Published on

Published in: Health & Medicine, Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide


  1. 1. Intrarater Reliability of Manual Muscle Testing andHand-held Dynametric Muscle TestingCAROLYN T. WADSWORTH,RUTH KRISHNAN,MARY SEAR,JEAN HARROLD,and DAVID H. NIELSEN Physical therapists require an accurate, reliable method for measuring muscle strength. They often use manual muscle testing or hand-held dynametric muscle testing (DMT), but few studies document the reliability of MMT or compare the reliability of the two types of testing. We designed this study to determine the intrarater reliability of MMT and DMT. A physical therapist performed manual and dynametric strength tests of the same five muscle groups on 11 patients and then repeated the tests two days later. The correlation coefficients were high and significantly different from zero for four muscle groups tested dynametrically and for two muscle groups tested manually. The test-retest reliability coefficients for two muscle groups tested manually could not be calculated because the values between subjects were identical. We concluded that both MMT and DMT are reliable testing methods, given the conditions described in this study. Both testing methods have specific applications and limitations, which we discuss. Key Words: Muscle contraction, Muscle performance, Physical therapy. Muscle strength assessment is an in- obtaining these measurements has not potential for human error. Up to 40%tegral aspect of physical examination. been ascertained. of the bodys muscles, however, are soKnowledge of muscle strength, that is, Manual muscle testing is the most small that the part of the grading sys-the muscles ability to produce tension, widely used clinical method of strength tem based on gravity resistance is in-assists an examiner in making a differ- assessment.1 Manual muscle testing is applicable.3,6,7ential diagnosis, measuring improve- based on a system of grading movement Iddings et al, in one of the few pub-ment or deterioration, determining against examiner or gravity resistance, lished studies that address the reliabilityfunctional impairment, and planning first used by Lovett in 1912.2,3 Several of MMT, found that MMT used in atherapeutic measures. Physical thera- aspects of this system have led to its clinical setting can be highly reliablepists have contributed to the develop- being classified as a semiquantitative despite differences in examiner trainingment of muscle strength assessment method of measurement. Some of the and testing techniques.8 Other research-methodologies and routinely use them MMT grades rely largely on an exam- ers, who appraised MMT for standardi-on a frequent basis. Not all methodol- iners judgment. For example, an ex- zation in the poliomyelitis vaccine trials,ogies, however, yield equally quanti- aminers subjective assessment of the also found it to be reliable.9,10 In con-fiable results. The need for accurate, amount of resistance applied is inherent trast, Beasley found that physical thera-reliable strength measurements is un- in the grading criteria for strengths in pists using MMT were unable to identifydeniable, but the best method for the Good to Normal range.4-7 In con- up to 50% of loss of knee extensor mus- trast, the grading criteria for muscle cle strength in patients with poliomye- strengths in the Poor to Fair range are litis.11 In this study, physical therapists Mrs. Wadsworth is Lecturer, Physical Therapy more objective because an examiner assigned Normal grades to muscles thatEducation, College of Medicine, The University of uses gravity as a standard. Another sub- were able to produce up to only half theIowa, 2600 Steindler Bldg, Iowa City, IA 52242(USA). jective aspect of MMT is that an exam- force on a cable tensiometer of age- Ms. Krishnan is Staff Physical Therapist, Moline iner must develop an internal basis for matched norms.11 Also, therapists usingGeneral Hospital, Moline, IL 61265. comparing test results, for example, MMT did not distinguish muscle Ms. Sear is Instructor in Physical Therapy, Uni-versity of Kansas, Kansas City, KS 66103. being able to sense normative values for strength differences of 20% to 25% be- Ms. Harrold is Staff Physical Therapist, Marian age, sex, or size of body parts. An ex- tween patients strong and weak sides.11Health Center, Sioux City, IA 51101. aminer also relies on personal experi- More studies are needed to substantiate Ms. Krishnan, Ms. Sear, and Ms. Harrold werestudents in the physical therapy program, The Uni- ence and skill to palpate muscles and one or the other contradictory conclu-versity of Iowa, at the time this study was con- tendons, detect substitution, and prop- sion of these studies.ducted. erly stabilize the patient for MMT. Use Dr. Nielsen is Associate Professor, Physical Ther- An alternative to MMT is a quanti-apy Education, College of Medicine, The University of the standardized MMT protocol, tative, instrumented system for assessingof Iowa. which specifies test positions and incor- muscle strength. The Graham-Desagu- This article was submitted May 2, 1986; was withthe authors for revision 15 weeks; and was accepted porates gravity as at least one objective liers force dynamometer, developed inNovember 3, 1986. Potential Conflict of Interest; 4. measurement criterion, helps reduce the London in 1763 to measure human1342 PHYSICAL THERAPY
