ClinicalTrials.gov identifier: NCT01346995.
Experimental knee joint pain during strength training and muscle strength gain in healthy subjects: A randomized controlled trial†
Article first published online: 28 DEC 2011
Copyright © 2012 by the American College of Rheumatology
Arthritis Care & Research
Volume 64, Issue 1, pages 108–116, January 2012
How to Cite
Sørensen, T. J., Langberg, H., Hodges, P. W., Bliddal, H. and Henriksen, M. (2012), Experimental knee joint pain during strength training and muscle strength gain in healthy subjects: A randomized controlled trial. Arthritis Care Res, 64: 108–116. doi: 10.1002/acr.20618
- Issue published online: 28 DEC 2011
- Article first published online: 28 DEC 2011
- Accepted manuscript online: 8 SEP 2011 03:39PM EST
- Manuscript Accepted: 26 AUG 2011
- Manuscript Received: 1 APR 2011
- The Danish Physiotherapists Research and Practice Foundations
- The Oak Foundation
- The Danish Council for Independent Research, Medical Sciences. Grant Number: 10-093704
Knee joint pain and reduced quadriceps strength are cardinal symptoms in many knee pathologies. In people with painful knee pathologies, quadriceps exercise reduces pain, improves physical function, and increases muscle strength. A general assumption is that pain compromises muscle function and thus may prevent effective rehabilitation. This study evaluated the effects of experimental knee joint pain during quadriceps strength training on muscle strength gain in healthy individuals.
Twenty-seven healthy untrained volunteers participated in a randomized controlled trial of quadriceps strengthening (3 times per week for 8 weeks). Participants were randomized to perform resistance training either during pain induced by injections of painful hypertonic saline (pain group, n = 13) or during a nonpainful control condition with injection of isotonic saline (control group, n = 14) into the infrapatellar fat pad. The primary outcome measure was change in maximal isokinetic muscle strength in knee extension/flexion (60, 120, and 180 degrees/second).
The group who exercised with pain had a significantly larger improvement in isokinetic muscle strength at all angular velocities of knee extension compared to the control group. In knee flexion there were improvements in isokinetic muscle strength in both groups with no between-group differences.
Experimental knee joint pain improved the training-induced gain in muscle strength following 8 weeks of quadriceps training. It remains to be studied whether knee joint pain has a positive effect on strength gain in patients with knee pathology.
Pain is the cardinal symptom in knee joint diseases such as knee osteoarthritis (OA) (1), and is a strong predictor of functional impairment in patients with rheumatic disease (2, 3). Lower extremity muscle weakness, particularly in the quadriceps (4, 5), is a common feature of knee joint diseases (6, 7), and contributes to functional impairment (4, 8).
Treatment of patients with knee joint diseases includes modalities that aim to relieve pain and restore physical function. Treatment efficacy is often evaluated based on the ability to reduce patients' pain (9). Exercise therapy has been shown to be as effective as pharmacologic treatment for pain reduction in patients with knee OA (9, 10), rheumatoid arthritis (11), and anterior cruciate ligament rupture (12), and is recommended as a frontline treatment approach. Exercise, including muscle strength training, may in fact be the most cost-effective treatment for knee OA (13).
It is generally accepted that the presence of pain reduces muscle function and is believed to counteract effective muscle rehabilitation (4, 14). Several neural mechanisms may underlie this compromise. Motoneurons receive input from joint nociceptors (15), and close relationships between group III and IV afferent fiber and muscle motoneurons have been described (15, 16). Nociceptive activity in these afferents may produce changes at motoneurons in the spinal cord (17) or at higher motor centers (18). Animal studies have shown that activation of group III and IV afferents facilitates the nociceptive flexor withdrawal reflex (19), increases Ib inhibition (20), leads to γ loop dysfunction (21), and provides inhibitory and excitatory postsynaptic potentials on the motoneuron. Furthermore, experimentally induced knee pain leads to changes in coordination between the quadriceps heads in humans during stepping (22). Based on these findings, it may be hypothesized that stimulation of the knee joint nociceptive afferents may reduce the capacity to activate muscles and thereby reduce the positive effect of training (16).
