Development of a French Isometric Strength Normative Database for Adults Using Quantitative Muscle Testing

Article Outline

• Abstract
• Methods
  • Sites
  • Participants
  • Instrumentation
  • Experimental Procedure
  • Reliability Study of Testing Procedures
  • Data Analysis
• Results
  • Between-Center Variability
  • Strength Value in a Referent Population
  • Intra- and Interrater Reproducibility
  • Case Report
• Discussion
  • Usefulness of a Normative Database
  • Methodologic Issues
  • Predictive Strength Reliability
  • Reproducibility Issues
• Conclusions
• Acknowledgment
• References
• Copyright

Abstract 

Hogrel J-Y, Payan CA, Ollivier G, Tanant V, Attarian S, Couillandre A, Dupeyron A, Lacomblez L, Doppler V, Meininger V, Tranchant C, Pouget J, Desnuelle C. Development of a French isometric strength normative database for adults using quantitative muscle testing.

Objective

To establish a normative database for isometric strength measured by quantitative muscle testing (QMT) for a French adult population.

Design

Measurement of maximal voluntary isometric contraction.

Setting

Four clinical centers involved in neuromuscular disorders.

Participants

A total of 315 healthy adults (147 men, 168 women) ages 20 to 80 years.

Interventions

Not applicable.

Main Outcome Measure

Isometric torque values of 14 muscle functions (13 bilaterally and neck).

Results

This study led to the development of a French isometric strength normative database for adults measured by QMT. For each muscle function, predictive regression models using age, sex, and weight are proposed. Some methodologic issues concerning strength measurement are discussed.

Conclusions

This database can be used to compute relative deficits in muscle strength for 27 muscle functions and also to estimate composite scores for follow-up of patients either during the natural history of their disease or during a therapeutic trial.

Key Words: Isometric contraction, Muscles, Outcome assessment (health care), Rehabilitation

ANY PATHOLOGY INVOLVING the neuromuscular system can be longitudinally investigated with 1 or several methods to follow degenerative effects on muscle strength during the natural history of the disease or to detect small changes during a therapeutic trial. As already described,¹, ² several methods may be used to assess muscle strength. Depending on the aim of the assessment, each presents several advantages and drawbacks. Methodologic issues are fundamental because patients present with different motor capacities, changes in strength may be fairly small over the duration of the trial, and different evaluators may be involved in different clinical centers.

The term quantitative muscle testing (QMT) implies that quantification of strength is performed by a measurement device or sensor.³ It can be performed by handheld dynamometers, strain gauges with 1 extremity fixed to a wall-mounted frame, or isokinetic ergometers. Strain gauges have been frequently used in clinical trials concerning various neuromuscular diseases.⁴, ⁵, ⁶, ⁷, ⁸ Strength measurements are performed in isometric conditions to assess the maximal voluntary isometric contraction (MVIC) at a given position. Although muscles produce linear forces, motions at joints are generally rotary. Strength generated around joints should be measured as torque (in newton meters) because the degrees of freedom of joints are mainly rotational. When strength estimates are made, the moment arm must be carefully measured. Otherwise, the reproducibility of measurements cannot be assessed even when anatomic reference marks are methodically respected. According to Munsat,⁹ MVIC measurement provides a “direct, reproducible, sensitive and practical” method to assess changes in the voluntary motor system. The reliability of QMT was questioned in a report on a multicenter trial in amyotrophic lateral sclerosis (ALS).¹⁰ The authors¹⁰ argued that the development of precise procedures understood and applied by all the involved evaluators is a prerequisite to achieve consistent measurements. However, compared with manual muscle testing (MMT), which can require multiple training sessions to obtain acceptable reliability, a high level of agreement can be obtained with a single session of training in QMT.¹¹

Andres et al⁴, ¹² developed standardized quantified tests known as the Tufts Quantitative Neuromuscular Exam (TQNE) for evaluating patients suffering from ALS. The TQNE includes measurements of pulmonary and oropharyngeal functions, timed motor activities, and QMT. The different measurements can be combined by using z scores based on the population mean and standard deviation to produce megascores. The TQNE was used for instance by Munsat¹³ and Conradi¹⁴ and colleagues to assess motor disability and disease progression in ALS patients. More recently, MVIC was used in children suffering from Duchenne muscular dystrophy to assess the efficacy of creatine or glutamine supplementation by using QMT and MMT.⁸ QMT was found to be more sensitive than MMT in detecting muscle loss of strength.