  2. 2. RESEARCHmuscle strength while eliminating syn- need to identify a reliable, clinically ac- 1. Wrist extensors—tested with the sub-ergistic activity, was the earliest reported ceptable method of muscle strength as- ject seated in a chair with his forearminstrumented system.12 In 1798, Regnier sessment, we designed our study to de- supported on an adjacent bed, butin Paris invented an all-purpose portable termine the intrarater reliability of with his hand unsupported. The sub-dynamometer, which he used for the MMT and DMT, including subjective jects elbow was flexed to 90 degrees,first recorded attempt to study muscle between-method comparisons of relia- his forearm was pronated, his wriststrength in individuals with diseases.12 bility. We also examined the advan- was extended to 30 degrees, and hisNumerous devices have been designed tages, disadvantages, and clinical utility fingers were relaxed. The rater stabi-since to provide objective, quantitative of these two methods. We expected both lized the subjects forearm on the bedstrength measurements. They register methods to be reliable because of their and resisted movement through con-the peak force generated by a muscle widespread clinical acceptance; how- tact over the dorsum of the sub-through loading in tension or compres- ever, we hoped this investigation would jects hand.sion. provide more information regarding the 2. Shoulder abductors—tested with the Myometers, cable tensiometers, dy- discriminating capabilities of each. subject seated in a chair with hisnamometers, and grip meters all have shoulder abducted to 90 degrees, inbeen shown to be statistically reliable METHOD zero degrees of rotation, and with hisinstruments for testing muscle strength elbow extended. The rater stabilizedunder various conditions.4,5,13-19 For ex- Subjects the subjects shoulder girdle with oneample, Mathiowetz et al, who studied The human subjects committees of hand just proximal to the glenohu-grip and pinch strength using hand- The University of Iowa and St. Lukes meral joint and resisted movementheld grip dynamometers and pinch Hospital, Cedar Rapids, Iowa, approved through contact over the distal endgauges, found interrater correlation this study before subject screening. Cri- of the humerus.coefficients to be .97 or above and test- teria for subject selection included a 3. Hip flexors—tested with the subjectretest reliability coefficients .80 or minimum of Fair strength (able to as- seated in a chair with his back sup-above; they emphasize that raters must sume and hold test positions against ported and his pelvis stabilized byfollow standardized test procedures to gravity) in the muscle groups to be the chair; the subjects arms wereobtain these results.18 In a comparison tested, no spasticity, and the ability to across his chest, and his hip wasof a cable tensiometer, a Wakim-Porter isolate movements on command. We flexed to 135 degrees. The rater re-strain gauge, a spring scale, and a selected 13 subjects who met the criteria sisted movement through contactNeuman myometer, Clarke found the and consented to participate in the study over the distal end of the femur.tensiometer to be the most precise in- from inpatients receiving physical ther- 4. Elbow extensors—tested with thestrument with a test-retest interrater cor- apy at St. Lukes Hospital. We per- subject lying supine on a bed withrelation coefficient greater than .90.13 formed initial tests on all 13 subjects, his shoulder flexed to 90 degrees, in In the sparse literature comparing the but because of circumstances unrelated zero degrees of rotation, and his el-results obtained through MMT and to their disorders, two subjects could not bow flexed to 45 degrees. The ratermuscle testing with a hand-held force- participate in the retest session. The sub- stabilized the subjects upper armmeasuring device, one study measured jects had various chronic orthopedic and resisted movement through con-the hip flexor and abductor muscle and neuromuscular diagnoses, produc- tact over the distal end of the ulna.strength of 128 patients with unilateral ing muscle strength deficiencies that es- 5. Kneeflexors—testedwith the subjectlower extremity pathological conditions. sentially were stable. Their mean age lying prone on a bed with his hip inAn