Injection of hypertonic saline into the infrapatellar fat pad (IPFP) replicates many aspects of clinical knee joint pain (23), and is a well-accepted, efficient, and safe method to induce knee joint pain (23). Experimental knee pain models make it possible to test the isolated effects of pain on motor functions in healthy volunteers without the influence of confounding disease-related factors (24). It has recently been demonstrated that experimental knee joint pain significantly reduces knee muscle strength (25) as well as causing gait changes comparable to changes seen in knee OA (26). Likewise, experimental knee joint pain induces changes in motor control of the quadriceps (22) and alters the motor unit discharge rate and recruitment strategies (27).
This study aimed to assess the effects of experimental knee joint pain on muscle strength gain after 8 weeks of quadriceps muscle strength training in healthy individuals. We hypothesized that the presence of knee joint pain during strength training would reduce the muscle strength gain from a muscle strength training protocol in healthy individuals.
Significance & Innovations
Knee joint pain and reduced quadriceps strength are cardinal symptoms in knee pathologies.
In people with painful knee joints, it is assumed that pain compromises muscle function and thus may prevent effective rehabilitation.
This study showed that experimental-induced knee joint pain actually enhanced the effect of quadriceps strength training in healthy subjects.
MATERIALS AND METHODS
A randomized controlled trial of an 8-week exercise intervention was conducted. The study was approved by the Scientific Ethical Committees for the Capital Region, Denmark (H-3-2009-129). Data collection took place at The Parker Institute, Copenhagen University Hospital Frederiksberg, and the exercise intervention was carried out at a physiotherapy clinic with training facilities. Participants gave written consent before entering the study and were randomized into 1 of the 2 following exercise interventions: supervised resistance training during pain induced by injection of hypertonic saline into the IPFP (pain group), or supervised resistance training with nonpainful injection of isotonic saline into the IPFP (control group) (Figure 1).
Recruitment of participants.
Participants were recruited at educational institutions by means of advertisements on notice boards and internet posts. Volunteers were included if they were ages 20–35 years, healthy, and untrained (defined as nonregular exercise participation [i.e., <1 day/week]). They were excluded if they had any symptomatic musculoskeletal diseases; history of traumatic injuries to the muscles, tendons, or joints of the lower extremity; or knee joint pain within a month prior to enrollment. Furthermore, they were not allowed to abuse alcohol, medicine, or drugs, or have been pregnant/breastfeeding in the preceding year.
After the baseline measurements, the participants were randomized to either the pain or control group using envelope-based randomization. A portfolio was generated containing 40 nontransparent envelopes with a slip inside with the number 0 (control group) or 1 (pain group) written on it. After sealing the envelopes, they were mixed and randomly placed in the portfolio. After the baseline measurements, an envelope was drawn in sequential order. The envelope was tagged with the participant's name, date of birth, participant number, and date, and handed to the treating physiotherapist for group allocation. The portfolio was kept in a sealed locker. The randomization procedure was carried out by the data-collecting physiotherapist, who was blinded to the exercise intervention assigned to the participants at the baseline measurements.
Exercise intervention and experimental pain protocol.
Two physiotherapists supervised the exercise intervention in both groups (no group assignment). Resistance training was performed with either experimental knee joint pain or the control condition without pain. The supervised training program was performed 3 times per week for 8 weeks (∼30 minutes per training session). Each training session began with a warm-up (10 minutes of submaximal cycling on a bicycle ergometer). Subsequently, ultrasound-guided bolus injections of 1 ml sterile pain-inducing hypertonic saline (5.8%) or nonpainful isotonic saline (0.9%) into the IPFP of the subject's right knees were performed using a 0.4 mm needle mounted on a 1 ml syringe. Both groups performed 2 quadriceps strengthening exercises (leg press and knee extension machine exercises) with their right leg for 3 sets of each exercise with loads corresponding to 80% of 1 repetition maximum (RM). Each set was performed to the point of muscular fatigue (inability to maintain the target load, ∼8–12 repetitions). The rest periods between the sets were 45–60 seconds. The training volume performed during a single training session was recorded with registration of the total number of repetitions and loads.
Weekly 1 RM estimation tests were performed (28), from which the training level for the upcoming week was determined. After the usual training session followed a 30-minute rest period before the 1 RM estimation tests were performed. The estimation tests were done without pain/control interventions. Attendance to the training sessions was registered throughout the entire intervention period.
The participants rated their pain intensities after completing each set of repetitions (6 scores per training session) using a 100-mm visual analog scale (VAS) anchored with “no pain” at 0 mm and “worst imaginable pain” at 100 mm.