Because the range of strength differs between the different muscle functions tested, the computation of composite scores requires the use of predictive models to express data in relative terms such as by z scores or as percentages of predicted values. In 1996, The National Isometric Muscle Strength Database Consortium established a normative database for the population of the United States in 493 adult healthy subjects by using 10 muscle groups.¹⁵ Separately, Tawil et al¹⁶ computed a regression model on the elbow flexion strength of 32 healthy subjects by using the variables age, sex, and height to show the possibility of computing composite scores (z scores) for the assessment of the strength of patients suffering from facioscapulohumeral dystrophy (FSHD). This first normative database was enlarged to 168 healthy subjects.¹⁷ Personius et al¹⁸ used these models for an FSHD natural history study. Recently, Meldrum et al¹⁹ published median and percentile predicted MVIC values for 9 muscle groups according to age and sex by using quantile regressions based on a sample of 494 healthy subjects in Ireland. It is, thus, possible to locate individual values for a given patient within or under the predicted range. However, the calculation of composite scores, which can be useful as outcome measures in the assessment of therapeutic interventions, is not possible.

The main objectives of this study were to establish a normative database for isometric strength measured by QMT for the French population and to propose regression models for clinical use for a larger number of muscle functions to assess patients’ weakness on follow-up. Intra- and interrater reproducibility was also questioned.

Back to Article Outline

Methods 

Sites 

Four clinical centers treating patients with neuromuscular disorders participated in the study. The centers were chosen because of their involvement in clinical trials and expertise in neuromuscular disorders.

Participants 

The study involved 315 healthy men (n=147) and women (n=168) aged 20 to 80 years who able to understand and perform the testing procedures. Subjects were uniformly distributed according to age. Exclusion criteria included muscle pathology; inflammatory disease or any disease involving joints; cardiovascular, pulmonary, or metabolic disease; use of regular medication over the past month (except oral contraceptives and hormone replacement therapy); or use of analgesic, anti-inflammatory, or sedative medication over the 2 days before testing and athletes involved in international sports competitions. Inclusion and exclusion criteria were verified through thorough interview and clinical examination by 1 of the medical investigators. Subjects were recruited from hospital personnel, relatives, patient families; advertisements placed in hospitals; or publications of patient associations. All the subjects signed an informed consent form before taking part in the study. The ethics committee of Nice, France, approved the study.

Instrumentation 

All centers used the same QMT system. Such systems are designed to measure muscle-force production during isometric contraction. The system included the following items: a wall-mounted frame,^a a load cell that used strain-gauge technology for measuring force,^b straps to attach the load cell to the frame and to the patient,^c a mobile examination table,^d a grip dynamometer,^e and a computer for feedback and recording usingQuantitative Muscle Assessment (QMA) software.^f Strength signals were sampled at 30Hz and recorded for further analysis.

Experimental Procedure 

We performed strength measurements for 14 muscle functions (13 bilaterally and neck flexion). Patients were placed on the mobile examining table. Wall-mounted traction bars were used to stabilize the pull straps attached to the strain gauge. Testing positions were standardized (table 1). The examiner provided appropriate stabilizations for maximal efforts of each function tested. QMT was performed bilaterally on the individual muscle listed groups. Each subject completed a series of 3 trials in which he/she was asked to produce a maximal voluntary contraction lasting 2 to 4 seconds. The subject was verbally encouraged. A 30-second rest period was allowed between each trial. If at least 2 trials differed by 10% or more, a further trial was performed. MVIC was taken as the maximum value of the trials. Test order was standardized as indicated in the second column of table 2. QMT was performed by recording force (in kilograms) through a direct computer interface linked to the strain gauge. The whole testing examination lasted 90 minutes on average.

Table 1. Procedures for QMT

Abbreviation: ASIS, anterior superior iliac spine.

Table 2. Torque Values

NOTE. Values are mean newton meters ± standard deviation (SD), except Neck flexion and Handgrip values, which are newtons. Raw data in kilograms are available on request.

