UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Clinical significance and reliability of two common grip strength measures Burns, Aglaia Jane 1998

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1998-0370.pdf [ 8.7MB ]
Metadata
JSON: 831-1.0099247.json
JSON-LD: 831-1.0099247-ld.json
RDF/XML (Pretty): 831-1.0099247-rdf.xml
RDF/JSON: 831-1.0099247-rdf.json
Turtle: 831-1.0099247-turtle.txt
N-Triples: 831-1.0099247-rdf-ntriples.txt
Original Record: 831-1.0099247-source.json
Full Text
831-1.0099247-fulltext.txt
Citation
831-1.0099247.ris

Full Text

THE CLINICAL SIGNIFICANCE AND RELIABILITY OF TWO COMMON GRIP STRENGTH MEASURES By A G L A I A J A N E B U R N S B.Sc. (Hons) , The University of Toronto, 1983 B .H.Sc . (PT) , McMas te r University, 1988 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S Schoo l of Rehabil i tation Sc iences W e accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F BRIT ISH C O L U M B I A September 1998 © Ag la ia Jane Burns, 1998 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^ U 1^*2- dtU W The University of British Columbia Vancouver, Canada Date Oct • WW DE-6 (2/88) Abstract This study surveyed therapist's use of grip strength measures in clinical practice and determined the test-retest reliability of two common grip strength tests to elucidate what drives clinical decision making with respect to grip strength measurements and to determine an effective measurement strategy to detect clinically important change. Survey responses were obtained from 120 facilities across Canada. Results indicated that the Jamar dynamometer was the instrument most commonly used to measure grip strength (87%), and the Standard Grip Strength Test (SGST) (3 trials at position #2) and the 5-Position Grip Strength Test (5PGST) (1 trial at each of 5 positions) were the most common tests used. A normative database for comparison of test results is not consistently used in clinical practice. From our survey results, there was considerable variation in what clinicians deemed a clinically important change in grip strength measurements. Fourteen healthy males (mean age 25.4 years) were involved in a test-retest reliability study of the S G S T and the 5 P G S T using the Jamar dynamometer. Generalizability coefficients (R) were used to determine test-retest reliability. Generalizability coefficients for the S G S T were greater for the mean of 3 trials (R =.67 for the right hand and R=.70 for the left). Handle position of the 5 P G S T was reliable for the left hand only at position 1(R=.87) and position 2 (R=.83). The greatest source of error was seen between subjects across test occasions. Grip strength fluctuations were between 2.8 to 6.6 kgf (6.4 to 12.8%) in the right hand and 2.8 to 5.9 kgf (5.1-10.8%) in the left over 2 test occasions. It appears that clinicians may be stating a true change in grip strength scores when changes in measurement may be due to error alone. From our study, it is recommended that a standardized test protocol be used consistently for grip strength measurements. The clinician needs to determine normal variability for their patients using the standard error of measurement (SEM) and confidence intervals. True grip strength may best be evaluated by taking multiple measurements over a number of test occasions. The use of a normative database should only be compared to when the same test protocol, sample population and instrument used by the clinician has been used to develop the normative data. iii Table of Contents Abstract ii Table of Contents iii List of Tables v List of Figures vi List of Abbreviations/Definitions viii Acknowledgement ix Chapter 1 Introduction and Literature Review 1 Theoret ical Constructs 3 Grip Strength 4 Measurement Tool 4 Factors Affecting Grip Strength 6 Rat ionale for Investigation 8 Reliabil ity of Grip Strength Measures 8 Cl in ical Signi f icance of Grip Strength Measu res 11 Conc lus ion and Purpose of Resea rch 15 Refe rences 16 Chapter 2 Survey Study 21 Introduction 21 Methods 22 Resul ts 24 D iscuss ion 32 Cl in ical Signi f icance 36 Limitations of Survey 39 Further Resea rch 40 Conc lus ion 40 Refe rences 42 Chapter 3 Reliability Study 43 Introduction 43 Purpose of the Study 45 Statement of Hypotheses 46 Methods 46 Statistical Ana lys is 51 Resul ts 53 D iscuss ion 59 Cl in ical Signi f icance 67 Limitations of Survey 69 Further Resea rch 71 Conc lus ion 71 Refe rences 73 Chapter 4 Summary and Conclusions 76 What is the normal variation in grip strength measu res? 77 How reliable are grip strength measures? 80 How can we interpret the results of grip strength measu res? 81 Recommendat ions 84 Suggest ions for further research 86 Conc lus ion 86 References 88 Append ices A : Descript ion of the S G S T , 5 P G S T 90 B: Survey Form 91 C : C o v e r Letter for Survey 93 D: R e s e a r c h Design 94 E: Standard ized Patient Instructions for Gr ip Strength Test ing 95 F: R a w Data 96 G : Statistical Ana l yses /Anova Tab les 100 V List of Tables Chapter 3 Tab le 1: S a m p l e means between test occas ions and across trials for the Standard Grip Strength Test 53 Tab le 2: General izabi l i ty Coeff icients (R) for the Standard Grip Strength Test 54 Tab le 3: Est imate of var iance components derived from the 3-way A n o v a for the Standard Grip Strength Test 55 Tab le 4: Summary of the Correlat ion Coeff icients for the 5-Posit ion Gr ip Strength Test 57 Tab le 5: Est imate of var iance components from a 2-way A n o v a for the 5-Posi t ion Gr ip Strength Test 57 Tab le 6: M e a n grip strength measurement fluctuations (kgf) for the 5-Posit ion Grip Strength Test and the Standard Grip Strength Test (n=14, 2 test occasions) 58 vi List of Figures Chapter 2 Figure 1: Survey responses indicating the types of grip strength used 25 Figure 2: Survey responses of grip strength measurement tools used in cl inical practice 25 Figure 3: Gr ip strength testing by patient population 26 Figure 4: Survey response depicting the intended purpose of grip strength measurements 26 Figure 5: Survey response data depicting the use of a normative da tabase in grip strength measurements 27 Figure 6: Survey responses depicting measures used to detect clinically important change in grip strength 28 Figure 7: Survey response indicating amount of change in kgforce to detect clinically significant change in grip strength measurements between hands 29 Figure 8: Survey response data indicating percentage difference between hands considered clinically significant with grip strength measurements 29 Figure 9: Survey responses depicting clinically significant percentage change between trials reported in grip strength measures 30 Figure 10: Survey responses depicting clinically significant change in kgforce between trials in grip strength measurements 30 Figure 11: Survey response data indicating percentage difference between grip strength scores and a normative database to be cons idered clinically significant 31 Figure 12: Survey response data indicating absolute difference between a normative database and grip strength measurements for cl inical s igni f icance 31 Chapter 3 Figure 1(a,b):Posit ioning for grip strength testing with the J a m a r dynamometer 49 Figure 2: Proport ion of high scores occurring on the first, second or third of 3 trials on the Standard Grip Strength Test 56 Figure 3: Test-retest reliability of 5 handle posit ions of the Jamar dynamometer 59 V l l l List of Abbreviations/Definitions Accuracy: Instrumentation accuracy refers to the exac tness of the measurement provided by an instrument by comparing the reading on the dev ice with a standard measure (i.e. hanging known weights on the dynamometer) . Correlation Coefficient: Mathematical express ion of the degree of relationship between two or more var iables. Generalizability Theory: Prov ides a method for assess ing the dependabi l i ty of a measure and al lows for the decomposi t ion of the error var iance into unique sources of variation. Interrater Reliability: The consistency among different tester's ratings of the s a m e object or response. Intrarater Reliability: The consistency with which one rater ass igns scores to a single set of responses on two or more occas ions . Reliability: The extent to which measurements are repeatable. Test-Retest Reliability: The degree to which the scores change or remain stable on repeated administration. Validity: The meaningfu lness of test scores as they are used for speci f ic purposes. The extent to which a test measures what it is intended to measure . SGST: Standard Grip Strength Test 5PGST: 5-Posit ion Grip Strength Test kgf: k i lograms of force SEM: Standard error of measurement Cl: Conf idence interval ICC: Intraclass correlation coefficient r: Pea rson Product Moment correlation R: General izabi l i ty correlation coefficient ix Acknowledgement I would like to thank my supervisor, Dr. Donna Maclntyre, for her knowledge, support and guidance throughout my graduate research. Her adv ice and constructive crit icism made the process a very rewarding exper ience. A s wel l , I would like to thank my committee members, Dr. Joe l S inger and Dr. Jan i ce Eng for their invaluable input and direction throughout the design and implementat ion of my research. I would a lso like to thank my col league, Mr. Pau l Stratford, for his ass is tance and help with the data analys is . The support of the Schoo l of Rehabil i tation Sc iences , University of British Co lumb ia , faculty and co l leagues throughout graduate schoo l was greatly apprec iated. Finally, I would like to thank the B C Heal thcare Scholarsh ip Fund for f inancial support. 1 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW The need for establ ishing objective measures in physical and occupat ional therapy and determining their reliability and clinical s igni f icance as rehabilitation tools w a s the motivating factor behind this investigation. Grip strength testing utilizing a standard J a m a r dynamometer (As imow Engineer ing Company , Los Ange les , C A ) 1 w a s selected for investigation due to its widespread use as a measurement tool in the cl inical rehabilitation sett ing. 2 The ability to document change with val id, reliable and objective outcome measurement tools is extremely important in any field of study. In rehabilitation medic ine, objective measures can be used for a variety of purposes including establ ishing basel ine pathology, document ing change, planning and evaluat ion of treatment programs, explaining performance and predicting rehabilitation ou tcome. 3 , 4 A measure is objective if it is not dependent on the e x a m i n e r 5 , involves the use of instrumentation, and is expressed in real numbers . 3 Object ive measures are tests or procedures whose results are independent of personal feel ings or prejudices. 6 Subject ive measures , on the other hand, are dependent on the patient and to s o m e degree, on the judgement of the examiner. Accord ing to Bohannon 3 , in phys ica l therapy, many, if not most, of the measures used are dependent on the patient and no matter how objective the test, variability within the patient and their level of effort is 2 bound to have an affect on the outcome. 3 A s Young et a l 7 stated, it is a common misconcept ion that all tests that generate numbers are necessar i ly objective; this is particularly true of grip and pinch strength measurements . 7 Object ive measures have potential advantages over subjective measures including indices of reliability and validity. Reliability is the degree to which test scores are free from errors of measurement and def ines an instrument's ability to measure consistent ly and predictably. 4 , 8 Validity is defined as the extent to which a test measures what it is intended to measure . Reliability is a prerequisite to establ ishment of validity but validity is often cited as the most important considerat ion when testing a measure as it refers to the appropr iateness, meaningfu lness and usefu lness of a measure . 9 W h e n using measurement tools in rehabilitation, exper ienced professionals should be famil iar with the reliability and validity for the speci f ic application of the tool as well as knowing the populat ions for whom the measure was des igned prior to attempting s tandard ized a s s e s s m e n t s or making clinical dec is ions regarding ou tcomes . 1 0 Theoretical Constructs Grip strength testing is commonly administered in both cl inical and evaluat ive work capaci ty sett ings. Measu res of grip strength are frequently used in rehabilitation medic ine to compare an individual's grip strength with establ ished norms or to compare affected and non-affected limbs within the s a m e individual. A s indicated by Hamil ton, Ba lnave and A d a m s 1 1 , the outcome of grip strength testing can be a major determinant of whether rehabilitation will be offered, what type of rehabilitation will be offered or as a measure of sincerity of effort in medico- legal i ssues . 1 1 The results of grip strength testing are often considered indicators for f i tness to return to work. 1 2 However , 3 measurements of strength are constructs that are in themselves not directly observab le . 8 There are no absolute standards to determine if current measu res of strength are valid indicators of what they are supposed to measure . This presents a real d i lemma in physical therapy as so much of our treatment centers on " increasing strength" and "determining weakness " in order to progress a client in rehabilitation. O n e of the theoretical constructs of physical therapy is the concept of movement dysfunct ion, or as Hislop 1 3 stated, pathokinesiology. Strength, within a pathokinesiological framework, relates to kinematics and kinetics. Kinetic var iables with estab l ished measurement units such as force and torque are often used in physical therapy to a s s e s s movement dysfunct ions. The value behind using these terms is that they are establ ished and commonly accepted. A s S m i d t 1 4 indicated, phys ica l therapists will be unable to develop theories of movement unless the phenomena measured are clearly def ined in universally acceptable terms. W h e n evaluat ing strength, it is imperative that the examiner feels confident that the measure of strength utilized is valid or meaningful . The examiner must a lso feel confident that the measure obtained is a true measure of maximal strength in order to accurately make dec is ions regarding rehabilitation. Whi le musc le strength is a s s e s s e d frequently by physical therapists, there appears to be cons iderab le variability in the methods and tools used . Manua l musc le strength testing is used by most therapists but it has not been determined to be re l iable. 1 5 Hand held dynamometers , hand grip dynamometers such as the Jamar , or isokinetic dynamometers such as the Kin-Corn are other methods of measur ing musc le strength. S o m e studies have indicated scores obtained from manual musc le testing correlate with 4 hand-held dynamometry . 1 6 Stratford and B a l s o r 1 7 , in a compar ison of the K in-Corn and a hand-held dynamometer for tests of muscle strength of the e lbow flexors, found that hand-held dynamometry is a viable alternative to costl ier tools such as the K in-Corn , provided that the assesso r ' s strength is greater than that of the speci f ic musc le group being measured . Hand held and isokinetic dynamometers, along with the J a m a r dynamometer , attempt to objectively measure musc le strength and are used in cl inical pract ice. However , as measurement tools, each has its own inherent set of limitations that cl inicians need to be aware of before using them in cl inical pract ice. Grip Strength Measurement Too l A n assessmen t of grip strength first introduced by B e c h t o l 1 8 in 1954, and recommended by the Amer ican Society of Hand Therap is ts 2 is widely reported in the literature. It incorporates the Jamar hand-held dynamometer (As imow Engineer ing C o . , Los Ange les , Cal i f . ) . 1 Whi le there are many hand dynamometers currently on the market, the J a m a r hand-held dynamometer remains one of the most popular cl inical tools to a s s e s s grip strength in the rehabilitation setting. Kirkpatrick 1 9 sugges ted the J a m a r test be used by the Amer ican Medica l Assoc ia t ion in the evaluat ion of all upper extremity impairment ratings. The literature indicates that the Jamar dynamometer is widely used in rehabilitation and is deemed the most accurate for determining grip s t rength. 2 0 Smith and Benge found in a survey of 195 occupat ional therapy facilit ies over 7 9 % of the cl inics commonly used the Jamar dynamometer in their set t ings. 2 1 The J a m a r dynamometer consists of a handle with 5 equal ly spaced adjustable sett ings. The dynamometer is presented to the individual by the examiner and the 5 individual is instructed to grip the handle as hard as possib le. In utilizing the J a m a r hand held dynamometer in grip strength testing, isometric strength is measured and the subject is unable to perceive the distance the handle has moved during grasp. During an isometric contraction, the potential tension that can be generated by musc le is c losely related to the length of the muscle with the potential strength decreas ing when the musc le length is either longer than or shorter than the resting length. 2 2 During grip, the b iomechanica l advantage of the finger joints and tendons is determined by the shape and s ize of the object being grasped therefore grip force will be affected by musc le l e n g t h . 1 8 2 2 Utilizing the Jamar dynamometer, the second (4 cm) or third (6 cm) posit ion usually al lows most individuals to apply maximal force comfortably. Accord ing to Bech to l 1 8 , grip strength is a function both of the s ize of the object being gr ipped and the ability of opposit ion between the thenar musc les and the four f ingers. 1 8 Genera l ly between the second and fourth setting on the dynamometer, the subject reaches a mechan ica l advantage depending on the s ize of their hand and it is at this setting that they are able to demonstrate the greatest static grip strength. 1 8 Instructions for the Jamar dynamometer do not provide any speci f ic protocols or norms 1 and it is regarded to be an instrument lacking in precision and accuracy for measur ing grip strength force according to Mr. Rick Seifarth at the National Institute of S tandards and Techno logy (NIST), Gathersburg, Mary land (personal communicat ion). The NIST (equivalent to the National Resea rch Counc i l in Canada ) is regarded as the national standard in the United States against which an assessmen t tool must be tested in order to determine accuracy within its measurement unit. 2 3 However, in communicat ion with a representative from NIST, it appears there are no s tandard ized 6 testing procedures relating specif ical ly to the Jamar dynamometer . C o m p a n i e s that calibrate dynamometers by hanging known weights from the dynamometer handle may exhibit a chain of traceability back to the NIST, but there is no gold standard with which to compare the force output of a standard Jamar dynamometer . Nonethe less , the J a m a r dynamometer is widely used as a measurement tool in rehabil i tat ion. 2 Factors Affecting Grip Strength Many factors influence grip strength itself including age, handedness , fatigue, pain, posit ioning and the client's cooperat ion. Normative data bases for grip and pinch strength relating to var ious population groups have appeared in the l i terature. 2 4 2 6 Stud ies have indicated that grip strength increases with a g e . 2 7 , 2 8 Mathiowetz et a l 2 5 in a study to determine normative grip and pinch strength data for adults indicated that the highest grip scores occurred in the 25 to 39 age groups. Ma les peaked in grip strength between the ages of 20-39 and women between 40 to 49 years of a g e . 2 9 Gr ip strength has been reported to diminish curvil inearly with age in elderly indiv iduals. 3 0 Ma les appear to be stronger than females in all age g roups . 2 7 3 1 H a n d e d n e s s plays a role in grip strength testing with the dominant hand largely thought to p o s s e s s 10% greater grip strength than the non-dominant hand (10% rule). Stud ies have shown that the 1 0 % rule appears valid for right handed persons only, with grip strength equivalent to the right in left handed i nd i v i dua l s . 3 2 , 3 3 Hand dominance w a s judged to be an unimportant variable in children aged 5 to 1 2 . 2 8 Harkonen et a l 3 1 in a sample of 204 Finnish adults aged 30 to 50 years, found no significant dif ference in strength between the dominant and non-dominant hands. The type of work that an individual does has an effect on their grip strength. Heavy manual workers exhibit the 7 greatest strength and the least difference between left and right hands while office workers have the weakest grips with the greatest difference between h a n d s . 3 4 Posit ioning of the upper extremity has been shown to have an effect on grip strength testing with the Jamar dynamometer and remains controversial . A wrist posit ion of 35 degrees of extension and 7 degrees of ulnar dev ia t ion 3 5 with the e lbow at 90 degrees of f lex ion 3 0 has been shown to produce the greatest force during grip strength testing. S u et a l 2 9 investigated shoulder and e lbow posit ions on grip strength in 160 men and women and found that grip strength was lowest with the shoulder in 0 degrees of f lexion and the e lbow at 90 degrees (standard protocol) as compared with the greatest grip strength found with the arm overhead (shoulder at 180 degrees of flexion) and the e lbow extended. Hand circumference in relation to grip strength is a lso important, particularly when testing all 5 posit ions on the J a m a r . 