Outcome measurements were taken at baseline, before randomization (during the week before initiation of the exercise intervention), and after the 8-week intervention period (during the week after completion of the program) in both groups. The data-collecting physiotherapist was blinded to the participant's exercise intervention assignment at the baseline measurement session.
Muscle strength measurements.
The primary outcome measure was change in maximal isokinetic muscle strength in knee extension and flexion. A 10-minute warm-up of submaximal cycling on a bicycle ergometer was completed before the muscle strength measurements. Maximal isokinetic torque produced in knee extension and flexion with the subject's right knee was recorded in a sitting position (Biodex System 3, pro set; Biodex Medical Systems). The method has been described elsewhere (25). Muscle strength was measured isokinetically at 60, 120, and 180 degrees/second to cover a reasonable and comfortable range of functional contraction velocities (e.g., sit-to-stand and stair ascent) in both knee extension and flexion in that order. Four maximal voluntary repetitions in each direction were performed, separated by ∼30 seconds. Four peak torque values were recorded from both knee extension and flexion at each angular velocity (60, 120, and 180 degrees/second), yielding a total of 24 peak torque values (12 in both knee extension and flexion).
One-leg rise test.
The secondary outcome measure was functional performance, which was assessed before and after the exercise intervention as the maximum number of one-leg raises that could be performed from a stool (48 cm) with the right leg (29). The one-leg rise test has been used to evaluate functional muscle strength capacity in patients with knee OA (29) and in healthy individuals (30). The one-leg rise test was done approximately 5 minutes after the isokinetic tests.
Quantification of training load.
Training volume is the total work performed within a specified time, and depends of the total number of sets, repetitions per training session, weight, and training frequency (31). The training volume was calculated as follows: for each set the load was calculated in newtons (e.g., knee extension; 1 set of 9 repetitions each with 17.5 kg: 17.5 kg × 9.8179 meters/second2 × 9 repetitions = 1,546.32 N). Total training volume for a session was calculated by summing the load from each training set (3 sets/exercise). At the end of the 8-week intervention the absolute training volume was calculated by summing the loads from all of the training sessions. A relative training volume was calculated as the absolute training volume normalized to the subject's body mass.
For the analysis of the primary outcome measure, a mixed linear model was applied using SAS software. All observations were used, and the directions of knee motion (knee flexion/extension) were analyzed separately. The analyses focused on the fixed effects of visits (baseline and posttest), group (pain/control group), and angular velocity (3 velocities: 60, 120, and 180 degrees/second), including each factor as the main effects, and their interactions. Sex and baseline values were included as covariates. Each individual contributed to the analyses as random effects, which allowed for intraindividual variation. Any statistically significant interactions were evaluated post hoc by exploring the pairwise differences comparing the groups using an unpaired t-test. The analysis was performed per protocol. Statistical significance was accepted at P values less than 0.05. The secondary outcome measure was compared between groups using an unpaired t-test at baseline and posttest.
The sample size for each group was estimated on the assumption that the 2 groups would show a 15% difference in improvement of muscle strength, and with a joint SD of 20%. Using these values, 14 subjects in each group would be sufficient to show a statistically significant difference using a 2-tailed test with an alpha level of 0.05, a beta level of 0.20, and power of 0.80. Recruitment was increased to 18 per group to allow for dropout.
A summary of age, sex, height, body mass, body mass index, and baseline measurements in the study population is shown in Table 1. A total of 36 subjects were included in the study (Figure 1). Twenty-seven subjects (pain group, n = 13 [10 men/3 women]; control group, n = 14 [6 men/8 women]) completed the exercise intervention and all of the measurements. Of the original 18 subjects randomly assigned to the pain group, 3 dropped out immediately after the baseline measurements before initiating the exercise intervention (2 due to lack of time and 1 due to poor tolerance of the injections), and 2 subjects dropped out after 1 week of exercise (due to a former knee trauma that was not disclosed during the inclusion procedure and due to lack of time). Of the 18 subjects randomly assigned to the control group, 4 subjects dropped out after the baseline measurements before initiating the exercise intervention (due to an ankle injury, knee trauma during running, fear of needles, and failure to attend training). There were no statistically significant baseline differences between the 2 groups (Table 1).