We recorded dominant side, weight (in kilograms) and height (in centimeters) for each subject. Body mass index (BMI) was computed (in kilograms per meter squared) as the ratio between the weight and the height squared. The distance between the rotation axis of the joint and the line of application of force (taken at the center of the strap) was measured. Raters were trained for reliable level arm measurement. When procedures and anatomic references are respected, the relative measurement error was generally less than 5% in our experience, which is lower than the expected intrinsic variability of strength measurements (10%–15% for healthy subjects). The straps were positioned in accordance with anatomic references as given in table 3. The torque was then computed (in newton meters) as the product of the recorded force and this moment arm.

Table 3. Anatomic Reference Points

Reliability Study of Testing Procedures 

Before the study, all examiners attended a training session. A reliability study was performed to assess intra- and interrater reliability and validate the procedure of testing established during training. Ten subjects participated in an intrarater reliability tests involving 4 physiotherapists, and 10 subjects participated in an interrater reliability study involving 6 physiotherapists. QMT was performed twice at a maximum interval of 1 week.

Data Analysis 

Strength values are given as torque in newton meters. Analysis of variance (ANOVA) was performed to analyze differences between centers. To take into account the number of comparisons, P-level significance was adjusted according to the Bonferroni procedure. Reliability results (inter- and intrarater agreement coefficients for each muscle group) are expressed as intraclass coefficients (ICCs) computed with a random-effects ANOVA model. To study the relation of strength with the following covariates: age, sex, height, weight, and BMI, stepwise multiple linear regressions were performed to identify significant parameters. The most common significant parameters found for the 27 muscle groups were age, sex, and weight. These covariates were, therefore, retained to calculate regression parameters for each muscle group by using multiple linear regressions. Ninety-five percent prediction intervals were calculated as the muscle strength value ± the square root of the mean square error. Statistical analysis was performed by using BMDP software.^g

Back to Article Outline

Results 

Between-Center Variability 

Despite precise measurement procedures and careful evaluator training, ANOVA revealed a significant center effect on torque estimates (P<.001) for 13 of 27 muscle functions. These differences were because 1 center reported lower values, even after adjusting for age, sex, and weight (although there were no differences between centers for these parameters). After close examination of the data and a complementary study examining how the different examiners followed the measurement procedures, we decided to exclude this center for further analysis. When data were reanalyzed without the excluded center, 3 muscle functions still presented a significant difference between the remaining centers: shoulder abduction and hip flexion and extension. The standard errors of the predicted models were not significantly increased by reducing the number of subjects (53/315).

Strength Value in a Referent Population 

Data from the 3 remaining centers were pooled. Two hundred sixty-two subjects were finally analyzed. Their characteristics are given in table 4. Mean strength values ± standard deviation (SD) are given in table 2. The right side was dominant for 86% of the subjects. The dominant side was significantly stronger than the nondominant side (P<.05) for all functions except hip flexion and extension, knee flexion, and shoulder internal rotation, for which strength was similar for both sides (not shown).

Table 4. Subjects Characteristics

NOTE. Values are mean ± SD or as otherwise indicated.

Regression parameters of the covariates age, sex, and weight are listed in table 5 for each muscle group. Height was not considered in the regression model because this parameter was nonsignificant in the model for most of the muscle functions. All regression models were significant at P less than .001. Using these equations, it was possible to compute for a patient of any age, sex, and weight a predicted strength value for each muscle group and, hence, a relative deficit with respect to a normative value. It was also possible to standardize the measurements using z scores, as used for example by Tawil et al,¹⁶ according to the following equation:

Such scores give a quantified “distance” from normative data in number of SDs with respect to the average referent performance. It was then possible to compute composite scores for the whole body and for the upper and lower limbs. The mean of z scores was calculated, taking as a rule that in cases of missing data (eg, because of pain or retractions), it was necessary to have at least half the functions available to compute the composite score. An illustration of the usefulness of these calculations is appended in an attached case study.

Table 5. Regression Parameters for Strength Prediction

NOTE. Values are newton meters, except Neck flexion and Handgrip values, which are newtons.