3 0 Time of testing and fatigability are important factors to cons ider with any musc le test ing. Activi t ies the individual has been doing prior to testing is important to note. Whi le fatigue may be a factor in grip strength measurements , the literature is inconsistent. Mathiowetz et a l 2 5 and Lusardi and B o h a n n o n 3 6 reported no difference within multiple trials but F e s s 3 7 indicated fatigue may be a factor during the Standard Grip Strength Test . Mathiowetz 3 8 did not find fatigue to be a clinically signif icant factor in the Standard Grip Strength Test of three grip strength trials with a 15 second inter-trial rest per iod. It has been suggested that two trials be recorded with a five minute rest between tr ials 2 1 , however there does not appear to be any standardizat ion in a review of the literature. Mar ion and Niebuhr 3 9 investigated the effect of warm-up trials prior to maximal grip strength testing and found that warm-up (submaximal gripping) had the 8 effect of improving maximal grip force va lues, however their sample s ize w a s smal l (n=15). Grip strength va lues do not appear to vary from morning to af ternoon 7 . The use of verbal encouragement , while ment ioned in the literature as part of the testing protocol, has not been specif ical ly studied with respect to grip strength test ing. McNa i r et a l 4 0 in a K in-Corn study looking at the effect of verbal encouragement on max imum voluntary musc le action of the e lbow flexors during isometric testing found that peak force increased when verbal encouragement w a s presented. The implications of verbal encouragement may be important when motivating subjects to give maximal effort during grip strength testing. Other instruments have been compared to the Jamar dynamometer for assess ing grip strength including the Balt imore Therapeut ic Equipment Work Simulator ( B T E ) 4 1 4 2 and the Modif ied Sphygmomanomete r 3 6 . Tools that utilize air compress ion in a c losed sys tem such as the Sphygmomanometer and the Martin vigorimeter have been crit icized as measur ing grip pressure and not grip fo rce . 3 0 However, in s o m e populat ions such as the elderly or those that have suffered a painful hand injury, these types of dev ices may be more appropriate than using a J a m a r hand dynamometer in order to obtain basel ine grip strength measures . RATIONALE FOR INVESTIGATION Reliability of Grip Strength Measures Reliabil ity refers to a measure that is free from errors of measurement 8 and is repeatable when administered on more than one occas ion . 9 Reliabil ity has severa l components that are often hard to separate including instrument reliability, intrarater 9 reliability, interrater reliability and intrasubject reliability. 9 Intrarater reliability, or the cons is tency with which one rater ass igns scores to a single set of responses on two occas ions measures error of the researcher . 9 Intrasubject reliability is assoc ia ted with actual changes in subject performance in repeated t r ia ls 9 . Most reliability studies or test-retest research reflect a combinat ion of instrument errors, tester errors and true subject variabil ity. 9 Instrument reliability for the Jamar dynamometer is tested through cal ibrat ion. Distributors of the J a m a r dynamometer calibrate the instrument in their labs and suggest that calibration should be done at least once a year. F e s s 4 3 s t ressed the need for compar ison of calibration with known standards such as the National Institute of S tandards and Techno logy (NIST). Whi le Jamar dynamometers have been in use s ince 1954, F e s s found that no one had checked them against NIST criteria for over three decades , assuming that calibration was accura te . 2 3 F e s s 4 4 , 2 3 , in a study on the reliability of 53 new and used Jamar dynamometers, found over 5 0 % of the used dynamometers and 2 3 . 8 % of the new dynamometers needed to be recalibrated accord ing to her method and recommendat ions for checking J a m a r dynamometer cal ibrat ion. 4 5 F e s s 4 5 states that the Jamar dynamometer is a highly reliable assessmen t instrument when it is correctly cal ibrated. High calibration accuracy has been reported for the Jamar dynamomete r . 4 6 4 7 F lood-Joy and M a t h i o w e t z 4 8 in a compar ison study of the measurements of three Jamar dynamometers , found that there appeared to be significant measurement dif ferences among the dynamometers . High inter-rater reliability has been repo r ted 2 0 , 4 4 4 9 for the standard J a m a r dynamometer . Hamil ton, Ba lnave and A d a m s 4 9 , in a test-retest study of grip strength in 10 healthy individuals using the Jamar dynamometer, found that the mean score of three trials w a s the most reliable (ICC=0.96 for the right hand and ICC=0.95 for the left, where ICC= intraclass correlation coefficient). MacDermid et a l 4 6 , in a interrater reliability study of patients with cumulat ive trauma disorders found the intraclass correlation coefficient w a s high (ICC> 0.87) for all strength measurements using either a s ingle trial or the mean of three repetitions. Mathiowetz et a l 2 0 showed that using the mean of three trials resulted in greater test-retest reliability compared to a s ingle trial or the best score of two trials (r=0.88 for the right hand and r=0.93 for the left for the mean of 3 trials). A l though all 5 handle posit ions on the Jamar dynamometer exhibited good reliability ( I C O 0 . 8 8 ) , Hamil ton, Ba lnave and A d a m s 1 1 found that posit ion 1 w a s significantly less reliable when compared with the other 4 handle posit ions on the J a m a r dynamometer (ICC=0.89 for the right hand at position 1 compared to 0.96 for posit ion 5 which exhibited the greatest test-retest reliability). Stratford et a l , 4 9 whi le using a different brand of hand dynamometer, conc luded that test-retest reliability w a s enhanced by multiple test sess ions as compared to increasing the number of repetit ions during a single test sess ion (R= 0.96 when taking three measurements per day over two days as opposed to R=0.90 when taking the est imate of a s ingle measurement) . A s indicated by Bear -Lehman and A b r e u 5 0 , hand assessmen t instrument reliability and validity studies are needed and it is imperative that the instruments be cal ibrated regularly and the s a m e instrument be used in pre and post-test measurements . Throughout the literature, there are a number of procedural di f ferences noted during grip strength testing including handle pos i t ion, 3 7 , 5 1 wrist and e lbow posi t ion 2 9 , number of 11 t r ia ls 3 8 , and the use of normative da tabases . 2 4 " 2 8 T h e s e factors make it very difficult to adequate ly compare the literature regarding the test-retest reliability of grip strength measures . The type of reliability statistics used depends on the type of reliability being measured and is often expressed as an intra-class correlation coefficient or a P e a r s o n product moment correlation when ratings are quantitative as in grip strength measurements . It is indicated in statistical theory that any measure will have a certain degree of reliability when appl ied to certain groups under certain condi t ions 4 9 . Of importance to cl inicians is the ability of the measure to reliably detect a clinically important change in order for the continued use of the measurement tool in the cl inical sett ing. Clinical Significance of Grip Strength Measures Most research involves testing a hypothesis through the appl icat ion of appropriate statistical tests to determine if there is a statistically significant difference between the measures of interest. However, a measure that is found to exhibit statistical s igni f icance may, or may not, exhibit cl inical s ignif icance. The clinical s igni f icance of a measure refers to the ability of that measure to dist inguish between important and unimportant change. For example, a 10 degree difference on a range of motion test of the knee may be found to be statistically significant, but, of interest in rehabilitation, is how signif icant that 10 degree difference is to the function of the patient or to the therapist 's evaluat ion and treatment of that patient. Therapists need to be able to determine when a meaningful change has occurred for their patients in order to explain per formance and predict outcomes. The definition of meaningful change may differ from patient population to population for a speci f ic measure . In the above illustration, for an athlete who requires full knee range of motion in order to return to competi t ive sport, 1 0 degrees of range of motion may be a meaningful change whereas another patient w h o s e activity level is less may not notice the difference in range of motion throughout their day to day activities. With all measures or assessmen t tools, it is important that therapists know what the normal variation in the measure of interest is, how reliable the measure is and how to interpret the results of the measure. Backed with this information, therapists can make informed clinical dec is ions regarding the evaluat ion and treatment of their patients. Minimal reference is given in the literature as to what constitutes a clinically important change with respect to grip strength measures . The absolute dif ference in grip strength scores between dominant and non-dominant hand is somet imes used to determine if a true w e a k n e s s exists. However, the Amer ican Medica l Assoc ia t ion 's G u i d e s to Evaluat ion of Permanent Impairment states that little ev idence exists for a signif icant difference between the dominant and non-dominant h a n d . 5 3 Most grip strength studies reviewed can be general ly c lassi f ied as 1 ) s tudies descr ib ing an instrument or compar ing means to a s s e s s grip strength; 2 ) s tudies descr ib ing the effects of posit ioning, environment and rest time on grip strength; 3 ) studies to determine maximum voluntary effort and the validity of per formance and 4) studies determining clinical norms for grip strength in certain ages , occupat ions or disabi l i t ies 4 2 . The latter group of studies involves efforts to develop s tandard ized data for grip strength in var ious populat ions. 13 Schmidt and T o e w s , 5 4 attempted to provide standardized grip strength norms using a large sample of employee appl icants at a manufacturing plant. However, it has been reported that descript ion of the methodology was insufficient for replication of this study and the handle of the dynamometer was slightly modified from convent ional dynamometers , therefore general izat ion from these results is l imited 2 5 . Kel lor et a l 5 5 in 1971 publ ished normative data for grip and pinch strength based on a samp le of 250 individuals partitioned in 3 large age groups. No standardized posit ioning or instructions were fol lowed and test-retest reliability was not reported, again making these norms limited in their usefulness. Mathiowetz et a l 2 5 in 1985 establ ished clinical norms for adults aged 20 to 75+ years on four tests of hand strength, including grip and pinch strength. T h e s e norms are most commonly cited in the literature. Math iowetz et a l 2 5 measured 628 volunteers with those in the 20 to 59 year age groups free from d i sease or injury that could affect upper extremity strength; inclusion criteria for subjects 60 years and above were outl ined. Mathiowetz et a l 2 5 used standardized posit ioning and instructions utilizing the Jamar dynamometer at handle posit ion number 2 to measure grip strength. The test scores for 3 success ive trials were recorded for each hand. Whi le not explicitly stated in the study, it was assumed that the mean of the three trials w a s used in the data analys is based on other studies by Math iowetz 2 0 . The average performance for men and women by age groups with respect to right/left hand and including the mean , standard deviation and standard error were repor ted. 2 5 The purpose of these norms was to provide a means of compar ing the score of individual patients to that of normal subjects of the s a m e age and sex. However, the interpretation of the scores and the magnitude of difference required to detect clinically important change were not explicitly addressed . 14 Class i ca l measurement theory a s s u m e s that every person has a true va lue for the measurement of interest (i.e. knee flexion) and that variations in a person 's scores are measurement errors about the true score . 8 In contrast, generalizabil i ty theory 8 recognizes that di f ferences in scores may be related to any number of different facets, or sources of variability. In our example of knee f lexion, facets of interest to the researcher might be the skill of the physical therapist in taking the measurement , the patient's level of relaxation or level of comfort with the examiner taking the measurement , and the accuracy of the goniometer. In physical therapy, the general izabi l i ty approach in the study of measurements acknowledges and provides a way to quantify the many sources of variability that cl inicians s e e in their patients from day to day 8 . Reliabil ity studies, using a generalizabil i ty theory approach, that report the standard error of measurement ( S E M ) and conf idence intervals (Cl) give a range of va lues for the reliability of the measure within the tested population. This is va luable to the cl inician as it provides a range of va lues for a particular measure and is more conduc ive to making dec is ions regarding clinical signif icance. For example , in our above example of knee range of motion, if we know that a measure of range of motion is reliable (high r value) and the 9 5 % conf idence interval about the S E M assoc ia ted with that measurement lies in the magnitude of 2 to 8 degrees, a 10 degree improvement could be interpreted as a clinically significant change in the patient's status. This information could be very helpful to the treating therapist. Nonethe less, upon review of the literature, there are no clear guidel ines as to what is deemed clinically signif icant change with respect to grip strength test measures . 15 CONCLUSION AND P U R P O S E OF R E S E A R C H In rehabilitation, grip strength testing is a common clinical assessmen t that is famil iar to many physical and occupat ional therapists. The literature reveals that there are a variety of different methods for measur ing grip strength including variat ions in measurement tools, the number of trials performed, posit ions of the e lbow and wrist during testing, the rest t ime in between trials, position of the Jamar dynamometer handle and the units of force used (pounds or k i lograms). 2 1 Gr ip strength testing protocols are often inadequately descr ibed in the literature and this apparent lack of s tandard ized terminology and protocols makes it very difficult to compare ac ross studies or to interpret results in the context of clinically important change. Prior to determining what constitutes a clinically important change, the tool or measure being used must demonstrate good reliability, 4 3 5 6 thus reliability studies are a prerequisite to the utilization of cl inical data. Measurements must be reliable if therapists are to be confident that changes in strength over t ime, or dif ferences in patient scores and normative va lues, are due to real dif ferences and not to measurement error. The purpose of this study was to survey therapists to determine what grip strength measurement strategies were currently being used to detect clinically important change in the rehabilitation setting. A s wel l , the reliability of two grip strength tests using the Jamar dynamometer was investigated in order to propose a measurement strategy to detect clinically important change. To our knowledge, no one has attempted to specif ical ly look at grip strength measures in this way. 16 R E F E R E N C E S 1. A s i m o w Engineer ing Company . Jamar Adjustable Hand Dynamometer . Los Ange les : A s i m o w Engineer ing. 2. F e s s E E , Moran C A . Clinical Assessment Recommendations. Garner , North Caro l ina : Amer ican Society of Hand Therapists 1981: 6-8. 3. Bohannon R W . Object ive measures . Physical Therapy 1989; 69(7): 590-593. 4. F e s s E E . The need for reliability and validity in hand assessmen t instruments. Journal of Hand Surgery 1986; 11A(5): 621-623. 5. Rothstein J M . Measurement in Physical Therapy: Clinics in Physical Therapy. New York, N Y : Churchi l l Livingston; 1985. 6. Buford W L . Cl in ical assessment , objectivity, and the ubiquitous laws of instrumentation. Journal of Hand Therapy 1995; 8(2): 149-154. 7. Y o u n g V L , Pin P, Kraemer BA , Gou ld R B , Nemergut L, Pel lowski M. Fluctuation in grip and pinch strength among normal subjects. Journal of Hand Surgery 1989; 14A: 125-9. 8. Domholdt E. Physical therapy research: principles and applications. Toronto: W . B . Saunders ; 1993. 9. C o l e B, F inch F, Gowland C , Mayo N. Physical Rehabilitation Outcome Measures. Toronto, Ont: Canad ian Physiotherapy Assoc ia t ion 1994. 10. Amer i can Phys ica l Therapy Assoc ia t ion 's Task Force on Standards for Measurement in Phys ica l Therapy. Standards for tests and measurements in physical therapy practice. Physical Therapy 1991; 71(8): 589-622. 11. Hamilton A , Ba lnave R, A d a m s R. Grip strength testing reliability. Journal of Hand Therapy 1994; 7(3): 163-170. 12. Matheson L N . Work Capaci ty Evaluat ion: Systemat ic Approach to Industrial Rehabi l i tat ion. Anahe im , C A : Employment and Rehabil i tat ion Institute of Cal i fornia; 1991. 13. Hislop H J . The not-so- impossible dream. Tenth Mary McMi l lan Lecture. Physical Therapy 1975; 55(10): 1069-1080. 17 14. Smidt G L . Walk ing the trail of physical therapy research. Physical Therapy 1986; 66(3):375-378. 15. Beas ley W F . Influence of method on est imates of normal knee extensor force among normal and postpolio chi ldren. Physical Therapy 1956; 36:21-41. 16. Bohannon R W . Manua l musc le test scores and dynamometer test scores of knee extension strength. Archives of Physical Medicine and Rehabilitation 1985; 67: 390-392. 17. Stratford P W , Balsor B E . A compar ison of make and break tests using a hand-held dynamometer and the Kin-Corn. Journal of Orthopaedic and Sports Physical Therapy 1994; 19(1):28-32. 18. Bechtol C O . Grip Test: The use of a dynamometer with adjustible handle spac ings . Journal of Bone and Joint Surgery (Am.) 1954; 36A: 832. 19. Kirkpatrick J E . Evaluat ion of grip loss. California Medicine 1956; 85:314-320. 20. Mathiowetz V , W e b e r K, Vo l land G , K a s h m a n N. Reliabil ity and validity of hand strength evaluat ions. Journal of Hand Surgery (Am) 1984; 9:222-226. 21 . Smith R O , Benge M W . P inch and grasp strength: Standardizat ion of terminology and protocol. American Journal of Occupational Therapy 1985; 39(8):531-535. 22 . Go ldman S , Caha lan T D , A n K. The injured upper extremity and the J a m a r five -handle posit ion test. American Journal of Physical Medicine and Rehabilitation 1991; 70(6): 306-308. 23. F e s s E E . Guide l ines for evaluating assessmen t instruments. Journal of Hand Therapy 1995; 8(2): 144-148. 24. Kel lor M, Frost J , Si lberberg N, Iversen I, Cummings R. Hand strength and dexterity. American Journal of Occupational Therapy 1971; 25:77-83. 25. Mathiowetz V , K a s h m a n N, Vol land G , W e b e r K, Dowe M, Rogers S . Gr ip and pinch strength: Normative data for adults. Archives of Physical Medicine and Rehabilitation 1985; 66:69-74. 26. T o e w s J V . A grip strength study among steelworkers. Archives of Physical Medicine and Rehabilitation 1964; 45:413-417. 27. Mathiowetz V , Wiemer D M , Federman S M . Grip and pinch strength: norms for 6-to 19-years-olds. American Journal of Occupational Therapy 1986; 40(10):705-11. 28. A g e r C L , Olivett BL , Johnson C L . Grasp and pinch strength in chi ldren 5 to 12 years old. American Journal of Occupational Therapy 1984; 38(2): 107-113. 18 29. S u C Y , Lin J H , Ch ien T H , Cheng K F , Sung Y T . Grip strength in different posit ions of e lbow and shoulder. Archives of Physical Medicine & Rehabilitation 1994; 75(7):812-5. 30. Desros iers J , Bravo G , Hebert R, Dutil E. Normative data for grip strength of elderly men and women . American Journal of Occupational Therapy 1995; 49(7):637-44. 31 . Harkonen R, Pi i r tomaa M and Alaranta H. Grip strength and hand posit ion of the dynamometer in 204 Finnish adults. Journal of Hand Surgery 1993; 18B(1): 129-132. 32. Pe te rsen P, Petrick M, Connor H, Conkl in D. Grip strength and hand dominance: chal lenging the 10% rule. American Journal of Occupational Therapy 1989; 43(7):444-7. 33. C rosby C A , W e h b e M A , Mawr B. Hand strength: normative va lues. Journal of Hand Surgery 1994; 19(4):665-70. 34. Josty IC, Tyler M P , Shewel l P C , Roberts A H . Grip and pinch strength variat ions in different types of workers. Journal of Hand Surgery (British Vo lume) 1997; 22(2): 266-9. 35. O'Dr iscol l S W , Horii E, N e s s R, Caha lan T D , Richards R R , A n K N . The relationship between wrist posit ion, grasp s ize and grip strength. Journal of Hand Surgery (Am) 1992; 17(1): 169-77. 36. Lusardi M M , Bohannon R W . Hand grip strength: Comparabi l i ty of measurements obtained with a Jamar dynamometer and a modified sphygmomanometer . Journal of Hand Therapy 1991; 4(3): 117-122. 37. F e s s E E . The effects of Jamar handle position and test protocol on normal grip strength. Journal of Hand Surgery 1982; 7A:308. 38. Mathiowetz, V . Effects of three trials on grip and pinch strength measurements . Journal of Hand Therapy 1990; 3(4):195-198. 39. Mar ion R, Niebuhr B R . Effect of warm-up prior to maximal grip contract ions. Journal of Hand Therapy 1992; 5(3):143-146. 