examiner using a hand-held device was 70 years, with a range of 21 to a neutral position and his pelvis sta-was able to detect consistently unilateral 96 years. bilized by the bed; the subjects kneeweakness as identified by another ex- was flexed to 75 degrees. The rateraminer using MMT.1 A similar study Procedure stabilized the subjects thigh againstcompared the data obtained from bilat- the bed and resisted movementerally testing hip flexors and abductors We used a test-retest design in which through contact over the distal endwith a hand-held device to the same one rater performed duplicate tests of his tibia.testers subjective identification of the (MMT and DMT) offivemuscle groups The testing position needed to bestronger limb. The testers subjective as- on the same patients, then repeated the modified for two subjects who could notsessment was correct in 82% of the tests two days later. We randomized the tolerate or assume the standard posi-cases, but he was unable to identify the type of testing and the order of the mus- tions. One subject was positioned pronestronger limb when the mean strength cle groups. A physical therapist with to test the elbow extensors, and the otherdifference between limbs was less than eight years of clinical experience served was positioned side lying to test the knee8.8% in abduction and less than 6.4% as the rater. She performed all tests on flexors. These modifications did not ap-in flexion.20 the patients involved (potentially weak- pear to hinder the rater. Few, if any, studies have determined ened) limbs. We examined five muscle We used the following protocolthe intrarater reliability of both MMT groups, selected for their ease of meas- throughout testing. An investigatorand hand-held dynametric muscle test- urement, that is, muscle groups with screened a subjects chart for pertinenting (DMT) to compare the two modes testing positions that could be assumed information and identified the side ofof testing. Most physical therapists as- readily by the patients and reproduced the involved limbs that were to be tested.sess muscle strength using one or both accurately and maintained by the ther- Just before testing, one investigator,methods. A reliability study would facil- apist. These muscle groups and their serving as a subject liaison, met with theitate their choice of method and confi- standardized testing positions are the subject to explain the procedure anddence in their results. Recognizing the following2,3: answer questions. She requested that theVolume 67 / Number 9, September 1987 1343
  3. 3. subject not converse with the rater in anattempt to eliminate bias regarding thesubjects status. She then prepared thesubject for testing. The rater entered theroom and began testing. Another inves-tigator, serving as a recorder, directedthe rater according to the preestablished,randomized mode and order of testing.She told the rater which side to test,which mode of testing to administerfirst, and the order in which to test themuscle groups. She also recorded allresults, verbalized by the rater, and com-puted the torque measurements ob-tained with the dynamometer. The rater performed MMT using thepositions previously described. She useda "break test," in which a subject ac-tively holds a body part in a prescribedposition, while an examiner attempts to"break" the hold by a manual counter-force (Fig. 1). The rater consistentlycommanded the subject to "hold, hold,dont let me move you" during the con-traction, which lasted about five sec-onds. The rater used the MMT grading Fig. 1. Testing kneeflexormuscles using manual muscle testing.system, described by Daniels andWorthingham2 and Kendall and Mc- 3 = Poor minus: The ability to moveCreary.3 We assigned ordinal values to the body part more than half waythe respective descriptive classification to completion of the test positionlevels for the purpose of statistical analy- with gravity lessened.sis as follows:11 = Normal: The ability to move the 2 = Trace plus: The ability to move the body part into the test position and body part less than half way to hold against gravity and maximum completion of the test position resistance. with gravity lessened.10 = Good: The ability to move the 1 = Trace: The ability to initiate a fee- body part into the test position and ble contraction or flicker of muscle hold against gravity and moderate or tendon movement that is visible resistance. or palpable, but does not move the 9 = Good minus: The ability to move body part. the body part into the test position Fig. 2. Hand-held dynamometer and tape 0 = Zero: No ability to contract the and hold against gravity and some muscle. measure used in dynametric muscle testing. resistance. 