|Pain group (n = 13)||Control group (n = 14)||P for group difference†|
|Age, years||26 ± 3.4||25 ± 3.2||0.64|
|Sex, total no. of men (%)||10 (77)||6 (43)||0.12|
|Height, cm||174 ± 8.3||173 ± 9.0||0.69|
|Weight, kg||74.5 ± 20.8||73.9 ± 19.9||0.95|
|BMI, kg/m2||24.3 ± 5.1||24.7 ± 6.0||0.85|
|Functional muscle strength test, total no. of one-leg rise||8 ± 8.4||10 ± 10.8||0.54|
|Peak torque (Nm/kg), knee extension|
|60 degrees/second||1.6 ± 0.06||1.7 ± 0.05||0.30|
|120 degrees/second||1.6 ± 0.05||1.6 ± 0.05||0.10|
|180 degrees/second||1.5 ± 0.05||1.5 ± 0.05||0.10|
|Peak torque (Nm/kg), knee flexion|
|60 degrees/second||1.1 ± 0.03||1.1 ± 0.03||0.31|
|120 degrees/second||1.0 ± 0.03||1.0 ± 0.03||0.45|
|180 degrees/second||1.0 ± 0.03||1.0 ± 0.03||0.51|
The mean VAS score during exercise in the pain group was 16.2 mm (95% confidence interval [95% CI] 15.6, 17.0) on the 100-mm VAS throughout the study, but ranged from 0–80 mm. The mean VAS scores across weekly training sessions in the pain group are shown in Figure 2. Two participants reported a VAS score of 0 between the fifth and the sixth set of repetitions at a single training session. In these cases, an additional injection was provided to ensure that pain was present for the entire training session. Nine participants reported a VAS score of 0 after the last set of repetitions, and all reported pain before performing the last set of exercises. The mean pain intensity did not differ between training sessions throughout the intervention period (P = 0.1340). No participants in the control group reported pain during the exercise sessions (data not shown).
Muscle strength measurements.
Maximal isokinetic muscle strength in knee extension was increased at all angular velocities for both groups after 8 weeks of resistance training (main effect: time; P < 0.0001). The improvement in isokinetic muscle strength at all angular velocities in knee extension was larger for the group who trained with pain compared to the control group (interaction: time × group; P < 0.0001). Strength improvements were independent of angular velocities (interaction: time × group × angular; P = 0.42). In the pain group, maximal isokinetic muscle strength increased by 24.6% at 60 degrees/second, by 21.6% at 120 degrees/second, and by 19.6% at 180 degrees/second. The strength in the control group improved by 7.5% at 60 degrees/second, by 5.0% at 120 degrees/second, and by 8.2% at 180 degrees/second (Figure 3).
Isokinetic muscle strength in knee flexion improved in both groups (main effect: time; P < 0.0001) and there were no significant differences between the groups (interaction: time × group; P = 0.80) (Table 2). Strength improvements were independent of angular velocities (interaction: time × group × angular; P = 0.24). The analyses were repeated with sex as the covariate, but the outcome did not change.
|Pain group (n = 13)||Control group (n = 14)||P for group difference*|
|Peak torque, Nm/kg|
|60 degrees/second||1.1 ± 0.03||1.1 ± 0.03||0.31|
|120 degrees/second||1.2 ± 0.03||1.1 ± 0.03||0.45|
|180 degrees/second||1.1 ± 0.03||1.0 ± 0.03||0.51|
|One-leg rise, total number||23.7 ± 23.5||22.4 ± 16.4||0.86|
|Knee extension, absolute training volume, N||83,944.93 ± 34,873.87||71,708.54 ± 28,327.56||0.33|
|Leg press, absolute training volume, N||427,773.83 ± 137.174||428,145.51 ± 124.340||0.99|
|Knee extension, relative training volume, N/kg||1,124.75 ± 359.23||965.13 ± 310.69||0.23|
|Leg press, relative training volume, N/kg||5,850.82 ± 1,463.46||5,798.45 ± 727.78||0.91|
|No. of training sessions||18.9 ± 1.9||20.1 ± 1.7||0.09|
One-leg rise test.
After 8 weeks of exercise intervention, the functional test showed significant improvement in both groups (P < 0.001), but there was no difference between the pain and control groups (95% CI −17.23, 14.61; P = 0.86) (Table 2).