40 . McNa i r P J , Dep ledge J , Brettkelly M, Stanley S N . Verba l encouragement : effects on max imum effort voluntary muscle action. British Journal of Sports Medicine 1996; 30(3):243-5. 19 41 . Beaton D E , O'Dr iscol l S W , Richards R R . Grip strength testing using the B T E work simulator and the Jamar dynamometer: A comparat ive study. Journal of Hand Surgery 1995; 20A(2):293-298. 42 . King J W , Berryhill B H . A compar ison of two static grip testing methods and its cl inical appl icat ions: A preliminary study. Journal of Hand Therapy 1988; 4 :204-208. 43 . F e s s E E . W h y trial-to-trial reliability is not enough. Journal of Hand Therapy 1994; 7(1):28. 44. F e s s E E . Reliabil ity of new and used Jamar dynamometers under laboratory condit ions. Comment . Journal of Hand Therapy 1990; 1:35. 45 . F e s s E E . A method for checking Jamar dynamometer calibration. Journal of Hand Therapy 1987; (4):28-32. 46. MacDermid J C , Kramer J F , Woodbury M G , McFar lane R M , Roth J H . Interrater reliability of pinch and grip strength measurements in cl ients with cumulat ive t rauma disorders. Journal of Hand Therapy 1994; 7: 10-14. 47. Harkonen R, Harju R, Alaranta H. Accu racy of the J a m a r dynamometer . Journal of Hand Therapy 1993; 6:259-262. 48. F lood-Joy , M, Mathiowetz,V. Grip-strength measurement : A compar ison of three J a m a r dynamometers . Occupational Therapy Journal of Research 1987; 7:4: 235 -243 . 49. Stratford P W , Normal G R , Mcintosh J M . General izabi l i ty of grip strength measures in patients with tennis elbow. Physical Therapy 1989; 69 :276-281 . 50. Bea r -Lehman J , Ab reau B C . Evaluat ing the hand: Issues in reliability and validity. Physical Therapy 1989; 69(12): 1025-1033. 51 . Niebuhr B R , Mar ion R. Detecting sincerity of effort when measur ing grip strength. American Journal of Physical Medicine and Rehabilitation 1987; 66:16-24. 52. Hopk ins K D , Hopkins B R , G l a s s G V . Bas i c Statist ics for the Behavioura l S c i e n c e s , 3 r d ed . Toronto: Al lyn and Bacon , 1996. 53. Amer i can Medica l Assoc ia t ion . Guides to the evaluation of permanent impairment, 4th ed . Ch icago , Illinois: Amer ican Medica l Assoc ia t ion , 1993. 54. Schmidt RT , T o e w s J V . Grip strength as measured by the J a m a r dynamometer . Archives of Physical Medicine & Rehabilitation 1970; 51:321-327. 20 55. Kel lor M, Frost J , Si lberberg N, Iverson I, Cummings R. Hand strength and dexterity. American Journal of Occupational Therapy 1971: 25:77-83. 56. F le iss J L . The Design and Ana lys is of Cl inical Exper iments, New York: John Wi ley and S o n s , 1986. 21 CHAPTER 2_ SURVEY STUDY INTRODUCTION W h e n reviewing the literature on grip strength testing, it becomes evident that whi le grip strength measures are commonly used in clinical practice, the protocols and terminology for testing are not standardized and differ widely from study to study and clinic to clinic. Posi t ioning, number of repetitions, equipment used , testing protocols, intertrial rest per iods, equipment calibration and interpretation of measurement scores all exhibit cons iderab le variation. This makes it extremely difficult to compare ac ross studies and to compare grip strength measures to normative da tabases that may not have used the s a m e testing procedures or instruments. The purpose of this study was to survey occupat ional therapists and physiotherapists who employ grip strength testing during their normal cl inical evaluat ion and a s s e s s m e n t s of patients in order to gain some insight into what measu res are currently being used . It was hypothesized that 1) most cl inicians employ ing grip strength tests will use the Standard Grip Strength Test ( S G S T ) and the 5-Posit ion Grip Strength Test (5PGST) (Appendix A) utilizing the standard Jamar dynamometer as a measurement tool; 2) that the main purpose of testing will be to measure grip strength in individuals with hand and upper extremity dysfunction and 3) that there will be a wide 22 var iance in what cl inicians deem a clinically important change with respect to analyz ing grip strength test results. The object ives of the study were as fol lows: 1. To determine which grip strength tests and which testing instruments are commonly used in cl inical practice. 2. To determine which patient populat ions are being measured and what is the purpose of the testing. 3. To determine if normative da tabases are used for compar ison of test results. 4. To investigate what parameters cl inicians use to determine clinically important change in grip strength measures . METHODS Subjects A self-report measure (Appendix B) was mailed to 140 sites ac ross C a n a d a . Faci l i t ies were chosen from amongst a larger sample of Canad ian University physical therapy and occupat ional therapy student placement information. Dec is ion to include a facility in the study was based on a descript ion of the treatment serv ices the facil it ies offered; inclusion criteria included hand rehabilitation and or thopaedics. A s wel l , 13 members of the Canad ian Society of Hand Therapists unaffiliated with the sites were sent a survey quest ionnaire. In total, 153 facilities (sites and therapists combined) were included in the test sample . Prior to mail ing, the survey was pilot tested at 5 different si tes in Vancouver , British Co lumb ia and modif ications were made with respect to format and content 23 based on their feedback. The final survey contained 6 closed-format mult iple-choice items with s o m e flexibility of response avai lable by including an "other" category. There were a lso 2 forced choice quest ions with a yes/no answer. A one-page format w a s chosen in order to increase compl iance and e a s e of filling out of the form. The University of British Co lumb ia Ethics Commit tee approved this research (#C97-0223). Testing Procedures A cover letter (Appendix C) was sent with each quest ionnaire outlining the study and providing contact information if there were quest ions. A s tamped, se l f -addressed enve lope w a s provided in the survey package and respondents were asked to return their completed surveys by mail or by fax. Data Collection Quest ionnai res were coded with a number that cor responded to the facility or individual on a master mailing list. R e s p o n s e s were treated as being anonymous and confidential . A cutoff date for survey return was establ ished once returns appeared to diminish and no further surveys were included in the study after this date. Data Analysis Survey responses were ana lyzed for f requency distributions. The percentage of subjects who responded to each category under a relevant survey quest ion w a s calcu lated. In the case of mult iple-choice quest ions where respondents checked off more than one choice, each choice was treated separately in the data analys is to avoid biasing the results. Where there were multiple surveys completed per site, one survey was randomly selected for inclusion in the data set. However, in analyz ing the data from quest ion number seven ("How large would the difference have to be for you to call it clinically significant'), all data (including multiple surveys from 1 site) were cons idered 24 b e c a u s e a proportion of the respondents to quest ion 7 did not indicate the units of measure and therefore, those responses could not be used. The qualitative data and written comments generated by the survey were reviewed and reported in the text of the results and the d iscuss ion. R E S U L T S O n e hundred and sixty six surveys in total were completed and returned including multiple responses from test sites. Three responses were not used due to late return. Of the 153 facilities contacted (140 sites, 13 hand therapists), 120 responses were obtained resulting in a 7 8 % response rate overal l . Breaking this down into subgroups, 7 8 % of the sites contacted responded (109/140) and 8 5 % (11/13) of the hand therapists responded. S e v e n respondents indicated they did not use grip strength measures in cl inical practice and their surveys were removed from further data analys is , revising the sample s ize from 120 to 113. Ninety four percent of all respondents (113 of 120 responses) indicated that they used grip strength measu res in their cl inical practice. Of the tests used, 8 3 % indicated that they used the Standard Grip Strength Test and 4 2 % stated they used the 5-Posit ion Grip Strength Test (Figure 1); 3 0 % used both tests. Twenty-f ive percent of respondents used the Rap id E x c h a n g e Gr ip Strength Test and 1 1 % used other grip strength measures . E ighty-seven percent used the J a m a r dynamometer to measure grip strength while 1 5 % used a blood pressure cuff (Figure 2). Other tools used to measure grip strength included the digital J a m a r dynamometer (10%) and the Balt imore Therapeut ic Equipment Work Simulator (BTE) (9%) (Figure 2). Respondents used grip strength testing for a variety of patient 0 to c o a <o o on c o L . CL 100 90 80 70 60 50 40 30 20 10 0 Survey Response - Type of Grip Strength Test Used S t a n d a r d Gr ip St rength Tes t • Al l R e s p o n s e s (n=113) • Hand Therapists Only (n=11) I • Si tes Only (n=102) 5 Pos i t ion Gr ip St rength Tes t R a p i d E x c h a n g e Gr ip Tes t O the r Grip Strength Test Figure 1: Survey responses indicating the types of grip strength used. Survey Response - Measurement Instrument Used 0) co c o O . CO 0) Q_ +•> c CJ o o 100 90 80 70 60 50 40 30 20 10 0 J a m a r n n Digital Jamar • Al l R e s p o n s e s (n=113) • Hand Therapists Only (n=11) • Si tes Only (n=102) B T E Blood Pressure Cuff Other Grip Strength Measurement Tool Figure 2: Survey responses of grip strength measurement tools used in cl inical practice. 26 Survey Responses - Grip Strength Testing by Patient Population o w c o a. in G) al * J c a u o CL 100 90 80 70 60 50 40 30 20 10 0 • All R e s p o n s e s (n=113) • H a n d The rap i s t s On ly (n=11) • S i t es On ly (n=102) Hand Injuries Upper Extremity Back /Neck Injuries Injuries Patient Population Figure 3: Grip strength testing by patient populat ion. Other Survey Responses - Test Purpose V) C o Q. V) o ai •*-> c o o 1_ Q. • All Responses (n=113) • Hand Therapists Only (n=11) • Sites Only (n=102) Determine Grip Strength Measure Measure A s s e s s function impairment sincerity of effort Test Purpose Figure 4: Survey response depicting the intended purpose of grip strength measurements . 2 7 Survey Response - Use of a Normative Database Q_ o CL O CO c o Q . CO CU c 0 o 100 90 80 70 60 50 40 30 20 10 • No • Y e s Al l R e s p o n s e s (n=111) Hand Therapists Only (n=9) Si tes Only (n=102) Use of a Normative Database Figure 5: Survey response data depicting the use of a normative da tabase in grip strength measurements . populat ions including 8 5 % for persons with hand injuries, 7 3 % for upper extremity injuries, 3 1 % for back and neck injuries and 3 2 % for other condit ions including arthritis and cerebrovascu lar accidents (CVA 's ) (Figure 3). W h e n asked what purpose they used grip strength testing for, the respondents indicated that 9 8 % measured grip strength itself, 3 5 % used grip strength testing to measure function, 5 8 % to measure impairment and 4 2 % to measure sincerity of effort (Figure 4). Descript ive comments indicated the Standard Grip Strength Test w a s the best test to determine grip strength and the 5-Posit ion Grip Strength Test w a s indicated as the best test for sincerity of effort. Thirty-nine percent of respondents stated they used a normative da tabase with which to compare results versus 6 2 % who did not; Math iowetz et a l 's 1 norms for grip and pinch strength were cited as the most common da tabase used (Figure 5). 28 Clinically Important Change O n a number of measures reported to determine clinically important change, 5 8 % of the respondents stated they used a percentage difference between hands, 6 0 % stated they used the absolute difference between hands, 2 7 % used the percentage difference between hands, 3 9 % used the absolute difference between trials, 1 6 % used the percentage difference between a normative da tabase and the raw score and 1 9 % stated they used the absolute difference between a normative da tabase and the raw score (Figure 6 ) . Other measures reported to detect clinically important change included using coefficients of variation. The actual amount of differrence in grip strength that cl inicians v iewed as a cl inically important change varied across cl inicians. Of those reporting the use of an absolute difference between hands to detect important change (n=53), 5 7 % indicated a difference of 5 kgforce was significant (Figure 7). Survey Response - Measures Used to Detect Clinical ly Important C h a n g e El Al l R e s p o n s e s (n=113) H a n d Therap is ts On ly (n=11) • S i tes Only (n=102) % diff btwn Abs diff % diff btwn Abs diff % diff btwn Abs diff hands btwn hands trials btwn trials NDB/score btwn NDB/score Other Measures Figure 6: Survey responses depicting measures used to detect clinically important change in grip strength measurement . Survey Response - Absolute Difference Between Hands (n=53) 100-, 2 kgforce 5 kgforce 10 kgforce 15 kgforce 20 kgforce 27 kgforce Kgforce deemed clinically significant Figure 7: Survey response indicating amount of change in kgforce to detect clinically significant change in grip strength measurements between hands. Survey Response - Percentage Difference Between Hands (n 100-, 80-0) rcent pons 60-0) W 40-Q_ o 20-0 M a c c C c C C c c c c C CD CD CD CD CD CD CD CD CD CD CD O O O O O O O O O O O L i_ \ \ i_ L— L— i_ CD CD CD CD CD CD CD CD CD CD CD CL- Q_ Q_ 0_ Q_ Q_ CL- CL CL CL CL IO O IT) O LO O IO O LO O O T— T— CM CN CO CO LO 00 Percent Difference Deemed Clinically Significant Figure 8: Survey response data indicating percentage difference between hands considered clinically significant with grip strength measurements . 30 Survey Response - Percentage Difference Between Trials (n=34) 100-, CD 5 Percent 10 Percent 15 Percent 20 Percent 25 Percent Percentage Reported Clinically Significant Figure 9: Survey responses depicting clinically significant percentage change between trials reported in grip strength measures . Survey Response - Absolute Difference Between Trials for Clinical Significance (n=23) 2kgf 5kgf 10kgf 15KGF Kgforce Reported Clinically Significant Figure 10: Survey responses depicting clinically significant change in kgforce between trials in grip strength measurements. Survey Response - Percentage Difference Between Normative Database (n=17) 100n 0 ) « 80-o w 60-a> u a> a. 5 Percent 10 15 20 25 30 35 40 60 Percent Percent Percent Percent Percent Percent Percent Percent Percentage Difference Reported Clinically Significant Figure 11: Survey response data indicating percentage dif ference between grip strength scores and a normative da tabase to be cons idered clinically significant. Survey Response - Actual Difference Between a Normative Database and Grip Strength Scores (n=9) 2 kgforce 5 kgforce 10 kgforce 15 kgforce 20 kgforce Kgforce Reported Clinically Significant Figure 1 2 : Survey response data indicating absolute difference between a normative da tabase and grip strength measurements for cl inical s igni f icance. 32 M e a s u r e s ranged from 2kgforce (15%) to 27 kgforce (2%). Of those using a percentage difference between hands (n=82), the range of va lues was between 5 % to 8 0 % . Thirty percent of respondents stated that a 1 0 % difference between hands would be cons idered significant and 2 4 % stated a 2 0 % difference would be signif icant (Figure 8). W h e n looking at between trial s igni f icance (n=34), 5 3 % of respondents indicated that a 1 0 % difference between trials was an important change (Figure 9). The absolute difference between trials (n=23) was reportedly significant at 2 kgforce (52%) and at 5 kgforce (43%) (Figure 10). Utilizing a normative database, 2 9 % (n=17) of respondents indicated that a 1 0 % difference between the score and the normative va lue w a s signif icant (Figure 11) while 2 4 % stated that a 2 0 % difference was significant. Sixty-seven percent (n=9) stated a difference of 5kgforce from the normative value was signif icant (Figure 12). Cl inical s igni f icance was also indicated by 1) using categor ies such as excel lent, average, below average etc. where below average and poor were clinically significant; 2) using consistent improvement over t ime; 3) using functional level; 4) using any change in the patient's status and 5) using the patient's self-report of function. It was a lso reported that there were too many var iables to be able to determine cl inical s igni f icance. DISCUSSION The high overal l response to the survey may indicate that grip strength measurement is an important and common clinical practice in physical therapy and occupat ional therapy. Marcuzz i , Kelly, Chang and Hannah 2 , in a survey of Canad ian hand therapists, found that evaluation of grip and pinch strength was rated one of the 33 most important cl inical activities performed by the respondents. In our survey, 8 7 % used the J a m a r dynamometer , 8 3 % used the Standard Grip Strength Test and 9 8 % indicated that the main purpose of grip strength testing was to measure grip strength. However , while cl inician use of the grip strength tools and measures was reportedly very high in our survey as noted above, there was ev idence of a range of responses indicating that s o m e variability exists in clinical practice as to equipment used , type of grip strength testing performed, patient populat ions tested and the purpose of each grip strength test. Of considerable interest was the range of interpretation of measurement scores (clinical signif icance) reported in clinical practice. T h e s e variat ions in survey responses could be attributed to a number of factors. With respect to measurement tools, the Jamar dynamometer emerged as the most frequently used grip strength measur ing tool. E ighty-seven percent reported using the J a m a r dynamometer in clinical practice while 3 1 % of all respondents indicated using more than one measurement tool. This cor responds to Smith and B e n g e ' s 3 survey of 195 occupat ional therapy cl inics where 7 9 % stated that the J a m a r dynamometer was the most commonly used measurement tool with 4 0 % of respondents using more than one tool. The Jamar dynamometer appears widely accepted as a grip strength measur ing dev ice in the literature 4 , 5 and the results of our survey support this statement in clinical practice. Compared to other instruments, it w a s the measurement tool of choice for grip strength amongst the hand therapists surveyed in our study. W h e n reviewing the survey data, it was evident that the use of other measurement tools reflected both the patient populat ions being tested and the availabil ity of measurement tools in the clinical setting. For example , respondents in 34 this survey working with elderly or arthritic populat ions stated that they used a blood pressure cuff to measure grip strength. The Jamar dynamometer may be too difficult to hold or requires too much force to register a value for this group of patients. A measurement tool has to "fit" the abilities of the patients and be appropriate for the type of populat ion tested in order to gain meaningful measurements . A s wel l , the affordability of certain tools often dictates their use in the cl inical setting. A J a m a r dynamometer is relatively inexpensive compared to more sophist icated and costly testing dev ices such as a B T E . The most common grip strength tests used were the Standard Gr ip Strength Test (83%) and the 5-Posit ion Grip Strength (42%) test as initially hypothes ized. However , a review of the comments noted in the survey responses suggested there w a s s o m e confusion as to the protocol for the Standard Grip Strength Test . Us ing one trial, the mean of three trials, or the highest score on three trials as the measure of grip strength were noted. Th is indicates the need for accurately defining test parameters and testing protocols. A survey looking at grip and pinch strength terminology and protocols sugges ted that "a significant number of respondents indicated they had no particular reason for using the speci f ic terminology, equipment or protocols that they used " in grip and pinch strength test ing. 3 Ninety-eight percent of respondents indicated the purpose of grip strength testing w a s to determine grip strength. Survey results a lso suggested that grip strength testing w a s used to measure function (35%) and impairment (60%). Hand function is a difficult construct to define as it can pertain to a variety of condit ions such as the ability to perform bas ic activities of daily living or the ability to perform a complex work task. There is no c lear consensus as to what is meant by the term "function" within physical 35 therapy but Roths te in 6 attempts to define function as the interaction of the individual with his or her environment. Grip strength testing itself involves the measurement of force during static (isometric) musc le contractions, in which virtually no joint movement takes p lace . 