We compared our data with published normative data on the U.S. population.¹⁵ On the whole, the strength measurements yield significant higher values for the American population except for hip extension, which was significantly higher for the French population. We computed the predicted strength for each muscle function in our study population with both regression equations from the American study and the present one. The French strength predictions were 10% lower on average compared with the U.S. predictions. The differences varied from −50% to 26% depending on the muscle function. The less comparable functions were hip (about −50% for flexion and 26% for extension), shoulder (about −15% for flexion and −17% for extension), and elbow extension (about −19%). The other functions gave smaller differences (close to ±5%). The ratio between right and left sides were higher when using the U.S. prediction models than the French ones.

Intra- and Interrater Reproducibility 

The ICC for intra- and interrater reliability data are given in table 6 for maximal values. Intrarater reliability can be considered as good to excellent on all functions (ICC>.75). When considering interrater reliability, only 3 functions had coefficients below .75: ankle dorsiflexion (left and right) and neck flexion.

Table 6. Results of Reproducibility Studies (ICCs)

NOTE. Values are kilograms.

⁎N=9.

Case Report 

To show the use of the normative database, we report the case of a 50-year-old, right-dominant male patient with an FSHD, weighing 64kg, who was seen at 1 of the centers for 3 visits at intervals of 3 months. The predicted values are listed in table 7 for each visit; absolute values are also expressed as a percentage of predicted values for each muscle function tested. At the first visit (M0), grip strength apart, the observed torques were severely reduced and ranged between 3.0% and 24.7% of normative values. Six months later (M6), strength was globally more impaired and ranged between 7.0% and 17.0%. Overall evaluation of progress was difficult from comparison of single-muscle functions. Computing composite scores allowed follow-up of the patient’s overall strength as assessed by the increase of z scores computed on the upper and lower limbs and for both combined (fig 1). Although considered to be only a slowly progressive disorder, it was possible to observe a decline in the strength of this patient with FSHD over 6 months. It was difficult for the same patient to assess any manifest decline with MMT scores.

Table 7. Strength Value for 1 FSHD Patient at 3 Successive 3 Monthly Visits

NOTE. Values are newton meters, except Handgrip values, which are newtons. In column 2 (predicted), values are the predicted value for each function according to the patient’s age, sex, and weight. For each visit, we provide the observed values at baseline and 3 and 6 months as well as the corresponding relative values as a percentage of predicted value, which gives an indication of the degree of impairment.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  Changes in composite MMT (points) and QMT (bars) scores for an FSHD patient at 3 successive visits (M0, M3, M6) 3 months apart. Composite scores were computed from QMT data for the (A) upper limb and (B) lower limb separately and (C) total for the whole body. Compared with table 7, for which it is difficult to see an overall trend in the patient’s levels of muscle strength because of the variability of results, the QMT data expressed as z scores gives a clear picture of the deteriorating strength if the patient.

Back to Article Outline

Discussion 

Usefulness of a Normative Database 

This study involved the development of a French isometric strength normative database for adults measured by using QMT. This will allow objective evaluation of patients with respect to these normative data, assessment of the degree of their neuromuscular deterioration, and collection of information on the clinical course of the disease. Using relative changes is not the same as using z scores. Indeed, for weak forces, a small change would lead to a strong relative change but only to a minor z-score change. Moreover, the use of z scores allows the patient to be situated with respect to normative values and reveals the profile of the deficit. For example, consider a patient presenting with a torque value for ankle flexion of 1.02Nm increasing to 2.42Nm 6 months later. This change corresponds to a variation of 137%. If the same values are expressed as z scores, the variation is only 4%.

A normative database also allows the computation of composite scores, which may be more robust and more sensitive than isolated muscle functions in the assessment of global improvements of patients involved in a therapeutic trial. As shown in the case study presented earlier, the analysis of individual muscle functions was insufficient for the overall evaluation of strength progression, which became possible by looking at the composite scores (see fig 1). Direct computation of composite scores from raw data is not valid because muscle strengths of different magnitudes measured around joints with different mechanical properties are not summative.¹⁷ This is still less valid in the follow-up of patients because different muscle functions can change inhomogeneously during a therapeutic trial.

We have also designed this protocol to get specific functions that have not been assessed yet in previous publications, such as shoulder internal and external rotation and neck flexion, which were required for use in a therapeutic trial in FSHD involving significant shoulder impairment.