The mean absolute and relative training volumes over the 8-week intervention period are shown in Table 2. There was no difference in these variables between the groups despite the induced pain during exercise in the pain group (Table 2). The individual absolute training volumes are shown in Figure 4. There was no significant difference in the number of training sessions between the 2 groups (Table 2).
To our knowledge, this study is the first to demonstrate the effect of experimental knee joint pain on muscle strength gain after 8 weeks of knee muscle resistance training in healthy individuals. Both the control and pain groups showed significant increases in maximal isokinetic muscle strength of the knee extensors and flexors. However, contrary to our hypothesis, the strength gain was greater for isokinetic knee extension at all angular velocities for the group that trained during pain compared to the control group. This difference cannot be explained by differences in demographics, baseline muscle strength, training volume, or compliance, as none of these values differed between groups. Therefore, the most likely explanation for the difference in outcome between the groups is the experimental pain induced in the IPFP during each training session in the pain group.
These data do not allow a comparison of the relative contribution of central (e.g., descending neural drive) and peripheral (e.g., muscle hypertrophy) contributions to strength gain. However, regardless of the explanation for the increase in strength, the effect of pain on training is likely to exert its effect via effects on neural input to the knee extensor muscles.
Most existing theories fail to provide an explanation for our findings. First, the “vicious cycle” theory hypothesizes that pain may increase muscle activity (32) and both animal and human data suggest that stimulation of group III and IV afferents excite γ motoneurons and thereby increase the sensitivity of muscle spindles (16, 33, 34). This could theoretically increase muscle activity, leading to greater contraction force during training with pain in the intervention group. However, as there was no difference in training volume between the groups, facilitation of muscle activation by pain is unlikely to explain the higher strength gain in the pain group. Second, the “pain adaptation” theory hypothesizes that pain inhibits agonistic muscles and facilitates antagonistic muscle activity (35). Excitation of group III and IV afferents can reduce muscle force, electromyographic activity, and discharge rate of some motor units (14, 17, 25, 27, 36), and reduce quadriceps activation. Our data do not support this hypothesis, as uniform inhibition of the quadriceps in the pain group would be expected to reduce the training volume and a parallel lower strength gain over time compared to the control group, but no differences in training volume and a larger strength gain were found in the pain group.
Several possible mechanisms can be proposed to explain the greater strength gain in the group that exercised with pain. The results of the present study support the model recently proposed by Hodges and Tucker (37). Rather than uniform effects (inhibition or excitation) of pain on the motoneuron pool, it is proposed that pain causes a redistribution of activity within and between muscles (37). As force is determined by the number of motor units recruited, their discharge rate, and the contractile properties of muscle (38), the ability to maintain force despite a decreased motor unit discharge rate can only occur if there is a simultaneous change either in muscle fiber contractile properties and/or changes in motor unit recruitment strategies (36). As changes in muscle contractile properties have been largely excluded (39), changed recruitment is the most likely explanation. Consistent with this, Tucker and Hodges showed multiple changes in quadriceps motoneuron activity with experimental knee joint pain, including cessation of discharge in some units and recruitment of other previously inactive units (27). Other studies also show decreased discharge rates of low and moderate threshold motor units with pain (40–42) and a parallel firing rate increase in high threshold units (27). Based on findings of nonuniform effects of pain on motoneuron discharge, it has been argued that pain may selectively inhibit low-threshold units (type I fibers) and at the same time favor recruitment of additional high-threshold units (which may include type II fibers) (41, 43). According to the size principle, higher threshold units produce greater force than lower threshold units (44), and these units are more responsive to resistance training than lower threshold units (45). The observation of greater strength gain in the pain group might be explained by preferential activation of higher threshold muscle fibers and simultaneous inhibition of low threshold units during resistance training. However, as there was no pain during the measure of muscle strength, the differential effects on low (possible inhibition) and high (possible facilitation) threshold units would not be present during outcome measurement, thus allowing all units to contribute to the maximal force generation.
One issue that requires consideration is that studies of differential effects of pain on motoneuron discharge have all used very low muscle activations and force levels (e.g., 2–9 recognized motor units and 10 N of knee extension ). In the present study, exercises were performed at high force levels (80% of 1 RM), and it is unknown if similar pain-induced changes in motor unit recruitment occur during near-maximal muscle activation.