7 A s suggested by Rothste in 's 6 definition of function, hand function would relate to movement patterns of the hand during a task. Whi le grip strength may be one component of hand function, there may be other s tandardized tests that are better suited and more valid est imators of hand function in each speci f ic patient situation. Accord ing to the Wor ld Health Organizat ion's International Classi f icat ion for Impairment Disabil ity and Hand icap 8 , as quoted in Phys ica l Rehabil i tat ion Ou tcome M e a s u r e s 9 , impairment refers to any loss or abnormality of psychologica l , physiological or anatomical structure or funct ion. 8 In physical therapy, impairment of the musculoskeleta l sys tem directly impacts funct ion 6 , and measures of impairment (often expressed as musc le strength or joint range of motion 9), are presumed to reflect functional ability. Thus grip strength is considered a measure of impairment in physical therapy pract ice although the Amer ican Medica l Assoc ia t ion 's Gu ides to the Evaluat ion of Permanent Impairment 1 0 does not p lace great emphas is on grip strength measurements when assess ing impairment. In select ing a measure to a s s e s s function or impairment, therapists must consider the purpose for which the instrument is being sought, the theoretical constructs surrounding the measure, the measurement d imens ion and the mode of administration required in order to select the appropriate test instrument. Whi le the majority of the survey responses reported persons with hand and upper extremity injuries as the patient population most often tested for grip strength (85 and 7 3 % respectively), 3 1 % indicated using grip strength measures for persons with 36 back and neck injuries. Forty-two percent of respondents indicated they used grip strength measures for sincerity of effort and comments indicated that the 5 Posi t ion Gr ip Strength Test and the Rap id Exchange Grip Strength Test were the tests most often used for this purpose. Whi le the issue of sincerity of effort is beyond the s c o p e of this study, the purpose of grip strength tests must be v iewed cautiously. The concept of what is actually being measured must be kept in mind and cl inicians must be careful not to use tests and extrapolate results for something other than what the tests are purported to measure . Clinical Significance Cl in ical s igni f icance was not clearly def ined in the survey due to omiss ion but the intent w a s to look at measurement va lues that cl inicians use to document change on which to base treatment dec is ions. Survey results indicated that the measures used to document clinically significant change and the interpretation of measurement scores exhibited considerable range and variation. For example, the difference between hands w a s the most common index reported to detect clinically important change in grip strength (58% percentage difference, 6 0 % absolute difference). Within these measures , the range of va lues extended from 5 % to 8 0 % and 2kgforce to 27 kgforce dif ference. Over half of the respondents (57%) indicated a 5kgforce was signif icant while 3 0 % indicated a 10% difference. However, Young et a l 1 1 , in a study of normal individuals using a J a m a r dynamometer to measure grip strength, found that mean grip strength f luctuated between 5.1 and 8.4 kgforce or between 19.2 to 23 .7%. If this information is appl ied to the survey responses, it appears that a majority of the 37 therapists are suggest ing clinically important change when the change may be due to normal subject variability a lone (Figures 7-12). The difference between trials was used less frequently than compar isons between hands (27% for percentage difference and 3 9 % for absolute difference). The range of responses for percentage difference included 5 % to 2 5 % with 1 0 % difference the most frequent response (53%). Of the absolute difference, the range extended from 2kgforce to 15kgforce with 5 7 % respondents indicating 5kgforce w a s clinically signif icant between trials. The problem foreseen with using a compar ison between trials is that the examiner has little information to know that the patient has given max imum effort on success ive trials. A difference perceived as clinically signif icant may in fact be due to motivation, subject variability from day to day, and/or measurement error. The use of a normative da tabase to compare results was used by 3 9 % of the respondents. R e s p o n s e s ranged from a 10% difference between a grip strength score and the normative da tabase to a 6 0 % difference with the most frequent responses being a 1 0 % (29%) and 2 0 % (24%) difference. A n absolute difference between a normative da tabase and grip strength score ranged from 2kgforce to 20kgforce with 6 7 % of respondents indicating a 5kgforce was clinically significant. It is important that the therapist be aware of the methodology used to develop the normative da tabase and to understand whether the therapist 's testing methodology, patient demograph ics and instrument are congruent with those recommended for the use of the da tabase . 1 Otherwise, the validity of the measure of clinically significant change when compared to a normative da tabase is quest ionable. 38 Hand therapists surveyed exhibited some variation in their responses to clinically signif icant change with 7 3 % indicating that they used the percentage dif ference between hands, 4 5 % the absolute difference between hands and 2 7 % the percentage dif ference between trials to report clinical s ignif icance. A s wel l , 5 6 % of hand therapists reported the use of a normative database compared to 3 9 % of the other respondents. It is noted though that the sample s ize of hand therapists w a s very smal l (n=11) compared to the other respondents (n=102). W e may have expected more c o n s e n s u s in their response to the quest ion of clinical s igni f icance with respect to grip strength scores due to the fact that hand therapists have had advanced training in hand therapy and assessmen t , and are deemed experts in this clinical area. However, they appeared to mimic the general sample population. The variation regarding cl inical s igni f icance throughout the survey responses indicates that there is not a consensus on what is clinically significant. The problem lies in the fact that we do not accurately know what normal variability is with grip strength measurements . Overa l l , the survey appeared to represent the area of interest (grip strength testing) for cl inicians targeted in our survey (general orthopaedic practice and hand therapists). Accord ing to our survey, measurement of grip strength is a common clinical pract ice but cons is tency among therapists with respect to measurement strategies is variable. Th is has implications not only when different therapists treat a patient in one setting but when a patient is transferred to another clinic where a different method of grip strength interpretation could be used. Whi le this makes it difficult to compare results, it could a lso have far reaching effects as grip strength testing is often used to determine further rehabilitation, return to work issues or disability in medical - legal 39 c a s e s . In these situations, widely accepted reliable and valid measurements are imperative to predict accurate outcomes. Limitations of survey Whi le overall compl iance in complet ing and returning the survey w a s excel lent, there were many inherent limitations in its des ign. The advantage to the survey method chosen w a s that it was efficient, responses remained anonymous, it was unlimited in terms of the geographic distribution of the respondents and it a l lowed the schedul ing of the complet ion of the quest ionnaire at the respondent 's conven ience . 1 2 The d isadvantages were that the depth of response was limited and the ability to clarify quest ions w a s not poss ib le . 1 2 A s well , c losed format quest ionnaires tended to restrict the range of possib le cho ices and responses. The purpose of the survey was to gain insight into what grip strength measures were currently being used during the normal cl inical evaluat ion and treatment of patients. The survey quest ions in some instances were not clearly def ined, leading to ambiguity in responses and problems with data retrieval in spite of the pilot survey. Quest ions could have been rephrased in order to more accurately gain the desi red information. Instructions for complet ing the survey needed to be more explicit. Chang ing the survey format and more clearly defining the quest ions would greatly aid in information gathering in the future. The survey did not target a certain patient population or age range, resulting in a wide variety of responses which may have been beneficial overall to gain a larger v iew of cl inical use of grip strength measures . The survey sample targeted for the study may have been b iased towards those cl inicians interested in hand rehabilitation, however it 40 can be argued that these cl inicians are the ones using the measurement tool. Of surveys that were returned indicating the respondent did not use grip strength measu res (n=7), it w a s because of the type of population they were treating and grip strength measurements were not relevant to their client base . Further research The information col lected from this survey enab les us to make the following recommendat ions for further research: 1. From our research, the use of different grip strength tests w a s determined. However, grip strength testing protocols need to be clearly def ined and standardized in order to improve consistent testing between and within therapists. This should include posit ioning, testing protocols, number of trials, intertrial rest periods and equipment cal ibration. Therapists need to be aware of and use these protocols in grip strength measurement . Further research is required to determine what protocols are being used and what is the most reliable. 2. The current methods of determining cl inical s igni f icance vary. In order to ensure clinical s igni f icance with grip strength measures , normal variability must first be establ ished. Then the varied validity of each form of assess ing clinical s igni f icance needs to be studied. 3. Normative da tabases currently in use in cl inical practice need to be reviewed. W h e n normative da tabases are used for patient compar ison , the use of the exact protocols that were reported in obtaining the normative data must be fol lowed. Further research is needed to determine if cl inicians are using normative da tabases properly. CONCLUSION This survey attempted to obtain information regarding grip strength testing during the normal cl inical evaluation and assessmen t of patients and specif ical ly, to investigate what parameters cl inicians use to determine clinically important change. To 41 our knowledge, no one has attempted to look at grip strength measurements in this way. The results of this study indicated that the Jamar dynamometer is commonly used to measure grip strength using the Standard Grip Strength Test and the 5-Posi t ion Grip Strength Test . Grip strength testing is used predominantly for upper extremity injuries. A normative da tabase for compar ison of test results is not consistently used in cl inical pract ice. Of interest in this study is the considerable variation in what c l in ic ians d e e m a clinically important change. 42 R E F E R E N C E S 1. Mathiowetz V , K a s h m a n N, Vol land G , W e b e r K, Dowe M, Rogers S . Gr ip and pinch strength:Normative data for adults. Arch ives of Phys ica l Medic ine and Rehabil i tat ion 1985; 66:69-74. 2. Marcuzz i A , Kel ly L, C h a n g M, Hannah, S . A survey of Canad ian hand therapists: Demograph ics , roles and educat ional needs. Journal of Hand Therapy 1998; 11:39-44. 3. Smith R O , Benge M W . P inch and grasp strength: Standardizat ion of terminology and protocol. Amer ican Journal of Occupat ional Therapy 1985; 39(8):531-535. 4. F e s s E E , Moran C A . Cl in ical A s s e s s m e n t Recommendat ions . Garner , North Caro l ina : Amer ican Society of Hand Therapists 1981: 6-8. 5. Mathiowetz V , W e b e r K, Vo l land G , K a s h m a n N. Reliability and validity of hand strength evaluat ions. Journal of Hand Surgery (Am) 1984; 9:222-226. 6. Rothstein J M . Measurement in physical therapy. New York: Churchi l l L iv ingstone; 1985. 7. Bohannon R W . The clinical measurement of strength. Cl in ical Rehabi l i tat ion 1987; 1:5-16. 8. Wor ld Health Organizat ion (WHO). International classif icat ion of impairments, disabil i t ies and handicaps. A manual of classif icat ion relating to the c o n s e q u e n c e s of d i sease . G e n e v a : World Health Organization 1980. 9. C o l e B, F inch F, Gowland C , Mayo N. Physical Rehabilitation Outcome Measures. Toronto, Ont: Canad ian Phys ica l therapy Assoc ia t ion 1994. 10. Amer i can Medica l Assoc ia t ion . Guides to the evaluation of permanent impairment, 4th ed . Ch icago , Illinois: Amer ican Medica l Assoc ia t ion , 1993. 11. Y o u n g V L , Pin P, Kraemer B A , Gou ld R B , Nemergut L, Pel lowski M. Fluctuat ion in grip and pinch strength among normal subjects. Journal of Hand Surgery 1989; 14A: 125-9. 12. Domholdt E. Phys ica l therapy research: principles and appl icat ions. Toronto: W . B . Saunders ; 1993. 43 CHAPTER 3_ TEST-RETEST RELIABILITY STUDY INTRODUCTION In rehabilitation, an important function of any measurement tool is to provide the therapist with relevant information concerning patient status and progress. To ach ieve this, a measurement tool must be able to measure accurately, diminish subject ive error and al low conc lus ions that are minimally affected by extraneous factors. 1 Reliabil ity def ines the tool 's ability to measure consistently and predictably. 1 A s F e s s 2 indicated, evaluat ion instruments that are neither reliable or accurate are not appropriate in our current world of ever- increasing demands for professional accountabil i ty. In physical therapy research, the generalizabil i ty approach in the study of measurements acknowledges and provides a way to quantify the many sources of variability that cl inicians s e e in their patients from day to day . 3 This approach is particularly useful in a test-retest research des ign . 3 The rationale for using a general izabi l i ty approach, as opposed to c lass ica l measurement theory to determine reliability, is the ability to identify the relative magnitude of the sources of error in the reliability equat ion. General izabi l i ty coefficients as outlined by Stratford, Norman and Mc in tosh 4 and the standard error of measurement ( S E M ) were used to evaluate grip strength measurements in this study. 44 Reliabil i ty is the consis tency of a measurement and consis ts of true variability (the degree of heterogeneity in the sample) and error (the difference between the obtained score and a true score due to uncontrolled factors that cannot be accounted for by the measurement protocol used) . 5 There are different factors that can threaten reliability and these include errors in the measurement tool itself (instrument variability), a lack of cons is tency in the patient response (subject variability) and errors made by those taking the measurements (tester variability). 6 Reliability is quantif ied in two ways , as either relative reliability or absolute reliability. 5 Relat ive reliability examines the relationship between two or more sets of repeated measures and is measured with a correlation coefficient, or in this study, a generalizabil i ty coefficient. 5 Abso lu te reliability, or the extent to which a score var ies on repeated measurement , is measured statistically by the Standard Error of Measurement (SEM) . The generalizabil i ty coeff icients (R) (variance attributed to dif ferences among subjects divided by the total var iance) generated in our study are unit less. Whi le the generalizabil i ty coefficient does not quantify the measurement error assoc ia ted with an individual score , the standard error of measurement ( S E M ) descr ibes reliability in clinically relevant terms and quantif ies measurement error of an individual's score. The S E M is like the sample standard deviat ion of an individual's theoretical performances and therefore its conf idence interval can be interpreted using the normal distribution. 7 For an observed score , the S E M quantif ies the range in which a true score might be expected to vary and therefore provides information required in order to make clinical dec is ions 7 Accord ing to El iasz iw et a l 6 , a reliability coefficient is just a point est imate based on one se lected sample and to be reliable a test must be able to discr iminate among individuals and error must be expressed in the s a m e units as the measurement . In 45 order to establ ish the true level of reliability in the population, as opposed to the sample , methods of data analys is must include tests of statistical s igni f icance appl ied to the reliability coefficients, the use of conf idence intervals and the calculat ion of the standard error of measurement ( S E M ) . 6 The use of generalizabil i ty coeff icients to measure reliability and the calculation of the S E M in the s a m e units as the measurements assist the researcher in differentiating true change from change assoc ia ted with error in test-retest des igns. It was felt this would be the best approach to ana lyze the data from this study based on the context of the research quest ion. Bea r -Lehman et a l 8 h a v e indicated that grip strength testing utilizing the J a m a r dynamometer is an objective rehabilitation outcome measure for which the reliability has not been thoroughly a s s e s s e d . Grip strength testing, utilizing a standard J a m a r dynamometer (As imow Engineer ing Company , Los Ange les , C A ) , 9 was se lec ted for investigation in this study due to its widespread use as a measurement tool in the cl inical rehabilitation set t ing. 1 0 The Amer ican Society of Hand Therapists recommended standardized positioning for grip strength measuremen ts . 1 0 Recen t literature has referred to this standardized testing m e t h o d 1 1 , 1 2 and therefore it w a s chosen as the method of testing in this study. Purpose of the Study The purpose of this study was to investigate the reliability of grip strength testing in healthy subjects. The test-retest reliability of the standard Jamar dynamometer w a s 46 investigated for two commonly administered grip strength tests, the Standard Gr ip Strength Test ( S G S T ) and the 5-Posit ion Grip Strength Test ( 5 P G S T ) (Appendix A) . Both the right and the left hand were investigated. It was hypothesized that 1) there would be no difference in the grip strength scores among 3 trials on the S G S T and 2) there would be no difference in the grip strength scores between test occas ions on both the S G S T and the 5 P G S T . Statement of Hypotheses Null Hypotheses : The S G S T is not reliable between trials (R* < 0.80). The S G S T and 5 P G S T is not reliable between test occas ions (R < 0.80). Al ternate Hypotheses : The S G S T is reliable between trials (R >0.80). The S G S T and 5 P G S T is reliable between test occas ions (R > 0.80) *R= generalizability coefficient METHODS Subjects Fourteen healthy male volunteers gave informed consent for testing. The sample of conven ience consis ted of physical therapy students and kinesiologists from the University of British Co lumbia . Subjects were exc luded from the study for any of the following reasons: 1) past history of hand injury, 2) lesions involving the cervical sp ine, 3) assoc ia ted neurological or musculoskeleta l injury affecting grip strength, 4) history of upper extremity surgery and 5) if the subject reported they were ambidextrous or left handed. The subjects ranged in age from 20 to 34 years with the average age 25.4 ± 47 3.8 years and the median age 24.5 years. Al l subjects were self-reported right hand dominant. With the except ion of one subject, all subjects had either no or minimal exper ience with the testing device (Jamar dynamometer) . The subjects reported no st renuous hand activity prior to testing when quest ioned at the beginning of each test sess ion . There were no inherent risks involved in this study. The subjects were ass igned a numeric code to ensure confidentiality of the data. Experimental Design In order to examine the reliability between trials and test occas ions , a repeated measures , within subject design was used (Appendix D). Eight grip strength trials were performed on 2 test occas ions for each hand. Re-test occas ions were within three days of the initial measurement . No clinically important change in the patient status w a s expected to occur between test days. Experimental Procedures Test Protocol The testing took place at the University of British Co lumb ia , Schoo l of Rehabi l i tat ion Sc iences and at Paci f ic Coas t Rehabil i tat ion Centre, a private practice physical therapy clinic in North Vancouver , B C . Three trials for the S G S T and 1 trial at e a c h of 5 posit ions for the 5 P G S T (a total of 8 grip strength trials) were taken on 2 test occas ions within three days of each other (range 1-3 days; mean = 1.5 days) . O n e examiner administered all testing. 48 Instrumentation A standard, adjustable handle, hydraulic Jamar Dynamometer (As imow Engineer ing C o . , Los Ange les , Calif.) was used to measure grip strength. The dynamometer w a s cal ibrated by the manufacturer prior to the study and was used exclusively for testing until the study had been .completed. The adjustable handle w a s used to test grip strength at grip span intervals (Posit ions 1-5) of 1.0, 1.5, 2.0, 2.5 and 3.0 inches (2.4 cm to 7.6 cm) for the 5 P G S T . The S G S T w a s tested at 1.5 inches (Posit ion 2). Gr ip strength was measured in ki lograms of force with the resolution of the dynamometer sca le read to the nearest 2 ki lograms (down from 1 kgforce). Positioning Differences in test procedures with respect to posit ioning, intertrial rest period, handle posit ion, number of trials and scores documented have been documented in the l i terature. 