992   .967
SLR            .950   .986   .983   .991
SLR/LR         .971   .982   .971   .975
SLR/ADD        .977   .991   .991   .982
QS (final)     .981   .992   .986   .928
(a)QS = quadriceps femoris setting; SLR = straight leg raising;
SLR/LR = straight leg raising with hip laterally rotated;
SLR/ADD = stravight leg raising combined with isometric hip
adduction,
(b) VMO = vastus medialis obliquus; VML = vastus medialis
longus; Vl=vastus lateralis; Rf=rectus femoris.

The data for each subject and each muscle group were normalized with respect to the highest mean EMG value recorded from that muscle group during any of the exercise conditions. For 10 of the 12 subjects, the QS exercises elicited the greatest mean EMG activity in each of the single-joint QF components (VMO, VML, and VL). Figure 4 illustrates the relative activity of each of the four QF components tested during each set of exercises. Figure 4 shows that each of the single-joint QF components (VMO, VML, and VL) demonstrated similar patterns of activity across the various exercises, whereas the two-joint RF exhibited a very different pattern of activity. There was a significant (P<.05) between-group (ie, between-exercise) effect for each of the four QF components studied. Post hoc analyses using paired-sample Bonferroni confidence intervals[30] revealed that for the VMO, VML, and VL, both the initial and final QS exercises elicited significantly greater activity than any of the three SLR variations (P<.05). There were no significant differences between the mean activity levels of any of the single-joint knee extensors when comparing the initial and final QS exercises or when comparing the three variations of the SLR exercise. In comparison with the single-joint QF components studied, the two-joint RF exhibited marked differences in activity patterns. The SLR elicited significantly greater RF activity than did the initial QS or SLR/ADD exercises (P<.05), and the SLR/LR exercise elicited significantly greater RF activity than did the SLR/ADD exercise or either set of QS exercises (P<.05).

In order to more closely examine the relative activity of medial and lateral QF components, VMO to VL and VML to VL ratios were calculated using the normalized mean EMG data. Figure 5 illustrates the comparison of normalized mean medial to lateral EMG ratios for both the oblique and longitudinal components of the vastus medialis muscle. A one-way ANOVA revealed that there were no significant exercise-dependent effects on either the VMO to VL or VML to VL ratio.

Discussion

Rationale for Design and

Implementation

This study was designed to quantify the relative activity levels of various QF components during exercises that have been proposed for preferential strengthening of the VMO. More generally, the study was designed to test the notion that a functional link exists between activity of the hip adductor muscles and a relative increase in activity of medial QF components.

The methodology chosen for this study was based on the implicit assumption that the relative magnitude of the EMG for each QF component during a given exercise is indicative of the strength gains that may be expected in that component as a result of training with that exercise. That assumption appears justified for several reasons. All data for each subject were collected in a single session in order to ensure that electrode placement and amplification were unchanged across exercises, and the data were appropriately normalized before analyzing data. Intraclass correlation coefficients demonstrated the trial-to-trial reliability of the EMG measures, and EMG magnitudes for initial and final QS sets showed virtually no change, indicating that effects attributable to fatigue, learning, or changes in skin-electrode interface characteristics were negligible in this study. Most importantly, all comparisons of EMG magnitude were carried out only for the isometric phase of each exercise, during which the magnitude of the rectified and smoothed EMG signal has been shown to be proportional to the force exerted by a given muscle[31,32] and thus, in accordance with the overload principle, to the strength gains to be expected from the performance of that exercise. The specific exercises that were chosen for study included two of the most commonly prescribed exercises for general QF strengthening, QS and SLR,[11] as well as two variations of the SLR that elicit concurrent hip adductor activity and that have been suggested as means of preferentially strengthening the VMO to a greater degree than the VL. The QS exercises were included because they are commonly used in the treatment of patellofemoral disorders and because our pilot studies, as well as data from previous investigations,[27,33] indicated that the QS exercise consistently elicits the greatest activity of the single-joint knee extensors in most subjects.

Comparison of QF activity across the three variations of the SLR provides the most direct test of the proposed functional link between VMO activity and hip adductor activity. The SLR/LR exercise has been specifically suggested as a means of preferentially strengthening the VMO[22] and has the advantage of being a relatively simple modification of the SLR exercise for the patient using a weighted boot or ankle-cuff weight for added resistance. Laterally rotating the hip, however, not only alters the muscular torque requirements at the hip, but also reduces the magnitude of knee extensor torque required to maintain complete knee extension. In the position used in this study (45 [degrees] of lateral hip rotation), the forces attributable to gravity acting on the shank, foot, and cuff weight contribute approximately equally to flexion and abduction torques about the knee. The abduction torque about the knee could be opposed by passive structures, such as the medial collateral ligament, rather than by active muscle forces. Direct comparison of RF EMG activity between SLR and SLR/LR conditions is further confounded by the possibility of length or moment arm changes for that two-joint muscle. For the reasons outlined above, we included the SLR/ADD exercise in this study. By simply adding to the SLR exercise a laterally directed force at the ankle, activation of the hip adductors could be achieved without any alteration in the required hip flexor or knee extensor torques and without altering either the length or moment arm of any of the QF components. Thus, comparison of medial to lateral activity ratios elicited by the SLR and SLR/ADD exercises provides the most direct test of the notion that VMO activity is enhanced by concurrent hip adductor activity.