Comparing our strength values with the American ones,¹⁵ significantly lower values were observed for our population, apart from hip extension. The predictive values deduced from our regression models were also lower than the predictive values computed from the American predictive regression models. The differences observed could have several origins such as the morphology of the subjects and the variables used in the models. The analyses in several studies on normative strength show a marked variation depending on the country, which can be partly explained by differences in methodologic procedures. However, differences could also be caused by morphologic, anatomic, ethnic, cultural, and social characteristics of the healthy populations that have been involved in such protocols. Interestingly, our strength values were slightly higher than or similar to ones presented by Meldrum et al¹⁹ for the Irish population.

When assessing strength around a joint, the measurement of force alone is not sufficient to provide a good estimate of muscle strength because it also depends on the lever arm. For example, a measurement error of 2cm on a lever arm measuring 20cm gives rise to an error of 10% in the force. It is, thus, essential to record measurement of strength as torque and to use the corresponding predictive regression models to give consistent comparisons between individuals. This point is particularly crucial when children are involved in the protocols.²⁰

Methodologic Issues 

The factors leading to variability in strength measurement are numerous and can be classified as technical, methodologic, environmental, and human factors.²¹ This study involved 4 centers and 10 clinical evaluators, increasing the risks of measurement variability.

For each protocol, functions to be evaluated should be carefully determined according to the disease, taking into account fatigue of patients, especially if other tests are also required. The time of day of testing must be defined. A learning session should also be organized before the trial (eg, during prescreening) so that patients can get familiar with the apparatus and procedures and initiate a relationship of confidence with the evaluator. Evaluation procedures must be described precisely to prevent more than 1 possible interpretation of the measurement process. In our protocol, the modus operandi was defined for positioning, installing, and stabilizing the subject and the strap. It was stipulated that the evaluators maintain the position of subject, even with straps, to avoid any compensation. In 1 center, this was not done systematically because the evaluator was not strong enough to stabilize the strongest subjects, resulting in lower estimates of MVIC. Stabilization methods are a critical concern for strong subjects and may be inefficient or inadequate. This issue has been discussed in some studies for particular functions such as hip extension,²² but this is likely to be true for any strong muscle function. When the aim is to assess the MVIC, it does not seem consistent to ask the subjects to stabilize for themselves the position of their body segments because the maximal force cannot be attained. However, it may be more reproducible to ask the subjects to stabilize themselves when they are strong. In general, measurement procedures should be adapted to the aim of the study and the populations involved. If a source of error is detected during a clinical trial, such as modifications in stabilization procedures or relative body part positions, it should not be addressed during the protocol but should be addressed in subsequent trials.

When assessing torques, lever arms must be precisely measured in the testing position because the joint rotation axis may move with respect to its position. This is, for instance, the case for shoulder abduction for which we observed a center effect probably related to various measurement procedures of the lever arm. This is also why precise anatomic landmarks must be defined and documented for each testing position.

When repeated evaluations are planned during a trial, it is highly recommended that each patient should be tested by the same examiner. Also, in a multicenter trial, repeated training sessions with all examiners should be organized at regular intervals (every 3 or 6mo).

The QMA software itself retains the highest value recorded on the force transducer during the effort. However, this maximal value can be situated on an overshoot or an artifact, which leads to an overestimation of the patients’ observed MVIC. Evaluators must make sure that such a situation does not occur.

Missing data can be a problem when computing megascores. Different situations can be responsible such as pain, retractions, or fatigue. If missing data are clearly caused by major impairment, the function should be recorded as 0. If other reasons apply, it is reasonable to compute the mean of the z scores if more than half the planned functions could be evaluated. However, simulations by bootstrapping have shown that scores computed from incomplete dataset are highly correlated with scores computed on complete datasets (R>0.9). This last observation is true only for healthy subjects as considered in the current study and is, however, certainly not true for patients because the composite score can be greatly biased depending on which functions on which sides are lacking. This issue deserves further work.

Predictive Strength Reliability 

We have proposed for each muscle function a predictive regression model performed using age, sex, and weight as also recently proposed for children by Eek et al.²⁰ In other predictive models, other variables were used instead of weight such as BMI¹⁵ or height.¹⁶, ¹⁷ In our study, height was not a significant predictor variable for most muscle functions. Although age, sex, and weight were significant for strength expressed either in kilograms or newton meters for most of the muscle functions tested, height was significant for only 3 of 27 functions when strength was expressed in kilograms and for 8 functions when strength was expressed in newton meters. We also tested the significance of BMI to explain muscle strength. In most cases (19/27 functions when strength was expressed in kilograms and 22/27 when strength was expressed in newton meters), this variable was not a significant predictor of strength. Moreover, as assessed by stepwise regression, height and BMI were less predictive variables with respect to age, sex, and weight. The regression coefficients were similar to previous studies with adults.¹⁵ These predictive regression models make it possible to assess the relative weakness of patients.