1 3 In this study, the standardized test protocol as outlined by the Amer i can Society of Hand Therapists (ASHT) was utilized to measure the static grip strength s c o r e s . 1 0 It w a s felt that a standardized method of testing would al low greater compar ison to the existing literature on the reliability of grip strength testing. The test procedure involved positioning the subject in sitting with the shoulder adducted and neutrally rotated, the e lbow at 90 degrees and the forearm and wrist in neutral (Figure 1a,b). The subject 's arm was not stabi l ized, and while efforts were made to conform to standard posit ioning, this al lowed slight variations in e lbow and wrist posit ions. The J a m a r dynamometer was p laced in the palmar gutter of the subject 's hand. The subject could not s e e the dynamometer dial or view the measurements during the study. 49 50 Testing Procedures The testing procedure was des igned to replicate the actual cl inical test situation and condit ions were as similar as possib le from day to day and from subject to subject. Study participants received standardized instructions (Appendix E) . The ethics commit tee of the University of British Co lumb ia approved this research. Subjects s igned a consent form prior to testing. Max imum time to complete the testing procedure was approximately 10 minutes for each testing sess ion for a combined total of 20 minutes. Subjects were randomly ass igned to either S G S T or 5 P G S T at the beginning of testing with the use of a computer ized random generat ion table. Between-day test sess ions were identical; subjects starting with S G S T on Day 1 started with S G S T on Day 2. Th is was done in order to eliminate order bias and ext raneous factors related to fatigue and learning. During the Standard Grip Strength Test ( S G S T ) , the subject w a s asked to maximal ly grip the Jamar dynamometer for 3 seconds at Posi t ion 2 on the adjustable handle with a 15-second rest in between trials as recommended by the Amer i can Soc ie ty of Hand Therap is ts . 1 0 Both left and right hands were tested, starting with the left hand a lways. Three grip strength trials were taken. During the 5-Posit ion Gr ip Strength Test ( 5 P G S T ) , the subject was asked to grip the dynamometer at each of the 5 posit ions for 3 seconds with a 15-second rest between trials (1 trial at e a c h position). A brief rest of 2 minutes was given between S G S T and 5 P G S T . Lengths of trials and rest per iods were t imed with a hand held stopwatch. No feedback regarding 51 performance w a s given, however, minimal verbal encouragement was provided during the testing (Appendix E) . This was kept consistent throughout the testing. Data Collection Test data including a test score for each of 3 trials and the mean score of 3 trials on the S G S T and a single score at each of the 5 posit ions on the 5 P G S T were col lected for each subject. Data for both left and right hands were recorded. Data were col lected on two separate occas ions within 7 days of each other with the mean length of t ime between test sess ions 1.5 days (sd=0.759, range =1 -3 days) . Sel f report activity information was col lected to ensure that the subject had not engaged in an activity immediately prior to testing that may have had an effect on grip strength. A n effort w a s made to retest the subjects at approximately the s a m e time e a c h day. Data Analysis The data were ana lyzed after 10 subjects for test-retest reliability for an est imate of samp le s i ze to meet statistical requirements. A s test-retest reliability using a Pea rson r correlation coefficient had been establ ished after 10 subjects ( r>8) , data col lect ion w a s s topped after 14 subjects. A power analysis for sample s ize was not done. Statistical Analysis Ana lys i s included a correlational analysis of relat ionships employing a general izabi l i ty a p p r o a c h 1 4 to determine the reliability of the measurement tool, the J a m a r dynamometer . A n analys is of var iance ( A N O V A ) was used to determine if the mean sco res differed significantly. The S G S T ana lyses were based on both a 3-way and 2-way A N O V A with an interaction term; whereas, the 5 P G S T ana lyses were based on a 2-way A N O V A without an interaction term (an interaction term could not be calculated because there was only one trial per sess ion) . Var iance components were calculated (Appendix G) and used to construct generalizabil i ty coeff icients. The standard error of measurement and the corresponding lower 1-sided 9 5 % conf idence limits were a lso calculated. Se lec ted var iance components were used to est imate the intertrial and interday standard error of measurement and their upper 9 5 % conf idence intervals. The ana lyses were done in MINITAB (Version 11) using the "Ba lanced A n o v a " option. Land is and K o c h 1 5 proposed a sca le to subjectively rate reliability where 0.00-0.20 indicates slight reliability, 0.21-0.40 fair, 0.41-0.60 moderate, 0.61-.80 substant ial and 0.81-1.00 almost perfect reliability. The generalizabil i ty coefficients in our study were compared using this subjective sca le . The raw data and A N O V A tables can be found in Append ix F and G respectively. 53 RESULTS Table 1: Samp le M e a n s Between Test Occas ions and A c r o s s Trials for the Standard Grip Strength Test (3 trials at Posit ion 2). Means n Right kgf Left kgf Test Occasion 1 42* 57.76 52.05 Test Occasion 2 42 60.12 55.24 Trial 1 28* 59.96 54.00 Trial 2 28 59.04 53.46 Trial 3 28 57.82 53.46 * 14 subjects X 3 trials X 1 test occasion = 42 • 14 subjects X 2 test occasions = 28 The sample means of all subjects (n=14) for the Standard Gr ip Strength Test are presented in Tab le 1. Tab le 2 represents the reliability study results of the Standard Grip Strength Test by presenting the right, left and right-left hand difference pooled est imates between trials (intertrial) and between test occas ion ( interoccasion) for 1 trial and the average of 3 trials. The generalizabil i ty coefficients (R) and the standard error of measurement ( S E M ) are indicated. The lower 1-sided 9 5 % conf idence interval is given for the generalizabil i ty coefficient as it is more important to the cl inician to know how low, or how far from perfect (R=1) the reliability is. Similarly, the upper 1-sided 9 5 % conf idence interval is given for the S E M , as it is important for the cl inician to know how large as opposed to how smal l a measurement could be due to error a lone. The results were presented this way for the purpose of exploring the reliability of cl inical measurement strategies such as looking at the result of taking a single grip strength 54 measurement (trial), taking the mean of 3 measurements (trials) or looking at the dif ference between hands. Table 2: General izabi l i ty coefficients (R) for the Standard Grip Strength Test . Inter-Trial Pooled Estimate Inter-Occasion Pooled Estimate R SEM (kgf) R S E M (kgf) RIGHT HAND Single Trial Average of 3 Trials .85 (.70)* .95 2.87 (3.40) f 1.66 .60 (.39)* .67 4.74 (4.76) f 4.12 L E F T HAND Single Trial Average of 3 Trials .84 (.66) .94 2.96 (3.51) 1.71 .62 (.40) .70 4.49 (4.59) 3.78 DIFFERENCE (R-L) Single Trial Average of 3 Trials .56 (.24) .80 3.46 (4.10) 2.00 .35 (.31) .49 4.21 (4.56) 3.18 'Lower 1-sided 95% confidence interval for the generalizability coefficient (R) tUpper 1-sided 95% confidence interval for the SEM kgf = kgforce The generalizabil i ty coefficients for the Standard Grip Strength Test (Table 2) were high for the inter-trial reliability of a single measurement on the right (R =.85) and left (R=.84) but the difference between right and left on a single trial exhibited moderate reliability (R=.56). The lower 1-sided 9 5 % conf idence intervals indicate a lower bound of 0.70 for the right hand, .66 for the left and .24 for the right-left dif ference. The average of 3 trials resulted in higher generalizabil i ty coefficients than for a single trial in all c a s e s . B a s e d on the within trial va lues of the Standard Error of Measurement ( S E M ) , there is a 9 5 % chance that the population value of the S E M is less than 3.40 kgf on the 55 right, 3.51 kgf on the left and 4.10 kgf between hands. The S G S T appeared reliable between trials (R>0.80) but not between test occas ions . Interoccasion, or test-retest est imates of generalizabil i ty coeff icients exhibited fair to moderate reliability with the left hand marginally higher than the right on a single trial (R=0.62 versus R=0.60 respectively). General izabi l i ty coefficients were slightly higher for the average of 3 trials. There was a 9 5 % chance that the populat ion va lue of the S E M is less than 4.76 for the right, 4.59 on the left and 4.56 for the right-left dif ference (Table 2). Est imate of var iance components (Table 3, Append ix G) indicated that the greatest source of variability, or dif ferences among the grip strength measurement scores , occurred in the subject by day interaction. High grip strength sco res occurred more often on the first of three grip strength trials in the right hand and appeared variable on the left hand (Figure 2). Table 3: Est imate of var iance components derived from the three-way A N O V A for the Standard Grip Strength Test. Intertrial Interoccasion Source Right Left R-L Diff Right Left R-L Diff Subjects 34.27 33.15 9.95 33.93 33.37 9.57 Trial 1.21 0 .22 Day 1.88 4.20 0 1.67 4.38 0 Subjects X Trial 0 0.66 0 Subjects X Day 12.40 7.22 6.87 12.54 6.98 6.11 Trial X Day 0 0.55 .23 Error 8.65 8.08 12.67 8.24 8.77 11.95 Note: Negative variances are set to zero. Please refer to Appendix G for calculations of the variance components. 56 High grip strength scores by trial for the Standard Grip Strength Test (3 trials at position 2 on the Jamar dynamometer) 100 Right Hand Test Right Hand Retest Left Hand Test Left Hand Retest Hand/Test O c c a s i o n Figure 2 : Proportion of high scores occurring on the first, second or third of 3 trials on the Standard Grip Strength Test. The 5-Posit ion Grip Strength Test correlation coefficients (Table 4 ) showed good reliability for J a m a r handle position 1 (R=0.87) and 2 (R=0.83) for the left hand only. The 5 P G S T did not appear to be reliable (R>0.80) between test occas ions . Var iance est imates of the 5-Posit ion Grip Strength Test including subject var iance, t ime var iance and error var iance are shown in Table 5. Figure 3 depicts the test-retest reliability di f ferences in the left and right hands across all handle posit ions; overall the left hand exhibited greater reliability. Table 6 indicates the test-retest measurement f luctuations and percent change for both the S G S T and the 5 P G S T . It appears that the fluctuation in individual scores across test occas ions is of the greatest concern in assess ing the s igni f icance of change over time. 57 Table 4: Summary of Correlat ion Coeff icients for the 5 Posit ion Grip Strength Test . Jamar Handle Correlation Coefficient S E M (kgf) Position (Lower 1-sided 95% Cl) (Upper 1-sided 95% Cl) Position 1-Left .87 (.69)* 2.79 (4.15) Position 1-Right .58 (.20) 5.42 (8.05) Position 2-Left .83 (.62) 2.71 (4.03) Position 2-Right .64 (.29) 4.38 (6.51) Position 3-Left .70 (.39) 5.06 (7.52) Position 3-Right .70 (.38) 3.62 (5.38) Position 4-Left .80 (.55) 3.09 (4.59) Position 4-Right .78 (.52) 3.28 (4.87) Position 5-Left .79 (.54) 3.29 (4.89) Position 5-Right .72 (.41) 2.57 (3.82) * SAMPLE CALCULATION (variance values from Table 5, Position 1 Left) Type 2,11CC = Subject Variance Subject Variance + Time Variance + Error Variance Type 2,11CC = 50.7 50.7 + 0 + 7.8 Type 2,1 ICC = .87 (lower 1-sided 95% confidence interval =.69) Table 5: Est imate of var iance components from a 2-Way A n o v a for the 5 Posi t ion Grip Strength Test. Source Position 1 Position 2 Position 3 Position 4 Position 5 Right Left Right Left Right Left Right Left Right Left Subjects 40 .95 50.75 34.50 41 .33 30.85 60.62 38.73 37.25 16.92 41 .85 Time 0* 0 0 .82 .08 0 .15 .05 0 .25 Error 29.46 7.84 19.18 7.354 13.13 25.60 10.74 9.55 6.63 10.82 •Negative variances set to zero. 0 S T j # CO TU o on ea ea =R 3 3 3 Q> 3 CD O 0 7 3 CO CD :ion F mean <0 O II ' S n F mean scon ale cn IE? cn scon c ?c C Q ' ' — - CD £0 CQ 3 " a cn CD co O 59.9 cn r—e O —i 3 59.9 OS CO Day — * O O 3 Day C Q cs re mi Q_ £ J 11 SG X re mi || OL CO 3 CO H O US "01 0 0 CD • O CD 7? II O CQ + ^ CD CO CD Da cn CO •< O O K> 11 -I CD II CD O OL CO CO CD CQ CD cn II cn C ^CR CD' O 1—»-^n 11 + cn + + 0 0 + O CD 7C CQ if CD 3 O CD CO H =f Si. CD C£ CQ ® (7) * c f cn CD P. ^ ^ C Q it -5' CO r o q -CD 3 CO C Q " »—1-O =!" o d 0 ) CD CO Q) o w CD • 3 CD 3 c o c CO o ' CO ^? C Q CD cn I T J o CO o ' (7) T J ' CO r-»--^  CD 13 C Q CD CO cu C L CD CO I— 0 ) C L 0 ) —1 C L CD Standard Grip Strength Test 0 ) CD W » O — h C O cn 0 0 C D O 0 O cn co bo 0 0 cn 0 0 cn C D 5- Position Grip Strength Test i.S 3 ^ cn r o 0 0 r o -vl r o co CD C D C O C O cn cn C D C D cn co " I f I % 0> =• O 3 CD (D (0 &> TJ I C_ O 0) 0) w 3 3 O CD ~* 3 9: " o q s o ' CO co CD 0) 3 Is CD 13 o CD c o c 0) l-t-o 3 to TJ 5' O 5? O CD , 3" O ?Hfi) 3 S <° 3 SS rt- W CQ 1 CD o ' 13 CO CD &) 3 0) CD 3 c c T o ^% CO I—»-CO o o Z2 CD °-13' 3 " C T si 0 ) 3 CD 13 13 O CD c o r-K c 0) rt> 5" 3 (/> TJ 5' o ^ O CD , 3" o ?Hai 3 & ® 3 S rt- W CQ T*" CD 2 > z o m -n H > z D 59 Test-Retest Reliability of Jamar Handle Position 1 Position 1- Position 2- Position 3- Position 4- Position 5- Position 1- Position 2- Position 3- Position 4- Position 5-Left Left Left Left Left Right Right Right Right Right Jamar Handle Position Figure 3: Test-retest reliability of 5 handle positions of the Jamar dynamometer. DISCUSSION Standard Grip Strength Test The general izabil i ty coefficients for the Standard Grip Strength Test in our study indicated almost perfect reliability (according to the subjective rating sca le as outl ined by Landis and Koch 1 5 ) on a single trial in the right hand (R=.85) and the left hand 60 (R=.84) (Table 2). This may lead one to surmise that taking a single grip strength measurement would be reliable in our study population. Bohannon and S a u n d e r s 1 6 investigated the use of a single trial for measur ing isometric e lbow flexion with a hand-held dynamometer in healthy subjects and found that the first measurement did not differ significantly from the maximal or the mean; reliability of all measures w a s comparab le . Bohannon and S a u n d e r s 1 6 warned about general iz ing their f indings to the cl inical setting though, noting that the examiner in their study w a s ski l led with the measurement tool and the measures were done in healthy individuals, who may not have exhibited as much subject variability as a patient population present ing with pathology. MacDermid et a l 1 2 examined the interrater reliabilities of pinch and grip strength in patients with cumulat ive trauma disorder and found that although the ICC ' s were slightly higher for the mean of three trials than for a single trial (0.96 versus 0.93 for symptomat ic patients and 0.98 versus 0.94 for asymptomat ic patients), there was no signif icant difference between the reliability coefficients suggest ing acceptab le scores could be obtained by using only one repetition. Whi le the use of a single grip strength measure may be desirable for the busy cl inician, Strat ford 1 7 caut ioned that with this approach, the therapist does not know if "the obtained single value reflects a stable measurement or is part of a trend". A l though the generalizabil i ty coefficients were high for a single trial in our study, they were higher for the average of 3 trials as indicated in Tab le 2 (right hand reliability .85 for a single trial compared to .95 for the average of 3 trials). Mathiowetz et a l 1 8 investigated the effects of three trials on grip strength by looking at the correlat ions for one trial, the mean of two trials, the mean of three trials and the highest score of three trials. They found that the mean of three trials ach ieved the highest correlat ions and a 61 single trial ach ieved the lowest correlations. B a s e d on this, they recommended the mean of three trials as a more accurate measure of grip strength than one trial or the highest score of three tr ials. 1 8 Hamil ton, Ba lnave and A d a m s , 1 1 investigating the test-retest reliability of grip strength using the Jamar dynamometer in healthy subjects, looked at the s a m e four methods to determine grip strength (the score of one trial, the mean score of two trials, the mean score of three trials, and the highest score of three trials) and conc luded that because all four methods were not significantly different in reliability, no one method could be recommended as more accurate. Stratford et a l 1 4 stated that averaging repetitions rather than taking the highest or lowest va lue in a test sess ion w a s the best strategy to obtain a representative est imate of grip strength. The results of our study indicate that the average of three scores was a more reliable measure of grip strength on all measures (right, left, difference between right and left and between days) (Table 2). Between day dif ferences, or test-retest reliability, exhibited fair to moderate reliability (right hand R=0.60, left hand R=0.62 for a single trial) according to the Land is and K o c h 1 5 sca le . Reliability between days was slightly higher for the average of three trials (0.60 for 1 trial versus 0.67 for the average of 3 trials in the right hand, 0.62 versus 0.70 in the left hand). Stratford et a l 1 4 , in a study of grip strength measurements in patients with tennis elbow, looked at the generalizabil i ty coeff icients for measurements taken within a test sess ion and from one sess ion to another. They conc luded that the greatest source of variation was between d a y s . 1 4 Overal l generalizabil i ty of results could best be enhanced by averaging grip strength measurements recorded from multiple test sess ions , as opposed to multiple test repetitions during a single test s e s s i o n . 1 4 Reliabil ity might be expected to be lower with a longer period between test sess ions 62 due to the possibil ity of a true change in the patient's cl inical status during that t ime. However , with our sample of normal uninjured individuals and a short length of t ime between test sess ions (mean of 1.5 days, range = 1 - 3 days) , the issue of t ime between testing and true change in our subjects is probably not relevant. Test familiarity would not be expected to change much in such a short t ime. Var iance for the trial and day exhibited smal l systemat ic dif ferences compared to the var iance for the subject /day interaction (Table 3). The largest source of var iance in our sample (Table 3) w a s the subject /day interaction, therefore, our results appear to support the research indicating that the greatest source of variation in grip strength measurements is in the subject by day interaction. The effect of fatigue and learning on grip strength measurement has been cited in the l i terature 1 9 and the important factors affecting fatigue and learning appear to be intertrial rest intervals and the number of repetitions per test s e s s i o n . 1 9 Pat terson and Bax te r 2 0 f ound maximum force occurred on the first, second , and third trial 3 5 % , 3 1 % and 3 4 % of the time respectively when using a 1 minute rest between trials. A s the rest period dec reased , max imum force occurred 6 1 % of the time on the first trial, 2 1 % on the second and 1 3 % on the third trial when using a 5-second rest between tr ia ls. 2 0 F e s s 2 1 found that during a 3-trial test high scores occurred 60 percent of the time on the first attempt and 24 percent of the time on the second attempt. In our study of the Standard Grip Strength Test, with a 15 second rest interval between trials, high scores on the right hand test occurred 4 4 % of the time on the first attempt, 3 3 % on the second attempt and 2 2 % on the third attempt (Figure 2). Retest scores on the right exhibited the highest score on the first attempt (47%), but in the left hand, the results were var iable. Math iowetz 1 9 found no practice or learning effect over three grip strength trials 63 with a 15 second rest in between trials and our results appeared to concur with Math iowetz 's results. W h e n investigating this further, a review of the sample means of all subjects (Table 1) indicates the difference between the means of the first trial to the means of the third trial is smal l ( 5 4 k g f - 53.46 kgf = 0.54 kgf) for the left hand, suggest ing a minimal effect on grip strength scores due to fatigue in the non-dominant hand. The right hand difference w a s greater (2.14 kgf), possibly suggest ing some fatigue. Math iowetz 1 9 found a dec rease of 0.41 kgf for the right hand and 0.86 kgf for the left over three repeat measurements of grip strength using the s a m e testing procedure as in our study. Hamilton et a l 1 1 found an overall difference between the first and third scores on all 5 handle posit ions of the Jamar to be 2.18 kgf for the group as a whole, leading them to suggest fatigue was evident within trials. F rom the literature, two factors appear important when looking at the effects of fatigue on grip strength and these include the length of each trial and the length of the intertrial interval. Theoret ical ly, the effects of fatigue could be minimized by increasing the rest t ime between trials, but this would substantial ly increase the time required for grip strength testing and may not be feasib le in a busy cl inical situation. Dunwoody, Tittmar and M c C l e a n 2 2 indicated that a 15 second to 6 minute variability in intertrial rest periods was evident in the literature. Dunwoody et a l 2 2 studied the effect of a 120 second intertrial interval on fatigue with a 3 second trial using the Martin Virgometer. They found that grip strength increased across trials from trial 1 to 4 and therefore a fatigue effect was not seen , but a learning effect w a s sugges ted . 