The decision to monitor EMG activity of two portions of the vastus medialis was based on anatomical evidence of differing fiber orientation and nerve supply to the inferior (VMO) and superior (VML) portions of that muscle.[13,18,19] Subtle differences in the activity patterns of the VMO and VML, illustrated by the nonsignificant differences between VMO to VL and VML to VL ratios in Figure 5, are congruous with this anatomical differentiation. With respect to contributing to patellar alignment, however, the much greater obliquity of the fiber orientation of the VMO (40 [degrees] - 55 [degrees] as compared with 15 [degrees] - 18 [degrees] for the VML) indicates that the VMO plays a much greater role than the VML, which appears to be more closely aligned with the vastus intermedius in terms of both innervation and fiber orientation.[19] As such, alterations in the VMO to VL ratio are much more likely to affect patellar tracking than comparable changes in the VML to VL ratio.

Comparison with Previous

Findings

Our findings with respect to the maximum relative activity of the QF components across the exercises tested are summarized in Figure 6, in which data from the initial and final QS sets have been combined and the relative activity of each of the muscle groups is compared for each of the four exercises. Presenting the data in this manner emphasizes the marked difference in activity patterns between the single-joint knee extensors (VMO, VML, and VL) and the RF, which acts both to extend the knee and to flex the hip. All of the single-joint knee extensors tested demonstrated significantly greater mean activity during the QS exercise than during any of the SLR variations. The RF showed a nearly opposite pattern, demonstrating significantly greater activity during both the SLR and SLR/LR exercises than during the QS exercise.

These findings are in agreement with those of Soderberg and Cook[33] who compared vastus medialis and RF activity during QS and SLR exercises. These data support a distinct functional division between the one- and two-joint QF components. When the RF is strongly activated in order to exert a flexor torque at the hip, the single-joint knee extensors demonstrate submaximal activity (despite instructions to maintain a maximal QS throughout the SLR exercise), whereas the opposite is true during the performance of the isolated QS exercise. Because the single-joint QF components account for approximately 84% of the cross-sectional area of the QF,[34] this finding appears to have important clinical implications. Although the SLR is frequently considered to be a progression from the QS for QF strengthening, that assumption appears questionable for the single-joint QF components, at least for the magnitude of resistance used in this study.

Comparison of normalized medial to lateral EMG ratios for the four exercises revealed no significant differences among the exercises in either the VMO to VL or VML to VL ratio. Though there are no previously published data with which these results may be directly compared, this study's findings are in agreement with anecdotal reports by Basmajian[25] and Wild and associates,[26] who stated that they had observed no change in the relative activity levels of the QF group when comparing SLR and SLR/LR exercises. Two other interesting trends may be noted in Figure 6. In contrast to the relatively low variability in the medial to lateral activity ratios associated with the QS exercise, all three variations of the SLR showed a relatively greater degree of variability across subjects. Although neither ratio showed statistically significant exercise effects, Figure 6 illustrates that the VMO and VML showed opposite trends, with the mean VMO to VL ratio actually being the lowest for the SLR/LR exercise, which has been advocated as a specific means of increasing VMO activity.

Our results do not appear to support the suggestion made by Hanten and Schulthies[23] that the VMO can be strengthened selectively by performing hip adduction exercises. This apparent discrepancy might be attributed to several differences in experimental protocol. Hanten and Schulthies' subjects were positioned with the knee in 60 degrees of flexion, and their subjects performed isometric hip adduction, which was resisted by a pad placed proximally to the knee. Their subjects apparently were not instructed to actively extend the knee, though the relatively high percentages of maximum QF EMG activity reported during the isolated hip adduction exercises (mean values of 61.75% for the VMO and 45.63% for the VL) would seem to indicate that subjects must have either generated significant knee extensor torques or co-activated the hamstring muscles along with the QF.

Advocates of combining hip adduction with knee extension exercises have typically focused on the anatomical link between the VMO and the adductor magnus as the rationale for such exercises.[14,22-24] An alternative rationale for hypothesizing an increase in the medial to lateral QF activity ratio during the performance of the SLR/LR and SLR/ADD exercises would be that muscle groups crossing the medial aspect of the knee joint (eg, the VMO and medial hamstrings) might be activated in response to stress of medial joint structures, such as the joint capsule and medial collateral ligament, in order to augment and protect those passive structures by providing dynamic support. The presence of sensory receptors in ligaments and the identification of ligamentous-muscular reflexes support this hypothesis.[35,36] Although this possibility is intriguing, experimental results obtained while applying abduction torques to the human knee have provided little support for this hypothesis.

The most direct examination of this hypothesis in humans was performed by Andriacchi and colleagues,[37] who tested the effect of an external abduction torque applied at the knee during isometric knee extension at angles of 10, 20, and 40 degrees of flexion; no hip adduction torque was required of the subjects because the thigh was stabilized in their study. They reported that vastus medialis activity actually decreased when abduction torques were applied with the knee in 10 or 20 degrees of flexion, but increased slightly when the knee was flexed to 40 degrees. Their findings are further supported by the results obtained in our study, during which no increase in activation of the VMO was observed with the knee held in hill extension, even though the lateral force applied at the ankle resulted in abduction torques at both the hip and knee.