Reproducibility Issues 

Although performed on few subjects and for few repeated measurements, our results concerning reproducibility are good to excellent for most of the muscle functions tested, in agreement with previous studies.¹¹ This good reliability underlines that the learning effect in healthy adults tested by trained raters is minor.⁸, ²² Only with standardized operated procedures and repeated training sessions can satisfactory reproducibility be attained.

Back to Article Outline

Conclusions 

This study has led to the development of an isometric strength normative database for French adults by using QMT. The database will be used to compute composite scores in therapeutic trials to follow a global index of strength.

No consensus exists on the various methods to use for strength measurement.²³, ²⁴, ²⁵ No method is perfect or ideal yet, and none will probably ever be. The challenge is to provide each clinical trial with appropriate, standardized, reliable and sensitive outcome measurements.

Because therapeutic trials may concern rare disorders, multiple centers are often involved to reach the statistical power required to show treatment efficacy. Thus, it is fundamental that all centers use the same methodologic procedures to assess outcome measure such as strength. Rigorous training and monitoring are required before and during any therapeutic trial so as not to compromise the quality of its results.

Suppliers

Back to Article Outline

Acknowledgment 

We are grateful to Denis De Castro, MD, for his kind assistance in the language revision of the manuscript.

Back to Article Outline

References 

1. Cook JD, Glass DS. Strength evaluation in neuromuscular disease. Neurol Clin. 1987;5:101–123
   • View In Article
   • MEDLINE
2. Andres P, Skerry L, Munsat T. Measurement of strength in neuromuscular diseases. In:  Munsat T editors. Quantification of neurologic deficit. Stoneham: Butterworth; 1989;p. 87–100
   • View In Article
3. Edwards RH. Quantitative methods of assessing skeletal muscle function for therapeutic trials. Acta Cardiomiol III. 1991;2:185–198
   • View In Article
4. Andres PL, Hedlund W, Finison L, Conlon T, Felmus M, Munsat TL. Quantitative motor assessment in amyotrophic lateral sclerosis. Neurology. 1986;36:937–941
   • View In Article
   • MEDLINE
5. Iannaccone ST, Hynan LS American Spinal Muscular Atrophy Randomized Trial (AmSMART) Group. Reliability of 4 outcome measures in pediatric spinal muscular atrophy. Arch Neurol. 2003;60:1130–1136
   • View In Article
   • MEDLINE
   • CrossRef
6. Kissel JT, McDermott MP, Mendell JR, et al. FSH-DY Group Randomized, double-blind, placebo-controlled trial of albuterol in facioscapulohumeral dystrophy. Neurology. 2001;57:1434–1440
   • View In Article
   • MEDLINE
7. van der Kooi EL, Vogels OJ, van Asseldonk RJ, et al. Strength training and albuterol in facioscapulohumeral muscular dystrophy. Neurology. 2004;63:702–708
   • View In Article
8. Escolar DM, Buyse G, Henricson E, et al. CINRG Group CINRG randomized controlled trial of creatine and glutamine in Duchenne muscular dystrophy. Ann Neurol. 2005;58:151–155
   • View In Article
   • MEDLINE
   • CrossRef
9. Munsat TL. Development of measurement techniques. Neurology. 1996;47:S83–S85
   • View In Article
   • MEDLINE
10. Hoagland RJ, Mendoza M, Armon C, et al. Reliability of maximal voluntary isometric contraction testing in a multicenter study of patients with amyotrophic lateral sclerosis (Syntex/Synergen Neuroscience Joint Venture rhCNTF ALS Study Group). Muscle Nerve. 1997;20:691–695
    • View In Article
    • MEDLINE
    • CrossRef