2 2 Dunwoody et a l 2 2 caut ioned that familiarization or practice with the grip strength test may reduce the chance of a Type 1 error of reporting significant results which may 64 reflect the subject 's " learning" how to use the measurement tool. Mar ion and N iebuhr 2 3 examined the effect of warmup prior to maximal grip contractions, hypothesiz ing that subjects who engaged in mild warm up exerc ises would exhibit higher grip force; warm up consis ted of one submaximal grip. They found a statistically significant increase in grip strength due to warm up and noted that the highest forces occur at the beginning of a ser ies of maximal efforts 2 3 , the latter of which supports the f indings of our study. There may be a practice or learning effect evident in our study as the mean sco res for days were slightly better on the second day of testing (Table 1) throughout the samp le set (eg 2.36 kgf difference higher on the second day of testing on the right, 3.19 kgf on the left). The potential motor learning effect over two success i ve testing days seen in our study may indicate the need for practice testing prior to taking actual grip strength test scores . The addition of a third test occas ion would have been appropriate to s e e if a true learning effect was evident. B a s e d on our test results, the effect of fatigue was minimal in healthy adult males over three repeat trials using a 15-second rest interval between trials; there may have been a learning effect between test occas ions . 5-Position Grip Strength Test Reliabil ity of the 5 Posit ion Grip Strength Test ( 5 P G S T ) w a s greater than 0.80 (Table 4) for Posi t ion 1, 2 and 4 on the left but was fair to moderate for all other posit ions (Figure 3). Hamilton et a l 1 1 found high test-retest correlat ions ( ICC > 0.88) for the five handle posit ions on the Jamar dynamometer. Al though within acceptab le limits, the test-retest reliability of position 1 was lower than that of other handle posit ions. This does not compare with our f indings where position 1 resulted in the highest reliability on 65 the left hand (R=0.87), however it was true for the right hand (R=0.58) (Figure 3). During testing in our study, subjects reported that they found posit ion 1 uncomfortable and difficult to grip adequately due to the smal l s ize therefore we would have expected to s e e greater variability at this posit ion. Posit ion 2, which is most often used in grip strength test ing, exhibited good test-retest reliability on the left hand (.83) and substant ial reliability on the right (.64) with a 9 5 % chance that the populat ion va lue of the ICC is greater than .62 and .29 respectively (Table 4). Posi t ion 3 and 4 on the right (dominant hand) in our subjects exhibited the best reliability (Figure 2) and this may be expla ined by the subject 's hand s ize, which may have given them the greatest grip advantage and comfort at these posit ions, therefore they were more consistent in their measurements . Test-retest reliability appeared to increase from Posi t ion 1 to 4 on the right hand (Figure 3). At position 2 on the right hand, the S E M was 4.38 kgf with an upper limit of 6.51 kgf (95% Cl) due to error a lone (Table 4). Right versus Left Hand Difference General izabi l i ty coefficients were ana lyzed for the right to left hand difference for the Standard Grip Strength Test (Table 2). The right hand was consistent ly stronger as demonstrated by higher trial scores compared to the left (Table 1, Append ix F). The reliability for the intertrial difference between right and left hands w a s .56 with a S E M of 3.46kgf. The interoccasion reliability was .35 with a S E M of 4 .21 . Reliabil ity w a s improved by taking the average of three trials with the intertrial reliability .80 ( S E M 2.00 kgf) and the interoccasion reliability .49 ( S E M 3.18 kgf) (Table 2). Hamil ton et a l 1 1 and Math iowetz 1 8 found that the test-retest reliability of the right hand was lower than that of the left, a finding substantiated in our study, although to a smal l degree (R=.67 right 66 hand versus R=.70 left hand for the test retest average of three trials). However , Hamil ton et a l 1 1 noted that they tested the right hand first and suggested an ordering effect may have occurred with the left hand achieving greater reliability as a result of a learning effect. In our study, the left hand was tested first and our subjects were all right hand dominant which tends to refute this statement. The 1 0 % rule is often referred to in the literature to define the average dif ference in grasping power between the dominant and non-dominant h a n d . 2 4 However , the origin of the accepted 1 0 % difference remains obscure and the means of calculat ing this rule (the instrument used and the population from which the average difference w a s obtained) has not been explained in the literature. Peterson et a l 2 5 indicated that the 1 0 % rule dated back to the work of Bech to l 2 6 in 1954 who observed that most patients presented with a 5 % to 1 0 % difference between their dominant and non-dominant hands on grip strength measurements , with the dominant hand being stronger. Lunde et a l 2 7 , in a 1972 study of the grip strength of col lege women , found a 1 3 % dif ference in grip strength between the dominant and non-dominant hands, however 2 4 % of the 107 grip strength measurements revealed strength readings for the non-dominant hand equal to or greater than the dominant hand. A study by Schmidt and T o e w s 2 8 found 2 8 % of men tested had a grip strength in their non-dominant hand that w a s equal to or greater than the grip strength in their dominant hand, therefore, they quest ioned the appl icat ion of the 1 0 % rule. Peterson et a l 2 5 found a 10.74% difference overal l between hands, however, when they separated out the data into right hand dominant and left hand dominant individuals, the right handed subjects exhibited a 12 .72% dif ference and the left handed subjects exhibited a - 0 . 0 8 % difference. Therefore, they conc luded that the 1 0 % rule is only valid for right handed persons; for left handed persons, grip 67 strength is equal to right. 2 5 Harkonen et a l 2 9 in a sample of 204 healthy Finnish adults, found no significant difference in strength between the dominant and non-dominant hands. Clinical Significance Tab le 6 indicates the mean grip strength fluctuations in kgforce for the 5-Posi t ion Grip Strength Test and the Standard Grip Strength Test. Cl inical ly, therapists will test a patient on one day, then retest them again at a later date to try and determine if a clinically important change has occurred. Our test-retest reliability study mimicked this with a smal l (mean = 1.5 days) number of days between test sess ions . However , the percentage change from day to day among healthy individuals was shown to range from 5 . 1 % to 12 .8% with a fluctuation (mean difference of highest to lowest scores) of 2.8kgf to 6.64 kgf. Therefore, in our study population, between 2.8kgf and 6.64 kgf dif ference in measurement between grip strength trials on a test-retest situation appears to be due to error a lone. Accord ing to Harkonen et a l 3 0 , the accuracy of the J a m a r dynamometer at all 5 test posit ions varied from 1.2 kgf to 1.4 kgf when tested against known weights. Error in ki lograms increased with heavier weights but the relative percentage error dec reased when the load increased. This indicates that there may be more error in those individuals that present with low grip strength scores , whether this is due to true w e a k n e s s , manual disability, gender, job type or poor effort. Therefore, cl inicians must be aware that there is a degree of measurement error, particularly with low grip strength scores , that should be factored into their interpretation of test results. 68 Grip strength testing measurements were more reliable when the average of 3 trials w a s taken compared to a single trial, suggest ing an effective measurement strategy would be to use multiple trials. Fat igue did not appear to be a factor in our study and therefore the use of a 15 second trial interval appears to be sufficient. B e c a u s e grip strength measures are relatively s imple and quick to perform in the cl inical sett ing, taking three trials with a 15 second rest between trials should not be too time consuming for the busy cl inician and will provide him/her with a reliable est imate of grip strength on that day. Overal l , the use of a standardized protocol would be beneficial for all c l inicians in order to accurately compare patient results ac ross cl inics and between treating physical therapists. Our study found that the greatest variability occurred in the subject by day interaction. Unfortunately, this is what cl inicians must measure , as they try to determine patient change over time. In our study, the test-retest standard error of measurement ( S E M ) w a s 4.12 kgf for the right hand and 3.78 kgf for the left hand for the average of three trials. If we now wanted to use this information to general ize to individuals that matched our sample population (healthy right handed males between the ages of 20 and 35) we could perform the following calculation using, for example , 9 5 % conf idence intervals (Cl) and the standard error of measurement ( S E M ) : Samp le Cl inical Measurement Strategy for a Hypothetical Patient Similar to our S a m p l e Populat ion The confidence in a value for the mean of 3 trials on the same occasion = measured value + (z-value for CI)(SEM) 31 If the hypothetical patient's right hand mean score based on 3 t r ia ls/occasion = 59 kgf 9 5 % conf idence interval z-value (from statistical tables) = 1.96 S E M (from our study) for right hand, average of 3 trials = 4.12 kgf 69 Therefore, from equation above: 59 kgf + (1.96 x 4.12) = 59 ± 8.08 (50.92 to 67.08) Therefore we would be 9 5 % confident that changes greater than (or less than) ± 8.08 kgf would signify changes greater than that expected by chance . For the hypothetical patient score in the example above, this change due to chance lies somewhere between 50.92 kgf and 67.08 kgf: scores greater than 67.08 kgf or less than 50.92 may represent true change in the subject 's performance. Limitations of the Study The results of our study did not exhibit as high test-retest reliability as has been reported in the literature. This may be due to a number of factors. Reliabil ity can be threatened by errors in the measurement tool itself (instrument variability), errors made by those taking the measurements (tester variability) and a lack of cons is tency in the patient being examined (subject variability). 6 The J a m a r dynamometer used in the study was calibrated prior to the start of the study but was not cal ibrated after to s e e if there was any change throughout the testing in its measurement accuracy thus affecting instrument variability. The J a m a r w a s only used for subject testing during the time of the study so there was no potential interference by other cl inician use. O n e examiner did all the testing in this study. It was not felt that calibration accuracy would change significantly over the course of 3 weeks of grip strength testing, however it would have been interesting to calibrate the dynamometer in the clinic pre and post test. Measurement scores were read to the nearest ki logram down and whi le every effort w a s made to be consistent in the taking of measurement , there may have been 70 s o m e error assoc ia ted with tester variability. The tester was very familiar with the J a m a r dynamometer and was skil led in its use, probably minimizing tester error. The methodology for grip strength testing was based on a s tandard ized procedure as outlined in the l i terature. 1 1 , 1 8 However, the examiner wanted to make this study as clinically relevant as possib le and therefore sterile laboratory condit ions were not imposed on the subjects. Whi le positioning protocol was adhered to by the examiner and corrected prior to the start of a grip strength trial, the subjects tended to slightly modify wrist position to be comfortable in their grip, which is what is seen in the cl inical sett ing. Minimal verbal encouragement was given to the subjects throughout testing al though it can be argued that a "coaching" tone of vo ice w a s used . The subjects in our study started with their left hand, as opposed to the right hand which may have had some effect. Whi le subjects were largely tested individually, there were a few instances where 2 subjects were in the testing area at the s a m e time (one waiting to be tested, the other being tested) and it is not known if the p resence of another individual had an effect on motivation or maximal effort. However, it can be argued that often there are other individuals present in a clinical setting when a therapist is taking grip strength measurements with their patients. The reason for using a generalizabil i ty approach in our analys is of the data for the test-retest reliability study was to be able to allow for the decomposi t ion of the error var iance into unique sources of variation, and express the relative magni tude of these sources of error. Overal l , the component of subject variability over test occas ions contributed the most to the variability seen in our study as ev idenced by the var iance est imates for the two grip strength tests tested (Tables 3 and 5). Cl inical ly, this subject 71 variability is difficult for therapists to control over repeated test occas ions , however they need to be aware of this source of error when interpreting grip strength measurements . Final ly, sample s ize was smal l in this study (n=14) resulting in large conf idence intervals, al though the sample s ize was homogeneous which should have lead to smal ler S E M ' s . Larger samples permit more accurate estimation of true populat ion va l ues 3 and sample s ize may have contributed to the variability in our study. Recommendations for Further Research The information from this test-retest reliability study enab les us to make the following recommendat ions for further research: 1. Recogn i zed grip strength testing protocols should be fol lowed. Protocols should be clearly def ined and standardized in order to improve consistent testing between and within therapists in the clinical setting. This should include posit ioning, testing procedures, number of trials, intertrial rest periods and equipment cal ibrat ion. 2. The mean of 3 measurement scores appears to be the most reliable based on the results of our study. Further research on this measurement strategy is recommended in var ious patient groups. 3. The grip strength difference between the dominant and non-dominant hand needs to be more fully investigated before making compar isons between hands with grip strength testing. 4. R e s e a r c h needs to be done on patient populat ions as opposed to healthy individuals in order to determine what the true var iance in patient grip strength scores is. It is assumed that a person with a pathology will mimic healthy individuals but present with lower measurement scores (pers.comm, Pau l Stratford). Th is may not a lways be the case . 72 CONCLUSION This study attempted to investigate the reliability of grip strength testing in healthy subjects using two commonly administered grip strength tests. The results of our study indicate that the Standard Grip Strength Test was reliable (R >0.80) between trials for the right hand (R=.85) and the left (R=.84) for a single trial and for the mean of 3 trials (R=.95 right, R=.94 left). The Standard Grip Strength Test exhibited substant ial reliability (R=0.61 to 0.80) between days. The 5-Posit ion Grip Strength Test w a s reliable between days at position 1, 2 and 4 on the left hand (R >0.80) and had substant ial reliability (R=0.61-0.80) at other posit ions and with the right hand. There w a s more subject variability between days on all tests. T h e s e f indings have implications for cl inical practice and for the development of a measurement strategy to aid in the detect ion of clinically important change. 73 R E F E R E N C E S 1. F e s s E E . The need for reliability and validity in hand assessmen t instruments. Journa l of Hand Surgery 1986; 11A(5): 621-623. 2. F e s s E E . Human performance: A n appropriate measure of instrument reliability ? Journa l of Hand Therapy 1997; 10(1): 46-47. 3. Domholdt E. Phys ica l therapy research: principles and appl icat ions. Toronto: W . B . Saunders ; 1993. 4. Stratford P W , Normal G R , Mcintosh J M . General izabi l i ty of grip strength measures in patients with tennis elbow. Phys ica l Therapy 1989; 69 :276-281. 5. Rothstein J M . Measurement in Phys ica l Therapy: Cl in ics in Phys ica l Therapy. New York, N Y : Churchi l l Livingston; 1985. 6. E l iasz iw M, Y o u n g S L , Woodbury M G , Fryday-Fie ld , K. Statist ical methodology for the concurrent assessmen t of interrater and intrarater reliability: Us ing goniometr ic measurements as an example . Phys ica l Therapy 1994; 74:8: 777-788. 7. B i rmingham T B , Kramer J F , Speech ley M, Cheswor th B, MacDermid J . Test-retest reliability of the coefficient of variation as a measure of sincerity of effort during isometric testing. Physiotherapy C a n a d a 1997; 49:3:184-190. 8. Bea r -Lehman J , Ab reau B C . Evaluat ing the hand: Issues in reliability and validity. Phys ica l Therapy 1989; 69(12): 1025-1033. 9. A s i m o w Engineer ing Company . Jamar Adjustable Hand Dynamometer . Los Ange les : A s i m o w Engineer ing. 10. F e s s E E , Moran C A . Cl inical A s s e s s m e n t Recommendat ions . Garner , North Caro l ina : Amer ican Society of Hand Therapists 1981: 6-8. 11. Hamil ton A , Ba lnave R, A d a m s R. Grip strength testing reliability. Journal of Hand Therapy 1994; 7(3): 163-170. 12. MacDerm id J C , Kramer J F , Woodbury M G , McFar lane R M , Roth J H . Interrater reliability of pinch and grip strength measurements in clients with cumulat ive t rauma disorders. Journal of Hand Therapy 1994; 7: 10-14. 13. Smith R O , Benge M W . P inch and grasp strength: Standardizat ion of terminology and protocol. Amer ican Journal of Occupat ional Therapy 1985; 39(8):531-535. 14. Stratford P W , Normal G R , Mcin tosh J M . General izabi l i ty of grip strength measures in patients with tennis elbow. Phys ica l Therapy 1989; 69 :276-281. 74 15. Land is J R , Koch G G . The measurement of observer agreement for categor ical data. 1977 Biometr ics 33:159-174 16. Bohannon R W , Saunders N. Hand-held dynamometry: A single trial may be adequate for measur ing muscle strength in healthy individuals. Physiotherapy C a n a d a 1990; 42:1:6-9. 17. Stratford P W . Summar iz ing the results of multiple strength trials; truth or consequence . Physiotherapy C a n a d a 1992; 44 :1 : 14-18. 18. Mathiowetz V , W e b e r K, Vo l land G , K a s h m a n N. Reliability and validity of hand strength evaluat ions. Journal of Hand Surgery (Am) 1984; 9:222-226. 19. Mathiowetz, V . Effects of three trials on grip and pinch strength measurements . Journa l of Hand Therapy 1990; 3(4): 195-198. 20. Pat terson R P , Baxter T. A multiple muscle strength testing protocol. A rch ives of Phys ica l Medic ine and Rehabil i tation 1988; 69:366-368. 21 . F e s s E E . The effects of Jamar handle position and test protocol on normal grip strength. Journal of Hand Surgery 1982; 7A:308. 22. Dunwoody L, Tittmar H G , M c C l e a n W S . Grip strength and intertrial rest. Perceptua l and Motor Ski l ls 1996; 83:275-278. 23 . Mar ion R, Niebuhr B R . Effect of warm-up prior to maximal grip contract ions. Journa l of Hand Therapy 1992; 5(3):143-146. 24. T o e w s J V . A grip strength study among steelworkers. Arch ives of Phys ica l Medic ine and Rehabil i tat ion 1964; 45:413-417. 25. Pe te rsen P, Petrick M, Connor H, Conkl in D. Grip strength and hand dominance: chal lenging the 1 0 % rule. Amer ican Journal of Occupat iona l Therapy 1989; 43(7):444-7. 26. Bechto l C O . Grip Test: The use of a dynamometer with adjustible handle spac ings . Journa l of Bone and Joint Surgery (Am.) 1954; 36A: 832. 27. Lunde B K , Brewer W D , Garc ia P A . Grip strength of col lege women . Arch ives of Phys ica l Medic ine and Rehabil i tation 1972; 491-493. 28. Schmidt RT , Toews J V . Grip strength as measured by J a m a r dynamometer . A rch ives of Phys ica l Medic ine 1970; 51:321-327. 29. Harkonen R, Pi i r tomaa M and Alaranta H. Grip strength and hand posit ion of the dynamometer in 204 Finnish adults. Journal of Hand Surgery 1993; 18B(1): 129-132. 75 30. Harkonen R, Harju R, Alaranta H. Accu racy of the Jamar dynamometer . Journal of Hand Therapy 1993; 6:259-262. 31 . Hopk ins K D , Hopkins B R , G l a s s G V . Bas i c Statist ics for the Behavioura l S c i e n c e s , 3 r d ed . Toronto: Al lyn and Bacon , 1996. 76 CHAPTER SUMMARY AND CONCLUSIONS In cl inical practice, therapists need to know if their treatment strategies are effective in order to plan and progress care for their patients. Of value to the therapist, is the ability to monitor change in their patients with a valid and reliable measurement tool . 1 Gr ip strength assessmen ts are commonly used in cl inical practice due to the objectivity of the measurement dev ices, the relative e a s e and cost-ef fect iveness of test administrat ion and the availability of normative da ta . 2 The Jamar dynamometer w a s chosen for assessmen t in our study due to its widespread use as a grip strength measurement tool in rehabil itation. 