8 = Fair plus: The ability to move the The rater used a calibrated 25-kg Cha- measured the distance from the center body part into the test position and tillon* hand-held dynamometer for of the dynamometer pad to the joint hold against gravity and minimum DMT (Fig. 2). She chose a cuff of ap- axis. The recorder documented each propriate size to fit a subjects limb. She force measurement and calculated the resistance. held the dynamometer perpendicular to torques as the products of the distance 7 = Fair: The ability to move the body the limb at the specified contact points. and the forces. part into the test position and hold She tested the subjects in the positions The subject liaison investigator re- against gravity. previously described for MMT. She used mained with the subject to answer any 6 = Fair minus: The ability to move the body part more than half way a "make test," in which a subject exerts questions after testing. The same pro- to completion of the test position maximum force against the stationary cedure was followed during the retest, against gravity. dynamometer (Fig. 3). She used the ver- which was conducted two days later. bal command "Push, push. Is that all 5 = Poor plus: The ability to move the youve got?" during the contraction, Data Analysis body part less than half way to which lasted about five seconds. The completion of the test position A simple computer procedure was rater performed three trials and then against gravity. used to rank the MMT ordinal data in 4 = Poor: The ability to move the body ascending order (from lowest to highest) part into the test position with * John Chatillon & Sons, 83-30 Kew Gardens before the statistical analysis. The statis- gravity lessened. Rd, Kew Gardens, NY 11415. tical analysis involved computing be-1344 PHYSICAL THERAPY
  4. 4. RESEARCH ferences during DMT for the other mus- cle groups. DISCUSSION As reflected by the between-test cor- relation coefficients and lack of signifi- cant mean differences, MMT for shoul- der abduction, hip flexion, and knee flexion demonstrated good intrarater re- liability in this study. These results com- plement the findings of Iddings et al8 and Lilienfeld et al,10 which showed that MMT had high interrater reliability. These results, therefore, suggest that MMT is a consistent evaluation tool. Our results, however, also reflect that MMT is less discriminating than DMT in identifying small differences in mus- cle strength. Because we adhered to the MMT grading system with inherently few increments, MMT was less sensitive than DMT to variability in strength of muscles in the Fair to Normal range. Beasley also identified this limitation ofFig. 3. Testing knee flexor muscles using dynametric muscle testing. MMT.11 Thus, physical therapists can use MMT with confidence in its relia-TABLE bility, but may not be able to detectTest-Retest Reliability Coefficients muscle strength increments as discretely Dynametric as with instrumented systems. Manual Muscle Dynametric muscle testing also dem- Muscle Muscle Groups Testing onstrated high intrarater correlation Testing (r) (r) coefficients. In contrast to the almost Wrist extensors a .88b identical test-retest mean MMT values, Shoulder abductors .98b .69 the retest mean DMT values were higher Hip flexors .74b .72* for all five muscle groups, with two tests Elbow extensors a .90* being statistically significant. We attrib- Knee flexors .63 .75* uted the observed increases to a learning a effect, which could be minimized by Ellipsis indicates that reliability coefficient could not be calculated for this muscle testbecause the MMT grade was essentially the same for all subjects for the test-retest data. With familiarizing subjects with the dyna-the exception of one subject who had an MMT grade of 10 on one test and a grade of 11 on mometer during a practice session. (Per-the other test, all subjects demonstrated the grade of 11 for both tests. haps further study is indicated to deter- * Statistically significant from zero (p < .05). mine the amount of practice needed before stabilizing of values.) Our results,tween-test Pearson product-moment re- elbow extensor muscles because, with thus, support physical therapists use ofliability coefficients and between-test the exception of one subject, the subjects a hand-held dynamometer as a reliableStudents paired t tests on the DMT and demonstrated a grade of 11 on both the muscle strength testing tool.ranks of the MMT data. Pearson prod- test and retest, invalidating this statisti- Manual muscle testing and DMTuct-moment correlation analysis and cal technique. The test-retest reliability were comparably reliable, given the con-Students t tests on the ranks of the coefficients for the DMT ranged from ditions described in this study, but eachordinal data are equivalent to Spear- .69 to .90. With the exception of the method has specific applications andmans rank order correlation analysis shoulder abductor muscles, the test-re- limitations. Manual