Clinical Relevance of Findings

There is widespread agreement that conservative treatment of patients with patellofemoral joint dysfunction should include active exercise aimed at strengthening the QF in general and the VMO in particular. In order to reduce the risk of exacerbating patellofemoral joint irritation, isometric QF exercises performed with the knee in full extension, such as the QS and SLR exercises used in this study, have been widely advocated. One of the questions addressed by this study is whether the likelihood of selectively strengthening the VMO is enhanced by modifying such exercises to elicit concurrent hip adductor muscle activity.

The results of our study do not support the suggestion that isometric QF strengthening exercises are more likely to preferentially activate the VMO when performed in conjunction with hip adductor activity. We found that one such exercise (ie, SLR/LR), which has been advocated for preferential strengthening of the VMO, actually elicited the lowest mean VMO to VL ratio of the four exercises studied. Moreover, performing the SLR/LR exercise with an ankle weight equal to 5% of body weight resulted in a mean VMO activity level of only 50.5% of the maximal activity that was typically attained during performance of the QS exercise. These findings suggest that, in cases requiring QF strengthening in the fully extended knee position, isometric QS may be the treatment of choice.

Limitations and Suggestions

for Further Research

Because these results were obtained by studying subjects with asymptomatic knees, rather than patients with patellofemoral dysfunction, caution is warranted in extrapolating these findings to patient populations. It should be noted, however, that a recent study by Soderberg and Cook[33] of QF EMG activity during QS and SLR exercises demonstrated no significant differences between healthy subjects and patients with knee pathologies. Moreover, we would point out that the anatomical link between the VMO and the adductor magnus, which is the theoretical rationale presented by advocates of exercises such as the SLR/LR, should be equally present in both healthy subjects and patients with patellofemoral dysfunction.

Other limitations of this study include the size of the sample and the magnitude of the external loads used. The results presented are based on a total of only 12 subjects (6 male, 6 female). Given the results of our pilot studies, the consistency of the findings across subjects, the identical findings for the male and female subgroups of the sample, and the general agreement with previous studies, we doubt that increasing the sample size would have altered the results. Future studies addressing this proposed functional link between the VMO and adductors could further address both the clinical relevance and reliability of these results by examining a larger sample that includes both healthy subjects and patients with knee pathology.

With respect to the loads studied, the decisions to use only a single level of external loading and to limit that level to 5% of body weight (for both the ankle weight and the laterally directed force) were based on the necessity of having all subjects complete the entire protocol of 25 QF exercises without fatiguing the QF group to a degree that would significantly alter the force/EMG relationship, the stability of which is critical to the validity of our conclusions. The 5% of body weight level was chosen on the basis of subjects' perception of the degree of effort required to complete the required tasks during pilot studies. Although the loads used for these healthy subjects were submaximal in all cases, subjects reported that the task of maintaining a strong QS set and consistent foot position throughout each of the exercises was not an easy task, and several subjects reported that the SLR exercise combined with adduction was quite difficult to perform at the load level used in this study. Furthermore, it is clear that the loads used in this study were sufficient to produce significant changes in the relative activity of the QF components, as evidenced by significant differences in activity of all QF components when comparing QS and SLR variations and by a significant decrease in RF activity during the SLR/ADD exercise as compared with the other two variations of SLR. Whether greater loads would alter the medial to lateral QF activity ratio remains to be determined.

Although none of the exercises studied significantly increased the relative activity of the medial QF components to a greater degree than the lateral components, the rationale for preferential strengthening of the VMO in patients with patellofemoral dysfunction appears sound. The methodology used in this study could be adapted to examine patterns of QF activity in terms of both timing and magnitude of EMG activity in medial and lateral QF components during various rehabilitation protocols and in relation to functional activities. Of particular relevance to the treatment of patellofemoral dysfunction is the manner in which other currently advocated therapeutic interventions, such as patellar taping or the use of weight-bearing exercises, might affect medial to lateral QF activity ratios. It is important to note, however, that the validity of studies using EMG data to infer the relative effectiveness of various strengthening protocols depends on careful design and implementation, particularly with respect to EMG processing and normalization procedures and the limitations of the force-EMG relationship.

Conclusions

The results suggest that exercises requiring combined hip adduction and knee extension torques, such as SLR/LR or SLR/ADD exercises, are not superior to the QS or the standard SLR exercise in eliciting an increase in the relative activity of the medial components of the QF. Our results also support and expand upon previous studies indicating that QS exercises tend to elicit a greater degree of activity of the single-joint knee extensors than do SLR exercises. These results suggest that for patients needing an increased medial to lateral force ratio acting on the patella in addition to generalized QF strengthening, isometric QS exercises may actually be more beneficial than SLR/LR or SLR/ADD exercises.

(*) Coulbourn Instruments, Box 2551, Lehigh Valley, PA 18001. (dagger) WATSCOPE, Northern Digital Inc, 403 Albert St, Waterloo, Ontario, Canada N2L 3V2. (doubler dagger) SYSTAT, version 4.2, SYSTAT Inc, 1800 Sherman Ave, Evanston, IL 60201.