3 The purpose of our study was twofold: 1) to survey therapist 's use of grip strength measures in cl inical practice and 2) to determine the reliability of grip strength testing in healthy subjects. Of interest w a s the investigation of what dr ives cl inical decis ion making with respect to grip strength measures . M e a s u r e s of strength must be sensit ive enough to dist inguish between subnormal and normal musc le strength and be precise enough to document both increases and dec reases in strength. 4 With all measures or assessmen t tools, it is important that the user know what the normal variation in the measure of interest is, how reliable the measure is and how to interpret the results of the measure . To be able to measure or detect change in a patient's status effectively, the measurement tool should be highly sensit ive to clinically important change. The following d iscuss ion attempts to incorporate the results of our survey study, the test-retest reliability study and a review of the existing literature on grip strength testing in order to provide s o m e direction to practicing therapists regarding the normal variation, the reliability and the interpretation of grip strength measurements with respect to clinically important change. What is the normal variation in grip strength measures? Y o u n g et a l 5 studied the variations in grip strength among 95 healthy men and women using all 5 posit ions of the Jamar dynamometer. Grip strength w a s measured in the morning and afternoon, 2 days a week for 3 weeks for a total of 12 testing per iods. Over the course of this testing, they found that mean grip strength f luctuated between 5.1 kgf and 8.4 kgf, or between 19.2% and 23 .7% when compared to the group means . Speci f ical ly for men in their study (n=34, 30 being right handed, age range 18 to 67 years) , the dominant hand exhibited an 8.4 kgf grip strength difference and the non-dominant hand 8.1 kgf difference (19.2% and 19.4% change respectively) over all 5 handle posit ions of the Jamar dynamometer. In our study of 14 healthy right hand dominant males (mean age 25.4 years), the dominant hand exhibited a 4.7 kgf (8%) grip strength difference with the mean of 3 trials at position 2 on the Standard Grip Strength Test and an overall 4.2 kgf (8.2%) difference over all 5 posit ions on the J a m a r (Chapter 3, Tab le 6). Whi le our va lues appear lower than that found by Y o u n g et a l 5 , the di f ferences in sample s ize, the number of test occas ions (2 in our study, 12 in Y o u n g et al 's) and the large age range of the subjects probably accounted for the larger fluctuation in grip strength measures seen in Young et al 's study. In our study, the standard error of measurement for the test-retest mean of 3 trials on the Standard Grip 78 Strength Test w a s 4.12 kgf for the right hand and 3.78 kgf for the left hand. A s wel l , the S E M of the difference between right and left hands was 3.18 kgf for the average of 3 trials. A s the subjects in our study were healthy individuals, and the time between testing w a s short enough (mean=1.5 days) that no appreciable change in the subject 's status should have occurred, fluctuation in grip strength measurements w a s thought to be due to error a lone in our study subjects. The results of our survey study indicated that respondents felt that a 5 kgforce difference or 1 0 % difference between trials was clinically significant to base treatment dec is ions on. However, based on the results of our reliability study in healthy individuals, it would appear that approximately 4.74 kgf for the right hand and 4.49 kgf for the left could be due to error a lone and may be a measure of normal variability in our sample . M e a n grip strength fluctuations in our healthy sample populat ion exhibited 4.6 kgf to 6.6 kgf f luctuations between trials on the S G S T in the right hand and 4.2 kgf to 5.9 kgf in the left (Chapter 3, Table 6). Grip strength fluctuations were less when the mean of 3 trials was considered (right = 4.7 kgf, left = 4.8 kgf). Percent dif ference between trials on the S G S T varied from 7.7% to 11.5 % on the right hand and 7 .9% to 10 .8% on the left hand. Therefore, cl inicians may be stating there is a true change based on their use of 5kgf or 10% difference between trials for cl inical s igni f icance when one has not really occurred and the change seen is due to normal subject variability from day to day. Survey respondents a lso indicated that a 5kgf or 10-20% difference between hands would be clinically significant. B a s e d on our study results of the S G S T , the S E M for the difference between right and left hands was 4.21 kgf for a single trial and 3.18 for the average of 3 trials. Peterson et a l 6 used a formula in their study to determine the 79 percent dif ference between hands. They took the score of the nondominant hand and divided it by the score of the dominant hand and this value was subtracted from 1. For example , for a right handed subject in their study, with a right hand score of 80 kg force and a left hand score of 72 kg force, the calculation would be 1 - (72/80) = .10 (10%). Peterson et a l 7 used the stronger of 2 grip strength trials, at either posit ion 2 or 3 on the J a m a r dynamometer (depending on patient preference and comfort). The subjects in our study were all right hand dominant and we used position 2 on the J a m a r for the Standard Grip Strength Test. Using Peterson et a l ' s 6 formula with data from our reliability study to obtain the percent difference between hands with a s ingle subject (LW) (see Append ix F), taking the highest of 3 trials as per Peterson et a l 6 , the calculat ion would be 1 - (52kgf/66kgf) = 2 1 % for test occas ion #1 and 1-(58kgf/72kgf) = 1 9 % for the retest occas ion . However, if we take the mean of all subjects trials (3 trials t imes 14 subjects) on the first test occas ion , the percent difference between hands is 1 0 % and the test-retest percent difference is 8%. Therefore, the 1 0 % rule 6 , which is vaguely descr ibed in the literature as the average difference in grasping power between the dominant and non-dominant hand, did not appear to apply to the individual subjects in our study, but more closely matched the group norms of our right handed subjects. However , in cl inical practice, the therapist is usually concerned with an individual patient and it would appear that normal variability between the dominant and non-dominant hands may not concur with the 10% rule in single subjects. In our test-retest reliability study, we found the greatest source of error w a s due to subject variation between test occas ions . Factors that affect subject variability include the patient's level of attention and/or motivation, the environmental stimuli surrounding the test site, the patient's familiarity with the testing procedure, and the 80 posit ioning of the patient such as the placement of the dynamometer in the patient's hand . 4 In our study of healthy individuals, a range of measurement f luctuations of 3 to 7 kgf (5-13%) was noted over 2 test occas ions for all posit ions on the J a m a r dynamometer . Individuals that present with true pathology may present with a greater variation ac ross test occas ions due to other factors not present in our study such as pain, medicat ion use, fear and apprehension or actual anatomical changes in grasping ability. A s cl inicians use test-retest occas ions to base their treatment dec is ions on, being aware of and attempting to minimize controllable factors that may affect subject variability (environment, patient posit ioning, time of day) should be taken during grip strength testing and uncontrol lable factors (pain level) should be documented . How reliable are grip strength measures? The literature indicates that grip strength measurements for the standard J a m a r dynamometer exhibit high (r >0.80) inter-rater re l iab i l i ty 7 8 9 and high test-retest reliability. 7" 1 1 In our study, the reliability of the Standard Grip Strength Test w a s less than 0.80 but greater than 0.35 for the test-retest reliability of the right hand, the left hand and the difference between the right and left hands on a single trial and the mean of 3 trials. In the 5 Posi t ion Grip Strength Test, left hand test-retest reliability ranged from R=0.70 to R=0.87, with Posit ion 1,2 and 4 exhibiting the greatest reliability. In the right hand, test-retest reliability ranged from R=0.58 to R=0.78 with Posi t ion 4,5 and 3 ( in order of decreas ing magnitude) exhibiting the greatest reliability. Whi le our study w a s limited due to the smal l sample s ize (n=14), it appeared that the average of 3 trials 81 ( S G S T ) y ie lded the greatest test-retest reliability (R=.67 right and R=.70 left) and the difference between hands the least test-retest reliability (R=.49). Overal l , grip strength measurements using the standard Jamar dynamometer exhibited substant ial reliability as a measurement tool. How can we interpret the results of grip strength measures? A valid measurement tool should be able to detect and quantify a clinically important di f ference. 1 With the assessmen t of grip strength, the therapist is often interested in determining if a weakness exists. If a weakness is identified, the therapist wants to determine whether a change in strength has occurred over t ime. 1 2 A compar ison of the contralateral hand is often used to establ ish grip strength w e a k n e s s and the compar ison of the subject 's current measurement score with that of a previous measurement determines if a change in strength has occurred over t ime. 1 2 Many cl inicians base their measurement interpretation on the difference between the "normal" and "injured" hand in grip strength testing, with the "normal" hand assuming the criterion measure for compar ison. The Amer ican Medica l Assoc ia t ion 's Gu ides to Impairment 1 3 indicate that there is little ev idence for a significant difference in grip strength between the dominant and non-dominant hand, but a 10% difference between the dominant and nondominant hand is often referred to by cl inicians with respect to grip strength measures . However, it appears that cl inicians must be careful when using the 1 0 % rule to apply to their patient populat ions for 2 reasons. One , the studies reporting di f ferences were done on normal subjects and therefore the dif ferences among those 8 2 that truly do exhibit a difference between their dominant and nondominant hands are unknown and secondly , the 10% rule appears to apply only to right handed indiv iduals 6 and should not be general ized to left handed or ambidextrous individuals. Our results did not appear to support the 10% rule for individual subjects. Never the less, the Amer ican Medica l Assoc ia t ion (AMA) base their impairments of grip strength on a calculat ion: (Normal Strength - Abnormal Strength)/Normal Strength for a Percen tage Strength Loss Index. 1 3 For example , if a patient's involved right hand measures 20kgf and that of his left hand, which is normal, is 45 kgf, the equat ion would be (45kgf-20kgf)/45kgf = 5 6 % Strength Loss Index. This 5 6 % strength loss is cons idered to be a 2 0 % impairment of the upper extremity according to their table of upper extremity impairment for loss of strength. Accord ing to the A M A strength loss tables, if a right handed patient presented with grip strength measures of 55kgf in the left hand and 50 kgf in the injured right hand, the calculat ion would work out to a 9% strength loss and the patient would probably be awarded a 1 0 % impairment loss. However , the 5 kgf difference could be due to error a lone based on the results of our study on the normal variability in normal individuals. This has ser ious implications when looking at medico legal issues and the rel iance of grip strength measures to award impairment and disability ratings. The cl inical s igni f icance of a measure refers to the ability of that measure to dist inguish between important and unimportant change. The therapist must feel confident that their measurements reflect a true measurement of their patient's abilit ies. A measurement strategy to more accurately determine clinically important change in a patient's status with respect to grip strength measures requires the calculat ion of the standard error of measurement (SEM) and a corresponding conf idence interval (Cl) for 83 that patient. For an observed score, the S E M quantif ies the range in which the true score might be expected to vary and this provides information that needs to be cons idered in cl inical dec is ion making. A true change in a patient's grip strength score would be calculated by the measured score ± (z-value for CI)(SEM). Us ing this type of statistical calculat ion for each patient would lead to an improved clinical dec is ion making p rocess . The interpretation of impairment measures such as grip strength often involve cl inical assumpt ions on the part of the therapist which include: • grip strength measurements are "hard" or objective data • impairment measures are related to the patient's symptoms, pathology or functional limitations • a change in grip strength will correlate with changes in functional status • treating impairments or grip strength weakness will improve funct ion. 1 4 Whi le these assumpt ions need to be further investigated, the therapist must be aware that there are many factors that may affect grip strength measures ; even in our healthy, motivated sample population we noted a variation in test scores across days . A s well , grip strength measures may not perform equal ly well at all levels of injury; for example , they may not be as sensit ive to change with those individuals presenting with very low or very high grip strength scores . Finally, the correlation of changes in grip strength sco res and functional status needs to be addressed in order to more accurately dist inguish between important and unimportant change, or use less or useful therapy to increase in grip strength. The therapist must be able to take all these factors into account when attempting to interpret the results of grip strength measures . 84 Recommendations for Grip Strength Measurements A s B o h a n n o n 4 indicated, many of the m e a s u r e s u s e d in rehabilitation are dependent on the patient and no matter how objective the test, variability within the patient and their level of effort is bound to have an effect on the outcome. 4 Ideally, e a c h patient should be treated individually when interpreting what is "normal" for that individual. Compared to other strength measures (i.e. the Kin C o m isokinetic dynamometer) , grip strength testing is relatively quick, economica l and e a s y to perform in the clinic. However, the busy clinician is a lways faced with the d i lemma of averaging scores over multiple test sess ions to obtain a greater est imate of variability in a patient's scores as opposed to saving time by taking a single measurement . A s wel l , busy cl inicians may feel they do not have the time or the mathematical expert ise to calculate the S E M and conf idence intervals for their measurement scores . However , an easy method to more accurately measure a patient's score would be to take the average of multiple measurements over a number of patient visits (i.e. 3 measurements , over 2 days = an average of 6 measurements) . The trade off of cost (in time) versus quality of the dec is ion is that therapists taking into account normal variability ac ross test occas ions in their patients would have to see a greater change in a patient's grip strength measure in order for that measure to be deemed clinically significant. In order to improve the ability to detect clinically important change , cl inicians should be aware of the normal variability in measurement when employing grip strength testing in the cl inical setting. The following recommendat ions are based on a review of the literature on grip strength measurements , the results of our survey study of 85 cl in ic ians and our test-retest reliability study of 2 common grip strength tests. 1. Test Knowledge: Whi le important in any testing but not directly examined in this study, before using any clinical test procedure, therapists should be aware of the purpose of the test, the theoretical basis for the test and the test setting in which the test has been deve loped and u s e d . 1 5 A n understanding of avai lable measurement tools and their properties is imperative to obtaining val id, reliable measurements and in turn, making effective clinical dec is ions. 2. Standardized Positioning: It is recommended that s tandardized posit ioning as per the Amer i can Society of Hand Therapists (patient in sitting with the shoulder adducted and neutrally rotated, e lbow at 90 degrees and the forearm and wrist in neutral, arm not stabil ized) be used consistently for testing. Test posit ion should be documented in the charting procedures to allow more accurate replication of measurements by another tester. 3. Standardized Test Protocol: A standardized test protocol should be adhered to by the cl inician. A 15 second intertrial rest appears sufficient to minimize fatigue between trials. Consis tent protocol should be fol lowed (i.e. s tandard procedures for test administration and scor ing, starting with the s a m e hand each time for testing, using the s a m e commands , testing in the same room, document ing activity prior to testing that may affect grip strength) to minimize subject variation between test occas ions as much as possib le. 4. Measurement Strategy: The mean of 3 trials was more reliable in our study and is recommended as a measurement strategy as opposed to taking one trial. 5. Establishing Normal Variability: T ime may be a factor in the cl inical sett ing, but ideally therapists should establ ish a basel ine of grip strength measurements for each individual patient when attempting to measure change over time. From the literature, it appears that determination of true grip strength is best made with measurements taken over a number of test occas ions . From our study, the greatest source of variation was due to subject variability over test occas ions . Averag ing sco res over a period of a few days to determine a basel ine may be helpful in detecting true change from change assoc ia ted with error (i.e. 3 measures per day averaged over 2 days (6 measures in total)). Whi le grip strength testing exhibits substant ial reliability, therapists must be cautious when basing their cl inical dec is ions on just one measure of grip strength. 6. Measurement error: The Standard Error of Measurement ( S E M ) is important to calculate with test-retest grip strength measurements as it g ives an est imate of error in the s a m e units as the measurement . Conf idence intervals for the S E M should a lso be calculated. The example as outlined by Stratford and Go ldsm i th 1 6 may be helpful for cl inicians in understanding and calculating the S E M and its corresponding conf idence interval. 86 7. Use of a Normative Database: A normative da tabase should only be compared to when the s a m e test protocol, sample population and instrument used by the cl inician has been used to develop the normative data. 8. Calibration: Whi le not a focus in this paper, calibration is important to improve instrument accuracy and therefore the conf idence of measurement scores . It is recommended that hand dynamometers be calibrated on a regular bas is (calibration recommended once a year by the manufacturer) or develop a method of calibrating the dynamometer with known weights as per the method outlined by F e s s . 1 7 Suggestions for Further Research Further research is needed to look at the subject variation in grip strength measurements among patient-specif ic population groups. Th is is important to determine the impact of treatment approaches on individuals and on patient speci f ic groups as a whole. A s wel l , the validity, or the extent to which grip strength tests measure what they are intended to measure, should be explored. The appropr iateness, meaningfu lness and usefu lness of grip strength measures requires further investigation before we can accurately use these tests to infer a patient's functional status. Conclusion The most important aspect of a measurement tool for the cl inician is the ability to reliably detect clinically important change. From the literature, and a survey of c l in ic ians, it appears that the Jamar dynamometer is a widely used and accepted clinical measurement tool in rehabilitation. Most cl inicians are familiar with the Standard Gr ip Strength Test and the 5 Posit ion Grip Strength Test and most use grip 87 strength measurements to determine grip strength in hand and upper extremity injured populat ions. Test-retest reliability of two common grip strength tests ( S G S T , 5PGST) w a s substant ial in our study of healthy males. However, measures reported to detect clinically important change were var ied, and it appeared from our study, that cl inicians may report a change in a patient's status when the change may be due to measurement error a lone. A s well , the reliability and variability in patient speci f ic populat ions is not known as many of the reliability studies have been done on healthy subjects. In conc lus ion, knowing the range in which the true score for an individual 's observed test can be expected to vary aids in interpreting the grip strength measurement scores . The use of the standard error of measurement and conf idence intervals greatly aid the therapist 's clinical decis ion making in determining if a true change has occurred. Further research is required to examine the issues of validity with respect to grip strength measures . 