muscle testing isand the Wilcoxon matched-pairs signed- test reliability coefficients for all muscle clinically versatile and inexpensive toranked test, respectively.21 groups during DMT were statistically administer, requiring only an appropri- significant (p < .05). ate supportive base for a subject. ThisRESULTS The paired t test revealed that no sig- system is devised to measure the entire nificant test-retest mean differences oc- range of muscle strength from Zero to As shown in the Table, the test-retest curred during MMT for all muscle Normal, but is limited to only 12 ordinalreliability coefficients for the MMT groups (p > .05) (Fig. 4). The paired t values that produce an inherent limita-ranged from .63 to .98 and were statis- test, however, did demonstrate signifi- tion in measurement resolution. Usingtically significant (p < .05) for the shoul- cant test-retest mean differences during this system, a tester can grade only clin-der abductor and hip flexor muscles. DMT for the wrist extensor and elbow ically detectable weakness and thus mayThe correlation coefficients could not extensor muscles (p < .05) (Fig. 5). We not be able to discriminate accuratelybe calculated for the wrist extensor and found no significant test-retest mean dif- small variations of strength, particularlyVolume 67 / Number 9, September 1987 1345
  5. 5. in the Good to Normal range. Deviationfrom standardized testing procedures Testand the subjectivity of the grading sys- Retesttem are potential sources of error. Muscle testing with a hand-held dy-namometer provides a more objectivemeans of muscle strength measurementthan MMT with a continuous range oftorque values. Use of an instrument,however, adds to the cost and complex-ity of testing. This method is most usefulfor measuring muscle strength greaterthan Fair, although some clinicians useit in gravity-eliminated positions.22 It isinapplicable for measuring muscle strength less than Poor. Use of a hand- held dynamometer also is limited in the upper muscle strength ranges where an examiner may have difficulty stabilizing the instrument and resisting the subject. An examiners own strength may affect Fig. 4. Manual muscle testing: test-retest mean grades and standard errors (N = 11). Al- his ability to use a dynamometer appro- though grades 1 through 11 were possible, because of criteria for subject selection, only grades priately with stronger subjects. 9 through 11 were present in subjects. None of the test-retest mean differences were statistically Recommendations for further studies significant (p > .05). include recognizing the numerous vari- ables that can affect the reliability of both manual and dynametric methods. Maximal muscle contraction is limited Test ultimately by structural properties, Retest neural activation, and feedback mecha- nisms.23,24 Numerous external factors also affect a subjects ability to exert a maximal contraction, such as pain, mo- tivation, cooperation, limb and body position, physical condition, feedback of results, instructions, competition, fear, and incentives.7,17,18,23,25 McGarvey et al even found a statistically significant change in isometric strength with time of day, ranging from 3.97% to 7.22% for different muscle groups.17 An examiners technique, which var- ies according to training, experience, strength, and standards, also may influ- ence the results of muscle strength meas- urements.5-8,25 Other factors that must Fig. 5. Dynametric muscle testing: test-retest mean torques and standard errors (N = 11). be considered during muscle strength Note: Asterisks indicate statistically significant test-retest mean differences (p < .05). testing include the type and velocity of contraction, warm-up activity, and ad- equate stabilization.24 Kroll proposed jects must have at least a Fair grade for days. We also minimized rater contact that the timing of trials may introduce all muscle groups tested placed a limi- with the subjects by having different experimental error; inadequate recuper- tation on the range of strength of the persons screen the subjects and record ation time between trials may cause en- subjects. This limitation decreased the the test values. ergy depletion with apparent strength variance and made assessment of relia- decrease, whereas successive daily bility more difficult. We propose ex- CONCLUSIONS sessions may produce a training or tending this investigation to include This study explored two common learning effect with apparent strength in- more subjects and expanding the criteria methods of assessing muscle strength, crease.14-16 Body position and joint an- for subject selection to allow a greater MMT and DMT. Given the restrictions gles have been shown to be major vari- variance in strength. described in this study, we consider both ables in muscle strength assessment, The potential for rater bias carrying methods of strength testing to be reliable thus advancing the use of standardized over from the initial test to the retest for the muscle groups we tested. We also test positions.5,26 exists in a study such as ours. We at- identified limitations of each method, This study was based on a relatively tempted to minimize this bias by sepa- noting that particularly in stronger sub- small sample. The criterion that all sub- rating the test and retest sessions by two jects the dynamometer becomes more 1346 PHYSICAL THERAPY