References

[1] Atha J. Strengthening muscle. Exerc Sport Sci Rev. 1981;9:1-73. [2] Jones DA, Rutherford OM. Human muscle strength training: the effects of three different regimes and the nature of the resultant changes. J Physiol (Lond). 1987;391:1-11. [3] Lindh M. Increase of muscle strength from isometric quadriceps exercises at different knee angles. Scand J Rebabil Med. 1979;11:33-36. [4] Parker RH. The effects of mild one-legged isometric or dynamic training. Eur J Appl Physiol. 1985;54:262-268. [5] Rutherford OM. Muscular coordination and strength training: implications for injury rehabilitation. Sports Med. 1988;5:196-202. [6] Tomberlin JP, Basford JR, Schwen EE, et al. Comparative study of isokinetic eccentric and concentric quadriceps training. Journal of Orthopaedic and Sports Physical Therapy. 1991; 14:31-36. [7] Antich TJ, Brewster CE Modification of quadriceps femoris muscle exercises during knee rehabilitation. Phys Ther. 1986;66:1246-1251. [8] Brunet ME, Stewart GW. Patellofemoral rehabilitation. Clin Sports Med. 1989;8:319-329. [9] Woodall W, Welsh J. A biomechanical basis for rehabilitation programs involving the patellofemoral joint. Journal of Orthopaedic and Sports Physical Therapv. 1990;11:535-542. [10] Mariani PP, Caruso I. An electromyographic investigation of subluxation of the patella. J Bone Joint Surg [Br]. 1979;61:169-171. [11] Shelton GL, Thigpen LK. Rehabilitation of patellofemoral dysfunction: a review of literature. Journal of Orthopaedic and Sports Physical Therapy. 1991;14:243-249. [12] Fisher RL. Conservative treatment of patellofemoral pain. Orthop Clin North Am. 1986; 17:269-272. [13] Lieb FJ, Perry J. Quadriceps function: an anatomical and mechanical study using amputated limbs. J Bone Joint Surg [Am]. 1968; 50: 1535-1548. [14] McConnell J. The management of chondromalacia patellae: a long-term solution. Australian Journal of Physiotherapy. 1986;32:215-223. [15] LeVeau BF, Rogers C. Selective training of the vastus medialis oblique muscle using EMG biofeedback, Phys Ther. 1980;60:1410-1415. [16] Wise HH, Fiebert IM, Kates JL. EMG biofeedback as treatment for patellofemoral pain syndrome. Journal of Orthopaedic and Sports Physical Therapy. 1984;6:95-103. [17] Riegler HF. Recurrent dislocations and subluxations of the patella. Clin Orthop. 1988;227: 201-209. [18] Bose K, Kanagasuntheram R, Osman MBH. Vastus medialis oblique: an anatomic and physiologic study. Orthopedics. 1980;3:880-883. [19] Thiranagama R. Nerve supply of the human vastus medialis muscle. J Anat. 1990; 170: 193-198. [20] Wheatley MD, Jahnke WD. Electromyographic study of the superficial thigh and hip muscles in normal individuals. Arch Physifed Rehabil. 1951;32:508-515. [21] Speakman HGB, Weisberg MA. The vastus medialis controversy. Physiotherapy. 1977:63: 249-254. [22] Brownstein BA, Lamb RL, Mangine RE. Quadriceps torque and integrated electromyography. Journal of Orthopaedic and Sports Physical Therapy. 1985;6:309-314. [23] Hanten WP, Schulthies SS. Exercise effect on electromyographic activity of the vastus medialis oblique and vastus lateralis muscles. Phys Therapy 1990;70:561-565. [24] Reynolds L, Levin TA, Medeiros JM, et al. EMG activity of the vastus medialis oblique and the vastus lateralis in their role in patellar alignment. Am J Phys Med Rehabil. 1983;62:61-70. [25] Basmajian JV. Muscles Alive. 4th ed. Baltimore, Md: Williams & Wilkins; 1978:265-266. [26] Wild JJ, Franklin TD, Woods GW. Patellar pain and quadriceps rehabilitation: an EMG study. Am J Sports Med. 1982; 10: 12-15 [27] Soderberg GL, Minor SD, Arnold K, et al. Electromyographic analysis of knee exercises in healthy subjects and in patients with knee pathologies. Phys Ther. 1987;67:1691-1696, [28] Zipp P. Recommendations for the standardization of lead positions in surface electromyography Eur J Appl Physiol. 1982;50:41-54. [29] Baumgartner TA. Norm-referenced measurement: reliability. In: Safrit MJ, Wood TM, eds. Measurement Concepts in Physical Education and Exercise Science, Champaign, Ill: Human Kinetics Publishers Inc; 1989:45-72. [30] Shott S. Statistics for Health Professionals. Phfladelphia, Pa: WB Saunders Co; 1990:167-180. [31] Lippold OCJ. The relationship between integrated muscle potentials in a human muscle and its isometric tension. J Physiol (Lond). 1952;117:492-499. [32]Bigland B, Lippold OCJ. The relation between force, velocity and integrated electrical activity in human muscles. J Physiol ( Lond). 1954;123:214-224. [33] Soderberg GL, Cook TM. An electromyographic analysis of quadriceps femoris muscle setting and straight leg raising. Phys Ther. 1983;63:1434-1438. [34] Lehmkuhl LD, Smith LK. Brunnstrom's Clinical Kinesiology. 4th ed. Philadelphia, Pa: FA Davis Co; 1986:398. [35] Brand RA. Knee ligaments: a new view. J Biomech Eng, 1986;108:106-110. [36] Brand RA. A neurosensory hypothesis of ligament function. Used Hypotheses. 1989;29: 245-250. [37] Andriacchi TP, Andersson GB, Ortengren R, Mikosz RP. A study of factors influencing muscle activity about the knee joint.j Orthop Res. 1984;1:266-275.