11. Mayhew JE, Florence JM, Mayhew TP, et al. Reliable surrogate outcome measures in multicenter clinical trials of Duchenne muscular dystrophy. Muscle Nerve. 2007;35:36–42
    • View In Article
    • MEDLINE
    • CrossRef
12. Andres PL, Finison LJ, Conlon T, Thibodeau LM, Munsat TL. Use of composite scores (megascores) to measure deficit in amyotrophic lateral sclerosis. Neurology. 1988;38:405–408
    • View In Article
    • MEDLINE
13. Munsat TL, Andres PL, Finison L, Conlon T, Thibodeau L. The natural history of motoneuron loss in amyotrophic lateral sclerosis. Neurology. 1988;38:409–413
    • View In Article
    • MEDLINE
14. Conradi S, Ronnevi LO. Pitfalls in the evaluation of isometric strength (TQNE) data in ALS. J Neurol Sci. 1996;139(Suppl):60–63
    • View In Article
    • Abstract
    • Full-Text PDF (276 KB)
    • CrossRef
15. The National Isometric Muscle Strength (NIMS) Database Consortium. Muscular weakness assessment use of normal isometric strength data. Arch Phys Med Rehabil. 1996;77:1251–1255
    • View In Article
    • Abstract
    • Full-Text PDF (717 KB)
16. Tawil R, McDermott MP, Mendell JR, Kissel J, Griggs RC. Facioscapulohumeral muscular dystrophy (FSHD): design of natural history study and results of baseline testing (FSH-DY Group). Neurology. 1994;44:442–446
    • View In Article
    • MEDLINE
17. The FSH-DY Group. A prospective, quantitative study of the natural history of facioscapulohumeral muscular dystrophy (FSHD): implications for therapeutic trials. Neurology. 1997;48:38–46
    • View In Article
    • MEDLINE
18. Personius KE, Pandya S, King WM, Tawil R, McDermott MP. Facioscapulohumeral dystrophy natural history study: standardization of testing procedures and reliability of measurements (The FSH DY Group). Phys Ther. 1994;74:253–263
    • View In Article
    • MEDLINE
19. Meldrum D, Cahalane E, Conroy R, Fitzgerald D, Hardiman O. Maximum voluntary isometric contraction: reference values and clinical application. Amyotroph Lateral Scler. 2007;8:47–55
    • View In Article
    • MEDLINE
    • CrossRef
20. Eek MN, Kroksmark AK, Beckung E. Isometric muscle torque in children 5 to 15 years of age: normative data. Arch Phys Med Rehabil. 2006;87:1091–1099
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (286 KB)
    • CrossRef
21. Hogrel JY, Ollivier G, Desnuelle C. [Manual and quantitative muscle testing in neuromuscular disorders (How to assess the consistency of strength measurements in clinical trials?). [French] Rev Neurol (Paris). 2006;162:427–436
    • View In Article
    • MEDLINE
22. Meldrum D, Cahalane E, Keogan F, Hardiman O. Maximum voluntary isometric contraction: investigation of reliability and learning effect. Amyotroph Lateral Scler Other Motor Neuron Disord. 2003;4:36–44
    • View In Article
    • MEDLINE
    • CrossRef
23. Sanjak M. Measurement of strength: pro. Amyotroph Lateral Scler Other Motor Neuron Disord. 2002;3(Suppl 1):S45–S47
    • View In Article
24. Sorenson E. Measurement of strength: con. Amyotroph Lateral Scler Other Motor Neuron Disord. 2002;3(Suppl 1):S49–S50
    • View In Article
25. Miller RG. Measurement of strength: summary. Amyotroph Lateral Scler Other Motor Neuron Disord. 2002;3(Suppl 1):S51–S54
    • View In Article

• a Rose & Krieger GmbH, Potsdamer Str 9, 32423 Minden, Germany.
• b SM-250; Interface Inc, 7401 E Butherus Rd, Scottsdale, AZ 85260.
• c Qbitus, Units 11 & 12 Victoria Park, Lightowler Rd, Halifax, HX1 5ND, UK.
• d Neuro 40; Plinth 2000, Wetheringsett Manor, Wetheringsett, Stowmarket, Suffolk, IP14 5PP, UK.
• e Jamar; Kinetec, Tournes, 08014 Charleville-Mézières Cedex, France.
• f The Computer Source, 6045 Circle of Light, Gainesville, GA 30506.
• g Statistical Solutions, 999 Broadway, #704, Saugus, MA 01906.