Inability to generate maximal force during pain (25) would lead to lower training load for the pain group, if not accounted for by the study design. The present study was designed to ensure the same training load in both groups. The RM was estimated weekly without pain and used to determine the training load. As a result, the absolute and relative training volumes were not different between the groups. Therefore, to overcome any reduction of motor drive during pain during near-maximal quadriceps activation (80% of 1 RM), participants in the pain group would need to have increased cortical activity and descending motor activation to recruit additional units to compensate for those that have reduced or no contribution to force during pain. In this case, an external load of 80% of 1 RM would translate into a higher descending drive during pain (i.e., a higher neural load). Descending motor drive is enhanced by resistance training (46) and explains some component of the strength gain. Greater improvements in muscle strength in the pain group in the present study could be mediated by a greater enhancement of descending motor activation compared to the control group. As we did not test maximal force output during pain, we cannot clarify whether this was reduced by pain, and further work is required to test this hypothesis.
Several methodologic issues require consideration. First, experimental knee pain does not replicate all aspects of “clinical” chronic pain, although injection of hypertonic saline into the IPFP produces pain that is described as similar in quality (23). Clinical knee joint pain arises from several structures, including the IPFP, and is a significant source of pain in patients with knee OA (47, 48). Furthermore, experimental pain from the IPFP has been shown to change motor functions similar to those identified in knee pathology (22, 23, 25, 26). However, experimental pain does not replicate the long-term aspects of clinical pain, including the psychosocial features. For these reasons, similar positive effects of pain during exercise may not be obtained with strength training in a clinical setting. It has recently been demonstrated that patients with knee OA who have quadriceps activation failure acquire greater volitional quadriceps activation and muscle strength gain from an intervention aimed at blocking pain sensation centrally in combination with exercise (49). Second, the average pain intensity was 16.2 mm, which is lower than that reported in previous studies (25–27, 36). The time course of experimental IPFP pain shows that maximum pain intensity is reached within 5 minutes, and then gradually declines over the following 10–12 minutes (23). In previous studies using experimental IPFP pain, measurements are typically done during the first 5 minutes, yielding higher average pain intensities. In the present study, subjects rated their pain during a session that lasted approximately 10–12 minutes. Therefore, pain intensity decreased to lower levels, leading to a decrease in the calculated mean intensity. Furthermore, it is likely that exercise-induced analgesia via different pain regulatory mechanisms would reduce pain sensations (50), thus lowering the pain intensity reported during exercise. Third, the functional muscle strength test improved in both groups, but without any differences between them. This lack of difference may be due to lack of sensitivity of the test (Type II error), issues related to lack of adjustment of the chair height, or because the functional test involves multiple muscles and the motor system has multiple pathways to produce a movement. Finally, 7 participants dropped out of the study after baseline measurements (Figure 1). However, the dropouts occurred either before exercise start (n = 5) or during the first week of exercise (n = 2). The present results are based on per-protocol analyses (i.e., only includes subjects that completed the study), and it is unlikely that the dropouts confound the results.
Short-term experimental knee joint pain improved the training-induced isokinetic muscle strength gain following 8 weeks of quadriceps muscle strength training in healthy individuals. Neuromuscular reorganization during pain may account for the observed changes and needs to be clarified in future studies. It remains to be clarified whether this effect can be replicated in clinical pain conditions.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Henriksen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Sørensen, Langberg, Bliddal, Henriksen.
Acquisition of data. Sørensen.
Analysis and interpretation of data. Sørensen, Langberg, Hodges, Henriksen.
- 10EULAR Recommendations 2003: an evidence based approach to the management of knee osteoarthritis. Report of a Task Force of the Standing Committee for International Clinical Studies Including Therapeutic Trials (ESCISIT). Ann Rheum Dis 2003; 62: 1145–55., , , , , , et al.
- 11Ottawa Panel. Ottawa Panel evidence-based clinical practice guidelines for therapeutic exercises in the management of rheumatoid arthritis in adults. Phys Ther 2004; 84: 934–72.
- 12Exercise for treating isolated anterior cruciate ligament injuries in adults. Cochrane Database Syst Rev 2005; 4: CD005316., , , .
- 28Strength testing: predicting a one-rep max from a reps-to-fatigue. J Phys Educ Recreation Dance 1993; 64: 88–90..
- 38Motorneurons, motor units, and the size principle. In: Rosenberg RN, editor. The clinical neurosciences. 6th ed. New York: Churchill Livingstone; 1983. p. 471–517., .