GM Karst, PhD, PT, is Assistant Professor, Division of Physical Therapy Education and Department of Physiology and Biophysics, University of Nebraska Medical Center, 600 S 42nd St, Omaha, NE 68198-4420 (USA), He was Assistant Professor, Department of Therapeutic Science, University of Wisconsin-Madison, 1300 University Ave, Madison, WI 53706, at the time of this study. Address all correspondence to Dr Karst.

PD Jewett, PT, is Staff Physical Therapist, Team Rehab, 141 N Meramec, Ste 103, Clayton, MO 63105. He was a senior student in physical therapy, Department of Therapeutic Science, University of Wisconsin-Madison, at the time of this study.

Commentary

There is a need in physical therapy to evaluate common rehabilitation practices, so it can be determined whether the desired outcome of a particular therapeutic procedure has been achieved. It is only after this sort of scrutiny that we will grow as a profession. The authors are to be commended for pursuing this aim. They have investigated the common clinical claim that particular modification of the straight-leg-raising (SLR) exercise causes preferential activation of the vastus medialis obliquus muscle (VMO). Specific activation of the VMO has been advocated for patients with patellofemoral pain to improve patellofemoral tracking. It must be questioned, however, whether this report adds much to our body of knowledge, as similar research was performed by Soderberg and colleagues. The difference with this study is that the authors have examined the effect of a lateral rotation component, as well as an adduction component, to the SLR maneuver. The rationale for this addition was that exercises combining activity of the hip adductors in conjunction with conventional knee extensor strengthening activities such as the SLR have been advocated as a means of preferentially activating and exercising the VMO.

The above statement contains some unchallenged assumptions that have formed the basis of quadriceps femoris muscle (QF) rehabilitation for many years. One such assumption is that activation of the VMO requires knee extension. The work of Lieb and Perry,[3] which was referenced by the authors, established that "no extension of the knee could be accomplished by applying weight to the VMO tendon. In fact, the femur was fractured in each case before any extension was accomplished." (Cadaver specimens were used in the study). Lieb and Perry concluded that the function of the VMO was to align the patella medially, thus overcoming "the malaligning effects of the vastus lateralis."[3] From Lieb and Perrys meticulous work on QF function, we should have learned at least two things to help us implement an appropriate VMO strengthening program. First, the VMO is not an extensor of the knee, but a medial stabilizer of the patella, and second, the VMO is active throughout the entire extensor range, not just in the last 10 to 15 degrees.

Recently, however, Petschnig and colleagues[4] demonstrated in asymptomatic individuals that the VMO exhibited more activity than the vastus lateralis muscle (VL) at 20 degrees of knee flexion than at 90 degrees. Subjects with patellofemoral pain showed a reversal of electromyographic (EMG) activity, with more VMO than VL activity being present at 90 degrees and less at 20 degrees. This preliminary study gives some justification for performing inner-range extension exercises, because it appears that there is a deficit in VMO activity in this range in individuals with patellofemoral pain.

To preferentially recruit the VMO, however, isometric hip adduction, rather than knee extension, should be emphasized in treatment, because the VMO is a medial patellar stabilizer that arises from the adductor magnus tendon.[5] This supports Hanten and Schulthies' rationale of selectively strengthening the VMO by isometric adduction at 60 degrees of flexion without active extension of the knee.[6] interestingly, according to the 36th edition of Gray's Anatomy, "the adductors are essentially synergists in the complex patterns of gait activity" and are rarely required to perform forcible adduction.[7] During a VMO exercise regimen, in weight bearing, where adduction is emphasized, perhaps it is this synergistic action of the adductor magnus muscle that is being retrained. Additionally, it must be remembered that the adductor magnus and adductor longus muscles assist in lateral rotation of the thigh, so basing the theoretical rationale for SLR exercises performed with the hip in lateral rotation on an anatomical link between the VMO and the adductor magnus muscle, as stated by the authors, seems to be unfounded.[7]

When utilizing isometric adduction to enhance a VMO contraction, however, it seems that the position of the limb is a critical factor. A recently completed study by Hodges and Richardson[8] supports the finding of the authors that the addition of isometric adduction of 5% of body weight (about 30% of a maximal contraction) has no differential effect on the activation of the VMO in the non-weightbearing situation. Hodges and Richardson found that a maximal contraction was required before any increase in activity in the VMO relative to the VL could be demonstrated. Yet, in weight bearing, only 20% of a maximal contraction of the adductors was required to differentially increase VMO activity relative to the VL." This finding suggests that we should consider the effect of limb position and the relationship of the synergists if we desire preferential activation of the VMO.