  6. 6. RESEARCH 5. Clarke HH, Elkins EC, Wakim KG: Relationship 16. Kroll W: Reliability of a selected measure ofdifficult to stabilize and MMT becomes between body position and the application of human strength. Research Quarterly 33:410-less discriminating. With confidence in muscle power to movements of the joints. Arch 417, 1962both methods, we recommend allowing Phys Med 31:81-89, 1950 17. McGarvey SR, Morrey BF, Askew U , et al:the subjects physical condition and the 6. Hosking GP, Bhat US, Dubowitz V, et al: Meas- Reliability of isometric strength testing: Tem- urements of muscle strength and performance poral factors and strength variation. Clin Or-clinical or laboratory settings test objec- in children with normal and diseased muscle. thop 185:301-306, 1984tives to determine the test of choice. Our Arch Dis Child 51:957-963, 1976 18. Mathiowetz V, Weber K, Volland G, et al: Re-results support the reliability of testing 7. Wakim KG, Gersten JW, Elkins EC, et al: Ob- liability and validity of grip and pinch strength jective recording of muscle strength. Arch Phys evaluations. J Hand Surg [Am] 9:222-226,methods that physical therapists com- 1984 Med 31:90-99, 1950monly use and thus contribute to the 19. Wiles CM, Kami Y: The measurement of mus- 8. Iddings DM, Smith LK, Spencer WA: Musclesignificance of the MMT and DMT testing: Part 2. Reliability in clinical use. Phys cle strength in patients with peripheral neuro-methods. Ther Rev 41:249-256, 1961 muscular disorders. J Neurol Neurosurg Psy- chiatry 46:1006-1013, 1983 9. Gonnella C, Harmon G, Jacobs M: The role of the physical therapist in the gamma globulin 20. Saraniti AJ, Gleim GW, Melvin M, et al: The Acknowledgments. We thank the poliomyelitis prevention study. Phys Ther Rev relationship between subjective and objectiverater, Phyllis Griffin, LPT, and John 33:337-345, 1953 measures of strength. Journal of Orthopaedic and Sports Physical Therapy 2:15-19, 1980Wadsworth, MA, LPT, Director of 10. Lilienfeld AM, Jacobs M, Willis M: A study of 21. Conover WJ, Iman RL: Rank transformationsPhysical Therapy, both of Saint Lukes the reproducibility of muscle testing and certain as a bridge between parametric and nonpara- other aspects of muscle scoring. Phys TherHospital, Cedar Rapids, IA, for their Rev 34:279-289, 1954 metric statistics. The American Statistician 35(3):124-132, 1981assistance in the completion of this 11. Beasley WC: Influence of method on estimates 22. Darcus HD: A strain-gauge dynamometer forstudy. of normal knee extensor force among normal measuring the strength of muscle contraction and postpolio children. Phys Ther Rev 36:21- and for re-educating muscles. Annals of Phys- 41,1956 ical Medicine 1:163-176, 1952REFERENCES 12. Peam J: Two early dynamometers: An histori- 23. Kroemer KHE, Marras WS: Towards an objec- 1. Marino M, Nicholas JA, Gleim GW, et al: The cal account of the earliest measurements to tive assessment of the "maximal voluntary con- efficacy of manual assessment of muscle study human muscular strength. J Neurosci traction" in routine muscle strength measure- strength using a new device. Am J Sports Med 37:127-134, 1978 ments. Eur J Appl Physiol 45:1-9, 1980 10:360-364,1982 13. Clarke H: Muscular Strength and Endurance in 24. Smidt GL, Rogers MW: Factors contributing to 2. Daniels L, Worthingham C: Muscle Testing: Man. Englewood Cliffs, NJ, Prentice-Hall Inc. the regulation and clinical assessment of mus- Techniques of Manual Examination, ed 4. Phil- 1966, pp 10-16 cular strength. Phys Ther 62:1283-1290, 1982 adelphia, PA, W B Saunders Co, 1980 3. Kendall FP, McCreary EK: Muscles: Testing 14. Kroll W: Reliability variations of strength in test- 25. Edwards RHT, McDonnell M: Hand-held dy- and Function, ed 3. Baltimore, MD, Williams & retest situations. Research Quarterly 34:50- namometer for evaluating voluntary-muscle Wilkins, 1983 55, 1963 function. Lancet 2:757-758, 1974 4. Borden R, Colachis SC: Quantitative measure- 15. Kroll W: A reliable method of assessing iso- 26. Elkins EC, Leden UM, Wakim KG: Objective ment of the Good and Normal ranges in muscle metric strength. Research Quarterly 34:350- recording of the strength of normal muscles. testing. Phys Ther 48:839-843, 1968 355, 1963 Arch Phys Med 32:639-647, 1951Volume 67 / Number 9, September 1987 1347