Unfortunately, specific timing in a weight-bearing position for the treatment of the patient with patellofemoral pain has been actively discouraged, because the flexed weightbearing position increases the patellofemoral joint reaction force (PFJRF) - the greater the flexion, the greater the compressive force.9,10 The PFJRF has been calculated to be one half body weight during level walking, three to four times body weight during stair climbing, and seven to eight times body weight in a squat.9,10 Rather than avoiding knee flexion activities, which are an integral part of a patient's daily living, the aim of physical therapy intervention should be to increase the surface area of contact of the patella on the femur. This distributes the load over a wider area and decreases any localized concentration of pressure. Abnormal concentration of stress, if prolonged, may result in chondromalacia patellae on the posterior surface of the patella, not, as the authors suggest, on the inferior surface.[9,11] In fact, patients with problems at the inferior surface of the patella have their symptoms exacerbated by the very exercises that are supposed to be beneficial, that is, the QF setting (QS) and SLR exercises. These maneuvers displace the inferior pole of the patella posteriorly, causing further irritation of an already inflamed fat pad.12

Correcting the patella position, by taping the patella into an appropriate position, for example, should increase the surface area of the patella in contact with the femur and result in an immediate decrease in the patient's symptoms.[13,14] This then allows the therapist to be specific with muscle training. Exercises can, therefore, be commenced in pain-free weightbearing positions, particularly the first 20 to 30 degrees of knee flexion, as this is the range at which the amount and timing of the VMO activity are critical in controlling the seating of the patella in the trochlea. As strength gains are considered to be largely an acquisition of skill, strength training must be specific to the movement pattern, limb position, velocity, and contraction type.[15,16] For many years, coaches and athletes have been aware of the benefits of specificity of training, yet in physical therapy we have been more reluctant to adopt this concept of training. Imagine a tennis coach asking his or her players to strengthen their playing arms by one-armed swimming. Sounds absurd, doesn't it, but, why is lying supine, bracing the knee, and lifting the leg in the air to strengthen the knee for stair climbing any less absurd? Why, too, do physical therapists regard SLR as a progression of QS for VMO strengthening, despite mounting evidence to the contrary?

It has been demonstrated in this and other studies that activity in the vastus muscles is less in SLR than in QS and the muscle that is preferentially activated by SLR is the rectus femoris muscle, which is not usually the aim of physical therapy intervention for patients with patellofemoral pain. How many more studies do we need before we discard the SLR as being ineffective for these patients? Perhaps a more appropriate study in the future (as there will be more) would be to measure, in healthy individuals and in individuals with patellofemoral pain, the EMG activity of the back extensors, the obliques, and all the hip flexors during an SLR with and without a weighted boot. This information could be used to determine which patients are at risk of developing back pain during their rehabilitation progmm. Clinicians need to be proactive and examine better methods of preferentially activating the VMO to hasten the rehabilitation of patients with patellofemoral pain.

A recent study by Ingersoll and Knight,18 which could be adapted further by physical therapists, demonstrated the effectiveness of EMG biofeedback in specifically recruiting VMO. The study involved three groups of healthy female college students - a control group, a group strengthening the QF with biofeedback, and a group performing progressive resistive exercises with no biofeedback. The group that received biofeedback training to increase VMO activity and to decrease VL activity demonstrated an improvement in congruence angle (ie, a reestablishment of fit of the patella in the trochlea) after 3 weeks of training. The training included QS and SLR exercises with biofeedback, as well as integrated functional activities of deep knee bends, step-ups, and bicycle riding with biofeedback. The control group, which received no training, showed no significant change in patellar position, whereas the group performing short-arc QF exercises with progressive resistance demonstrated a deterioration in the congruence angle, even though their QF had increased in strength by 170%. The authors concluded that "terminal extension progressive resistive exercises (in non-weight bearing) do not produce medial relocation of the patella and may actually predispose individuals to the likelihood of lateral subluxation of the patella."[18] Although this study has significant implications for physical therapists in the rehabilitation of the QF, it cannot be determined whether the combined effect of all the training or a particular part of the training was more effective in altering the congruence angle.

Researchers in physical therapy examine current clinical practice. It is our (the clinicians') responsibility to direct the research focus. To do this, we must critically evaluate treatment outcome to determine whether our intervention has substantially effected change in the patient's condition. The findings must be published to disseminate the information. This allows the researchers to test more clinical hypotheses. However, we must act on research findings, discarding, if necessary, ineffective techniques, modifying existing techniques, and developing further treatment procedures to improve patient management. With the rising cost of health care and the shrinking of the health care dollar, this is essential for our survival as a profession. My final plea is let's relegate to the annals the SLR (and all variations on the theme) and rejoice in its departure from patellofemoral pain management.

References

[1] Soderberg GL, Cook TM. An electromyographic analysis of quadriceps femoris muscle setting and straight leg raising. Phys Ther. 1983;63:1434-1438. [2] Soderberg GL, Minor SD, Arnold K, et al. Electromyographic analysis of knee exercises in healthy subjects and in patients with knee pathologies. Phys Ther. 1987;67:1691-1696. [3] Lieb FJ, Perry J. Quadriceps function: an anatomical and mechanical study using amputated limbs. J Bone Joint Surg [Am]. 1968;50: 1535-1548. [4] Petschnig R, Baron R, Engel A, et al. Objectivation of the effects of knee problems on vastus medialis and vastus lateralis with EMG and dynamometry. PMR. 1991;2:50-54. [5] Bose K, Kanagasuntheram R, Osman MBH. Vastus medialis oblique: an anatomic and physiologic study. Ortbopedics. 1980;3:880-883. [6] Hanten WP, Schulthies SS. Exercise effect on electromyographic activity of the vastus medialis oblique and vastus lateralis muscles. Phys Ther 1990;70:561-565. [7] Williams P, Warwick R. Gray's Anatomy. 36th ed. London, England: Churchill Livingstone; 1980. [8] Hodges P, Richardson C. An investigation into the effectiveness of hip adduction in the optimisation of the vastus medialis oblique contraction. Scandj Rebabil Med. In press. [9] Fulkerson J, Hungerford D. Disorders of the Patellofemoraljoint, 2nd ed. Baltimore, Md: Williams & Wilkins; 1990. [10] Reilly D, Martens M. Experimental analyses of the quadriceps muscle force and patellofemoral joint reaction force for various activities. Acta Orthop Scand. 1972;43:126-137. [11] Goodfellow J, Hungerford D, Zindel M. Patellofemoral joint mechanics and pathology, I and 2. J Bone Joint Surg [Br]. 1976;58:287-299. [12] McConnell J, Fat pad irritation: a mistaken patellar tendonitis. Sport Health. 1991;9:7-9. [13] McConnell J. The management of chondromalacia patellae: a long-term solution. Australian Journal of Physiotherapy. 1986;32:215-223. [14] Gerrard B. The patellofemoral pain syndrome: a clinical trial of the McConnell programme. Australian Journal of Physiotherapy. 1989;35:70-80. [15] Sale D, MacDougall D. Specificity of strength training: a review for coach and athlete. Canadian Journal of Applied Sports Sciences. 1981;6:87-92. [16] Rasch PJ, Morehouse CE. Effect of static and dynamic exercises on muscular strength and hypertrophy, j Appl Pbysiol 1957; 1 1:29-34. [17] Wild JJ, Franklin TD, Woods GW, Patellar pain and quadriceps rehabilitation: an EMG study. Am J Sports Med. 1982; 10; 12-15 [18] Ingersoll C, Knight K. Patellar location changes following EMG biofeedback or progressive resistive exercises. Med Sci Sports Exerc. 1991;23:1122-1127.

Author Response

We wish to thank Ms McConnell for taking the time to write her commentary. The main criticism in Ms McConnell's commentary appears to be that she believes our paper deals primarily with a comparison of the straight-leg-raising SLR) and quadriceps femoris muscle setting (QS) exercises, and thus adds little to the existing literature. Specifically, she cites two studies by Soderberg and associates[1,2] as being similar to the research we have presented.

The primary aim of our research was, as the title denotes, to test specific exercises that have been proposed for preferentially strengthening the vastus medialis obliquus component (VMO) to a greater degree than the vastus lateralis component (VL). Neither of those exercises, the SLR in the laterally rotated hip position (SLR/LR) or the SLR combined with isometric hip adduction (SLR/ADD), were mentioned in either of the studies by Soderberg and associates.[1,2] Furthermore, neither of those studies attempted to draw any conclusions about the relative activity of the VMO and VL during any exercise, because no electromyographic data were even reported for the VL. We did report that our results concerning the relative activity of the rectus femoris (RF) and vastus medialis muscles during QS and SLR were in agreement with previous findings by Soderberg and associates,[1,2] but we did not imply that such a finding was new, and we certainly did not intend for that to be the principal message of the manuscript. That comparison was included to enhance the credibility of our findings by demonstrating that, insofar as our data could be compared with data from other laboratories, our findings were compatible with previously published results based on larger populations of healthy subjects[1,2] as well as with results from subjects with knee pathologies.[2]

Ms McConnell takes issue with "unchallenged assumptions" in our contention that exercises combining activation of the hip adductors and knee extensor strengthening activities have been advocated as a means of preferentially activating the VMO. We wish to point out that we are simply stating a fact: The use of such exercises bas been advocated in a number of published works,[3 13] including one authored by Ms McConnell, in which she states that "the addition of adduction while performing knee extension might facilitate VMO activity during the early stages of rehabilitation."[6] The heretofore "unchallenged assumption" is whether those exercises actually result in preferential activation of the VMO to a greater degree than the VL. Our data clearly do not support that claim. In this context, we would also note that of the 11 references we found advocating hip adduction as a means of facilitating VMO activity,[3-13] none actually involved study of the SLR/LR or SLR/ADD, and only one[8] provided any experimental data addressing a possible interaction between hip adduction and VMO activity. Thus, it seems that the notion of a functional link between hip adduction and selective VMO activation is becoming "conventional wisdom" in physical therapy practice merely on the basis of repetitive hearsay, rather than scientific evidence.

On a similar note, we are puzzled by Ms McConnell's statement that "the theoretical rationale for SLR exercises performed with the hip in lateral rotation [that are based] on an anatomical link between the VMO and the adductor magnus muscle, as stated by the authors, seems to be unfounded," when earlier in the same paragraph, as well as in previous writings,[6] she argues for the use of isometric hip adduction as a means of preferentially activating the VMO by citing an anatomic study by Bose et al.[14] Again, we wish to note that this anatomically based rationale for a supposed functional link between the adductor magnus and the VMO has been cited repeatedly in the literature,[5,6,8-10,12] but is neither advocated by us nor supported by the results of this study.

Ms McConnell does raise a pertinent point in her citation of a recent study by Hanten and Schulthies,[9] which suggests that the VMO may be activated to a relatively greater degree than the VL when subjects perform isolated hip adduction with the knee in 50 degrees of flexion. The possible conflicts between their findings and those of our study have been dealt with in the "Discussion" section of our manuscript, and will not be repeated here.

We do wish to address, however, Ms McConnell's contention that "the VMO is not an extensor of the knee" and her endorsement of isolated hip adduction based on the results of the study by Hanten and Schulthies.9 The fact is that all available data, including the findings of Hanten and Schulthies, clearly indicate that maximal recruitment of the VMO occurs when the subject is instructed to extend the knee. Hanten and Schulthies' data show that asking subjects to maximally adduct the hip resulted in a mean VMO activity of less than 62% of that obtained when they asked the subjects to perform isometric knee extension. Moreover, the large standard deviation of that mean (45.69%) indicates that a number of subjects must have achieved substantially less than 62% of maximal VMO activation when instructed only to adduct the hip. Finally, Ms McConnell's endorsement of the exercise studied by Hanten and Schulthies (subject seated, with knee flexed 50', and instructed to adduct the thigh) seems at odds with her pleas for task-specific functional strength training. In light of these considerations and the results of our study, hip adduction exercises, with or without concurrent knee extension, do not appear to deserve a blanket endorsement as a means of selectively strengthening the VMO.

Though the remainder of Ms McConnell's commentary is more an overview of her philosophy on treatment of patellofemoral dysfunction than a critical review of our report, she does reiterate the point made in our discussion that other interesting possibilities for preferentially activating the VMO remain to be tested. The use of electromyography, either for biofeedback or as a means for the therapist to tailor exercise programs to individual patients, certainly seems to hold promise. Likewise, careful evaluation of various weight-bearing exercises in terms of medial-to-lateral activation ratios is essential.

Finally, we found Ms McConnell's closing statements regarding research and physical therapy to be both thought provoking and disturbing. Her contention that "researchers in physical therapy examine current clinical practice .... it is our (the clinicians') responsibility to direct the research focus" is disturbing for several reasons. First, it fosters the notion that one is eithber a researcher or a clinician, and never the twain shall meet, a notion that, in our opinion, is antiquated and detrimental to the profession. Furthermore, the idea that untested "clinical hypotheses" are a sufficient basis for implementing (and advocating) an unproven form of treatment is the complete antithesis of modern medical research, in which new treatment principles derive from the application of sound anatomical and physiological principles and are deemed worthy of acceptance into current clinical practice only after having undergone critical evaluation (ie, studies published in appropriate peer-reviewed journals). If physical therapy is to advance as a profession, we need to break down the imagined wall between "clinician" and "researcher" and develop a great many more clinician-researchers be they individuals or teams) who not only generate scientifically based hypotheses but, more importantly, take the responsibility for critically evaluating those hypotheses before advocating their incorporation into clinical practice.

Again, we would like to thank Ms McConnell for her commentary, and in particular for her comments regarding research in physical therapy, the rebuttal of which provided us with a soapbox from which to express our own opinions. We hope that both the results of our study and our reply to her commentary will stimulate research and discussion that will ultimately promote the scientific basis of our profession.

References

[1] Soderberg GL, Cook TM. An electromyographic analysis of quadriceps femoris muscle setting and straight leg raising. Phys Ther. 1983;63:1434-1438. [2] Soderberg GL, Minor SD, Arnold K, et al. Electromyographic analysis of knee exercises in healthy subjects and in patients with knee pathologies. Phys Ther. 1987;67:1691-1696. [3] Reynolds L, Levin TA, Medeiros JM, et al. EMG activity of the vastus medialis oblique and the vastus lateralis in their role in patellar alignment. Am J Phys Med Rehabil. 1983;62: 61-70. [4] Walsh WM, Huurman WW, Sheiton GL. Overuse injuries of the knee and spine in girls' gymnastics. Clin Sports Med. 1984;3: 829-850. [5] Brownstein BA, Lamb RL, Mangine RE. Quadriceps torque and integrated electromyography. Journal of Orthopaedic and Sports Physical Therapy. 1985;6:309-3i4. 6 McConnell J. The management of chondromalacia patellae: a long-term solution. Australian Journal of Physiotherapy. 1986;32:215-223. [7] Antich TJ, Brewster CE. Modification of quadriceps femoris muscle exercises during knee rehabilitation. Phys Ther. 1986;66:1246-1251 [8] Shelton GL. Principles of musculoskeletal rehabilitation. In; Mellion MB, ed. Office Management of Sports Injuries and Athletic Problems. Philadelphia, Pa: Hanley & Belfus Inc; 1988:160-185. [9] Hanten WP, Schulthies SS. Exercise effect on electromyographic activity of the vastus medialis oblique and vastus lateralis muscles. Phys Ther 1990;70:561-565. [10] Shelton GL, Thigpen LK. Rehabilitation of patellofemoral dysfunction: a review of literature. Journal of Orthopaedic and Sports Physical Therapy. 1991;14:243-249. [11] Doucette SA, Goble EM. The effect of exercise on patellar tracking in lateral patellar compression syndrome. Am J Sports Med. 1992;20:434 40. [12] Shelton GL. Conservative management of patellofemoral dysfunction. Printart Care. 1992;19:331-350 [13] Wieder DL. Patellofemoral tracking syndrome. Rebab Management. 1993;6:99-102. [14] Bose K, Kanagasuntheram R, Osman MBH. Vastus medialis oblique: an anatomic and physiologic study. Orthopedics. 1980;3:880-883.