88 R E F E R E N C E S 1. C o l e B, F inch F, Gowland C , Mayo N. Physical Rehabilitation Outcome Measures. Toronto, Ont: Canad ian Phys ica l therapy Assoc ia t ion 1994. 2. Smith R O , Benge M W . P inch and grasp strength: Standardizat ion of terminology and protocol. American Journal of Occupational Therapy]985; 39(8):531-535. 3. F e s s E E , Moran C A . Clinical Assessment Recommendations. Garner , North Caro l ina : Amer ican Society of Hand Therapists 1981: 6-8. 4. Bohannon R W . The clinical measurement of strength. Clinical Rehabilitation 1987; 1:5-16. 5. Y o u n g V L , P in P, Kraemer B A , Gou ld R B , Nemergut L, Pel lowski M. Fluctuation in grip and pinch strength among normal subjects. Journal of Hand Surgery (Amer ican volume) 1989; 14:1:125-129. 6. Pe te rsen P, Petrick M, Connor H, Conkl in D. Grip strength and hand dominance: chal lenging the 10% rule. American Journal of Occupational Therapy 1989; 43(7):444-7. 7. Mathiowetz V , W e b e r K, Vo l land G , K a s h m a n N. Reliabil ity and validity of hand strength evaluat ions. Journal of Hand Surgery (Am) 1984; 9:222-226. 8. Bea r -Lehman J , Ab reau B C . Evaluat ing the hand: Issues in reliability and validity. Physical Therapy 1989; 69(12): 1025-1033. 9. Hamil ton A , Ba lnave R, A d a m s R. Grip strength testing reliability. Journal of Hand Therapy 1994; 7(3): 163-70. 10. MacDerm id J C , Kramer J F , Woodbury M G , McFar lane R M , Roth J H . Interrater reliability of pinch and grip strength measurements in clients with cumulat ive t rauma disorders. Journal of Hand Therapy 1994; 7: 10-14. 11. Stratford P W , Normal G R , Mcin tosh J M . General izabi l i ty of grip strength measures in patients with tennis elbow. Physical Therapy 1989; 69 :276-281. 12. Stratford P W . Summar iz ing the results of multiple strength trials: Truth or consequence . Physiotherapy Canada 1992; 44:1:14-18. 13. Amer i can Medica l Assoc ia t ion . Guides to the evaluation of permanent impairment, 4th ed . Ch icago , Illinois:, Amer ican Medica l Assoc ia t ion 1993. 89 14. Binkley J , Stratford P. Measurement of outcome in orthopaedic physical therapy pract ice; A step-by-step approach for cl inicians. Sent inel Assoc ia tes , 1997. 15. Amer i can Phys ica l Therapy Assoc ia t ion 's Task Force on Standards for Measurement in Phys ica l Therapy. Standards for tests and measurements in physical therapy practice. Physical Therapy 1991; 71(8): 589-622. 16. Stratford P W , Goldsmith C H . U s e of the standard error as a reliability index of interest: A n appl ied example using e lbow flexor strength data. Physical Therapy 1997; 77:7:745-750. 17. F e s s E E . A method for checking Jamar dynamometer calibration. Journal of Hand Therapy 1987; (4):28-32. 90 Appendix A GRIP STRENGTH TEST DESCRIPTION Standard Grip Strength Test The standard grip strength test, while not a lways referred to by this name in the literature, in our study consists of 3 grip strength trials at Handle Posi t ion #2 on the J a m a r dynamometer . The subject is posit ioned according to the Amer i can Soc ie ty of Hand Therapists, in sitting with the shoulder adducted and neutrally rotated, the e lbow at 90 degrees and the forearm and wrist in neutral. The subject 's arm is not stabi l ized. T ime between trials was 15 seconds . The left hand w a s tested first, and hands were alternated for testing (left, then right). The subjects were encouraged to give maximal effort by stating this verbally at the start of the testing and by giving minimal verbal encouragement during the testing. 5-Position Grip Strength Test The 5-Posi t ion Grip Strength Test involved similar subject posit ioning as indicated in the Standard Grip Strength Test above. O n e grip strength measurement was taken at each of the 5 handle posit ions on the Jamar dynamometer , starting with the smal lest position (handle position #1) and progressing to the largest (handle posit ion #5). The left hand was tested first, then the right at each posit ion. A 15 second rest between trials was implemented. 91 Appendix B Survey - Grip Strength Measures Please fill out this survey and return it by fax to Jane Burns at (604) 986-8492, or mail it in the self addressed, stamped envelope provided. Thank you for your time. 1. Are you currently using grip strength measurements in your clinical practice? • Yes • No If No, why not? 2. What standard grip strength tests do you currently use in your clinic? | | Standard Grip Strength Test | | 5 Position Grip Strength Test | | Rapid Exchange Grip Strength Test | | Other (please specify) 3. What instrument(s) do you use to measure grip strength? | | Jamar Dynamometer | | Digital Jamar Dynamometer | | Tekdyne Dynamometer • BTE [ | Blood pressure cuff , [TJ Other (please specify) 4. What patient populations are you using grip strength testing with? | | Hand injuries | | Upper extremity injuries | | Back/neck injuries | | Other conditions/uses 5. For what purpose(s) do you use the grip strength tests referred to in Question #2? Which test is best for each purpose? | | Determine grip strength | | Measure function (i.e. ADL) | | Measure impairment (i.e. % loss) | | Assess sincerity of effort 6. Which of the following measures do you use to determine clinically important differences on which to base your treatment decisions (you may check more than one). | | Percentage difference between hands | | Absolute difference between hands | | Percentage difference between trials | | Absolute difference between trials | | Percentage difference between normative database and raw score | | Absolute difference between normative database and raw score | | Other (please specify): 92 7. How large would the difference have to be for you to call it clinically significant? (Indicate the differences for the measures that you checked in the question above only). | | Percentage difference between hands % | | Absolute difference between hands kg/lbs force | | Percentage difference between trials % | | Absolute difference between trials kg/lbs force | | Percentage difference between normative database/raw score % | | Absolute difference between normative database/raw score kg/lbs force | | Other (please specify): 8. Do you use a normative database to compare results to? • Yes • No If yes, which one? 94 Appendix D R E S E A R C H DESIGN Testing Protocol - Day 1 -» Testing Protocol - Day 2 STANDARD GRIP S T R E N G T H T E S T 3 Trials at Posi t ion #2 (15 second rest between trials) 2 MINUTE REST 5 POSITION GRIP S T R E N G T H T E S T 1 trial at each of 5 posit ions (15 second rest between trials) STANDARD GRIP STRENGTH T E S T 3 Trials at Posi t ion #2 (15 second rest between trials) 2 MINUTE REST 5 POSITION GRIP STRENGTH T E S T 1 trial at each of 5 posit ions (15 second rest between trials) •Randomly assigned to S G S T or 5PGST at beginning of study •Test dates within 1 week of each other •Start with left hand first in all tests •Repeated measures, within subject design 95 Appendix E Standardized Subject Instructions 1. The subject will be asked to sign the consent forms and the study will be expla ined to them. Order of testing (Standard Grip Strength Test, 5-Posi t ion Gr ip Strength Test) will be determined by a computer ized random number table at the start of test ing. 2. The J a m a r dynamometer will be shown to the subject. Instructions to the subject will be "We are doing a study to test grip strength. I want you to squeeze this handle as hard as you can until I say stop. I want you to try as hard as you can and give maximal effort. Do you have any questions ?" 3. Both hands will be tested, starting with the left hand. 4. The test procedure involves positioning the subject in sitting with the shoulder adducted and neutrally rotated, the e lbow at 90 degrees and the forearm and wrist in neutral. The subject 's arm will not be stabi l ized. The J a m a r dynamometer is p laced in the palmar gutter of the subject 's hand. The subject cannot s e e the dynamometer dial or v iew the measurements during the study (data col lect ion will be sc reened from the subject 's view). 5. The test will begin by placing the dynamometer in the subject 's hand. Proper p lacement must be maintained in order to give the subject maximal ability to grip the dynamometer . Instructions to the subject will be "Are you ready ? Squeeze as hard as you can.. Squeeze..Squeeze...Squeeze and Stop". The next hand will then be tested in the s a m e manner. The sequence of testing is as fol lows, keeping in mind that the order will be dependent on randomizat ion at the start of the testing sess ion : Standard Grip Strength Test Jamar Handle Position: Position #2 Start: Left hand first Procedure: 3 trials on each hand (alternating) Length of Grip: 3 seconds Rest between Trials: 15 seconds Score: Each of 3 trials and the mean of 3 trials 5 Position Grip Strength Test Jamar Handle Position: Start at Position #1 and progress to Position #5 Start: Left hand Procedure: One trial at each of the 5 handle positions(alternating hands) Length of Grip: 3 seconds Rest Between Trials: 15 seconds Score: 1 trial at each of 5 positions 6. Rest between tests will be 2 minutes as timed by a hand held stopwatch. The s a m e procedure will be repeated on the retest day, within one week of the initial testing. The subject will start with the s a m e test as on day 1. 96 Appendix F SURVEY RESULTS RESPONSES SURVEY QUESTION N=120 n=11 n=109 # All Sites % All Sites #Hand Only % Hand Only # Sites Only % Sites Only CLINICIAN USE Using Grip Strength Measures 113 94 11 100 102 94 Not Using Grip Strength Measures 7 6 0 0 7 6 120 n=113 n=11 n=102 TESTS USED Standard Grip Strength Test 94 83 10 91 84 82 5 Position Grip Strength Test 47 42 7 64 40 39 Rapid Exchange Grip Strength Test 28 25 6 55 22 22 Other 12 11 1 9 11 11 181 INSTRUMENT USED Jamar 98 87 9 82 89 87 Digital Jamar 11 10 0 0 11 11 Tekdyne 0 0 0 0 0 0 BTE 10 9 1 9 9 9 Blood Pressure Cuff 17 15 0 0 17 17 Other 24 21 3 27 21 21 160 PATIENT POPULATION TESTED Hand Injuries 96 85 10 91 86 84 Upper Extremity Injuries 83 73 7 64 76 75 Back/Neck Injuries 35 31 2 18 33 32 Other 36 32 1 9 35 34 250 PURPOSE OF TEST Determine Grip Strength 111 98 10 91 101 99 Measure Function 40 35 3 27 37 36 Measure Impairment 66 58 5 45 61 60 Assess Sincerity of Effort 48 42 8 73 40 39 265 # All Sites % All Sites # Hand Only % Hand Only # Sites Only % Sites Only MEASURES TO DETERMINE CIC % diff between hands 65 58 8 73 57 56 Absolute diff between hands 68 60 5 45 63 62 % diff between trials 31 27 3 27 28 27 Absolute diff between trials 44 39 3 27 41 40 % diff between NDB/score 18 16 2 18 16 16 Absolute diff between NDB/score 21 19 2 18 19 19 Other 10 9 3 27 7 7 257 USE OF NORMATIVE DATABASE Yes 43 39 5 56 38 37 No 69 62 4 44 65 64 No Response 1 2 CO o Mean 4*. 73 m CO > ro ~0 O r -o c _ I CO —\ o CO O O z cn O c n O CD 4*. m 7) CO —i 75 ro I H r— Subject Test 0.697 In _^ o cn 00 CT) C J -4 ro cn -U CD c n o cn 4=> 0 0 cn c n 0 0 cn o c n 0 0 cn cn cn CD 4*. CD Ol CD H CD CA i-f 73 ca' zr »-»• S>' o i3 0.697 o> o ro CT) ro 00 CT) ro cn c n ro c n CD cn ro c n --J cn o c n 4^ cn c n cn o c n CO cn 0 0 c n 4=> ro 73 CD CD CA 0.739 -~j "vi C J ro cn ro C J cn -j - j CO o c n 0 0 c n ro c n ro c n cn c n 4^ . 0 0 4^ c n 0 0 c n CO 4*. 4^ -~J H CD CA f-t-r-CD 0.739 C J to ->j cn cn CT) C J cn c n 4*. c n o c n cn c n CD c n o cn o c n o c n ro c n 4^ c n ro cn ro 4^ cn Ol CO 73 <D CD (A i-+ 0.710 CJ> Ol cn •vi ID ro to - j c n cn 4^ CD 4^ CO cn ro CD cn cn 0 0 c n 0 0 c n ro cn ro cn CD c n o cn cn H CD CA 73 c5' s» 55' Ci *3 ro 0.710 00 o - J . CT) p C J CO cn CO CD CD o cn ro c n ro cn o c n CO O ) CO c n cn c n CO O ) Ol c n 4^ cn CO TJ CD i-+ CD CA p* 0.624 oi 4*. o> cn C J cn o CD ro c n ro c n CO c n ro c n ro c n ro c n ro 4* ro c n 0 0 Ol 0 0 4^ cn 4^ CO H CD CA f-+ I -CD 0.624 iu cn cn cn 00 CT) c n ro c n c n c n c n c n c n c n CD c n cn ro c n cn c n ro cn ro c n o c n Ol 73 CD t-f CD CA i-t-0.431 CT) to O o cn CT) 4*. c n cn cn 4*. a c n o cn o 4^ CO c n CO o c n o o i -r*. cn cn CO c n ro c n H CD (A 73 ca' 3 V) u' G — CO =tt —1 w 0.431 - J In co o cn CO Lo ro to cn CO cn ro c n cn ro cn ro c n -fc* cn ro c n Ol cn ro c n c n o cn c n 4v ro o 73 CD i-t-CD CA f-t-0.717 CT) cn ro cn ro cn cn c n ro ro c n c n --4 c n o c n cn c n CD a> 4^ 0 0 Ol CO Ol co 4^ 4^ c n ro H CD CA (-»• I -CD 0.717 ro to o cn ro oo CT) cn c n co 4*. c n CD c n cn c n ro cn ro c n c n oo 4^ co J5> CO cn ro 4^ o Ol ro 73 CD i-t-CD CA r* 0.704 ro o ro cn CT) I*g 00 o cn o co CD cn CO cn o c n CD c n cn c n c n cn c o cn CO c n o cn H CD CA i-t-73 (•' 3" 3 CD £U = c/> o Ci w H t w' 0.704 CT) io C J C J CT) O b CO c n c n c n ro cn o cn ro c n cn cn CO c n c n 4^ cn cn c n o o 73 CD CD CA i-f 0.800 CT) io to cn ro C J Li. 4*. cn cn cn o c n c n c n 4^ c n c n c n c n c n 4*. CD 4*. c n Ol cn c n 0 0 4^ cn CD H CD CA f-+ I -CD 0.800 CT) lo C J C O cn cn io CO CT) CO c n c n c n c n c n cn o c n c n - J c n CO Ol cn ro 4^ Ol c n c n 73 CD CD CA m co H T J m H m T J > r -TJ m T J > T J m co c CO O -n m m T J O -o co O O d 2 O o 2 < CD m TJ T J ro CO m TJ c> m D > -< CO m co T N O x o CO % I Ol B> TJ 3 O Q . CO CD H TJ O 22. o " 3 % X Ol M 0) TJ 9z CO CD H TJ o <n 5' 3 TJ I £ O ft) TJ 2. = O =r. £ co O CD H 3 CJ TJ I cn O ft) TJ <2, 3 CD =r. £ CO o CD H 3 TJ I Ol O ft) TJ <2. 3 G ) 2 ; OL CO O (P H 3 Ol 100 Appendix G STATISTICAL RESULTS FOR THE STANDARD GRIP STRENGTH TEST Analysis of Variance (Balanced Designs) Analysis of Variance for RIGHT Source DF SS MS F P SUBJECTS 13 3242 536 249 426 * TRIAL 2 64 667 32 333 * DAY 1 116 679 116 679 * SUBJECTS*TRIAL 26 171 000 6 577 0 76 0 755 SUBJECTS*DAY 13 596 155 45 858 5 30 0 000 TRIAL*DAY 2 0 857 0 429 0 05 0 952 Error 26 224 810 8 647 Total 83 4416 702 * No exact F-test can be calculated. Source Variance Error Expected Mean Square component term (using unrestricted model) 1 SUBJECTS 34 .2729 * (7) +3(5) + 2(4) +6(1) 2 TRIAL 1.2134 * (7) + 14 (6) +2(4) + 28 (2) 3 DAY 1.8819 * (7) + 14 (6) + 3(5) + 42 (3) 4 SUBJECTS*TRIAL -1.0348 7 (7) + 2(4) 5 SUBJECTS*DAY 12 .4038 7 (7) + 3(5) 6 TRIAL*DAY -0.5870 7 (7) + 14(6) 7 Error 8.6465 (7) * No exact F-Test can be calculated. Means DAY N RIGHT 1 42 57.762 2 42 60.119 TRIAL N RIGHT 1 28 59.964 2 28 59.036 3 28 57.821 Analysis of Variance (Balanced Designs) - Standard Grip Strength Test Analysis of Variance for LEFT Source DF SS MS F P SUBJECTS 13 2989 619 229 971 * TRIAL 2 5 357 2 679 * DAY 1 213 762 213 762 * SUBJECTS*TRIAL 26 244 310 9 397 1 16 0 352 SUBJECTS*DAY 13 386 571 29 736 3 68 0 002 TRIAL*DAY 2 31 595 15 798 1 96 0 162 Error 26 210 071 8 080 Total 83 4081 286 * No exact F-test can be calculated. Source Variance Error Expected Mean Square component term (using unrestricted model) 1 SUBJECTS 33.1529 * (7) + 3(5) + 2(4) +6 (1) 2 TRIAL -0 . 5156 * (7) + 14 (6) +2(4) +28(2) 3 DAY 4.1978 * (7) + 14 (6) + 3(5) + 42 (3) 4 SUBJECTS*TRIAL 0 . 6584 7 (7) + 2(4) 5 SUBJECTS*DAY 7 .2189 7 (7) + 3(5) 6 TRIAL*DAY 0.5513 7 (7) + 14(6) 7 Error 8.0797 (7) * No exact F-Test can be calculated. Means DAY N LEFT 1 42 52.048 2 42 55.238 TRIAL N LEFT 1 28 54.000 2 28 53.464 3 28 53.464 Analysis of Variance (Balanced Designs) - Standard Grip Strength Test Analysis of Variance for RIGHT Source DF SS MS F P SUBJECTS 13 3242 54 249 43 5 44 0 002 DAY 1 116 68 116 68 2 54 0 135 SUBJECTS*DAY 13 596 15 45 86 5 57 0 000 Error 56 461 33 8 24 Total 83 4416 70 Source Variance Error Expected Mean Square component term (using unrestricted 1 SUBJECTS 33 . 928 3 (4) + 3 (3) +6 (1) 2 DAY 1. 686 3 (4) + 3 (3) + 42 (2) 3 SUBJECTS*DAY 12 . 540 4 (4) + 3(3) 4 Error 8 .238 (4) Means DAY N RIGHT 1 42 57.762 2 42 60.119 Analysis of Variance (Balanced Designs) - Standard Grip Strength Test Analysis of Variance for LEFT Source DF SS MS F ,P SUBJECTS 13 2989 62 . 229 97 7 73 0 000 DAY 1 213 76 213 76 7 19 0 019 SUBJECTS*DAY 13 386 57 29 74 3 39 0 001 Error 56 491 33 8 77 Total 83 4081 29 Source Variance Error Expected Mean Square component term (using unrestricted 1 SUBJECTS 33 .372 3 (4) +3(3) +6 (1) 2 DAY 4 .382 3 (4) +3(3) + 42 (2) 3 SUBJECTS *DAY 6.987 4 (4) + 3(3) 4 Error 8 . 774 (4) Means DAY N LEFT 1 42 52.048 2 42 55.238 Analysis of Variance- Standard Grip Strength Test Analysis of Variance for RLDIFF Source DF SS MS F P SUBJECTS 13 1140 06 87 70 2 90 0 033 DAY 1 14 58 14 58 0 48 0 500 SUBJECTS*DAY 13 393 58 30 28 2 53 0 008 Error 56 669 33 11 95 Total 83 2217 56 Source 1 SUBJECTS 2 DAY 3 SUBJECTS*DAY 4 Error Variance Error Expected Mean Square component term (using unrestricted model) 9.5702 3 (4) + 3(3) + 6(1) -0.3736 3 (4) + 3(3) + 42 (2) 6.1078 4 (4) +3(3) 11.9524 (4) 104 STATISTICAL RESULTS OF THE 5 POSITION GRIP STRENGTH TEST Variance Components Estimation - 5PGST Position 1 Left ANOVA Type III Sum of Mean Source Squares df Square Corrected Model 1426.571 14 101.898 Intercept 41195.571 1 41195.571 S U B J E C T 1421.429 13 109.341 TIME 5.143 1 5.143 Error 101.857 13 7.835 Total 42724.000 28 Corrected Total 1528.429 27 Dependent Variable: G 1 L Variance Estimates Component Estimate Var(fclUb!JtlJI) Var(TIME) Var(Error) 50.753 - .192 a 7.835 Dependent Variable: G 1 L Method: A N O V A (Type III Sum of Squares) a - For the A N O V A and M I N Q U E methods, negative variance component estimates may occur. Some possible reasons for their occurrence are: (a) the specified model is not the correct model, or (b) the true value of the variance equals zero. Variance Components Estimation - 5PGST Position 1 Right ANOVA Type III Sum of Mean Source Squares df Square (Corrected Model 1454.714 14 103.908 Intercept 47726.286 1 47726.286 SUBJECT 1447.714 13 111.363 TIME 7.000 1 7.000 Error 383.000 13 29.462 Total 49564.000 28 Corrected Total 1837.714 27 Dependent Variable: G1R Variance Estimates Component Estimate Var(t!Llb!JbCI) Var(TIME) Var(Error) 40.55-1 -1.604a 29.462 Dependent Variable: G1R Method: ANOVA (Type III Sum of Squares) a For the ANOVA and MINQUE methods, negative variance component estimates may occur. Some possible reasons for their occurrence are: (a) the specified model is not the correct model, or (b) the true value of the variance equals zero. Variance Components Estimation - 5PGST Position 2 Left ANOVA Type III Sum of Mean Source Squares df Square corrected Model 1189.071 14 84.934 Intercept 82840.321 1 82840.321 S U B J E C T 1170.179 13 90.014 TIME 18.893 1 18.893 Error 95.607 13 7.354 Total 84125.000 28 Corrected Total 1284.679 27 Dependent Variable: G 2 L Variance Estimates Component Estimate Var(SUb!Jb(J I) Var(TIME) Var(Error) 41.330 .824 7.354 Dependent Variable: G 2 L Method: A N O V A (Type III Sum of Squares) Variance Components Estimation - 5PGST Position 2 Right ANOVA Type III Sum of Mean Source Squares df Square Corrected Model 1158.000 14 82.714 Intercept 100320.6 1 100320.6 SUBJECT 1146.429 13 88.187 TIME 11.571 1 11.571 Error 249.429 13 19.187 Total 101728.0 28 Corrected Total 1407.429 27 Dependent Variable: G2R Variance Estimates Component Estimate Var(yUb!Jb(JI) Var(TIME) Var(Error) 34.500 -.544a 19.187 Dependent Variable: G2R Method: ANOVA (Type III Sum of Squares) a- For the ANOVA and MINQUE methods, negative variance component estimates may occur. Some possible reasons for their occurrence are: (a) the specified model is not the correct model, or (b) the true value of the variance equals zero. Variance Components Estimation - 5PGST Position 3 Left ANOVA Type III Sum of Mean Source Squares df Square corrected Model 1914.000 14 136.714 Intercept 79929.143 1 79929.143 S U B J E C T 1908.857 13 146.835 TIME 5.143 1 5.143 Error 332.857 13 25.604 Total 82176.000 28 Corrected Total 2246.857 27 Dependent Variable: G 3 L Variance Estimates Component Estimate Var(SUBJE(J I) Var(TIME) Var(Error) 60.615 -1 .462 s 25.604 Dependent Variable: G 3 L Method: A N O V A (Type III Sum of Squares) a For the A N O V A and M I N Q U E methods, negative variance component estimates may occur. Some possible reasons for their occurrence are: (a) the specified model is not the correct model, or (b) the true value of the variance equals zero. Variance Components Estimation - 5PGST Position 3 Right ANOVA Type III Sum of Mean Source Squares df Square corrected Model 987.143 14 70.510 Intercept 93960.143 1 93960.143 S U B J E C T 972.857 13 74.835 T IME 14.286 1 14.286 Error 170.714 13 13.132 Total 95118.000 28 Corrected Total 1157.857 27 Dependent Variable: G 3 R Variance Estimates Component Estimate Var (SUBJhCI) Var(TIME) Var(Error) 30.852 8.242E-02 13.132 Dependent Variable: G 3 R Method: A N O V A (Type III Sum of Squares) Variance Components Estimation - 5PGST Position 4 Left ANOVA Type III Sum of Mean Source Squares df Square corrected Model 1103.071 14 78.791 Intercept 66543.750 1 66543.750 S U B J E C T 1092.750 13 84.058 TIME 10.321 1 10.321 Error 124.179 13 9.552 Total 67771.000 28 Corrected Total 1227.250 27 Dependent Variable: G 4 L Variance Estimates Component Estimate Var(SUBJbCI) Var(TIME) Var(Error) 37.253 5.495E-02 9.552 Dependent Variable: G 4 L Method: A N O V A (Type III Sum of Squares) Variance Components Estimation - 5PGST Position 4 Right ANOVA Type III Sum of Mean Source Squares df Square Corrected Model 1159.357 14 82.811 Intercept 74366.036 1 74366.036 SUBJECT 1146.464 13 88.190 TIME 12.893 1 12.893 Error 139.607 13 10.739 Total 75665.000 28 Corrected Total 1298.964 27 Dependent Variable: G4R Variance Estimates Component Estimate Var(SUUJbCI) Var(TIME) Var(Error) 38.725 .154 10.739 Dependent Variable: G4R Method: ANOVA (Type III Sum of Squares) Variance Components Estimation - 5PGST Position 5 Left ANOVA Type III Sum of Mean Source Squares df Square Corrected Model 1243.000 14 88.786 Intercept 47726.286 1 47726.286 S U B J E C T 1228.714 13 94.516 TIME 14.286 1 14.286 Error 140.714 13 10.824 Total 49110.000 28 Corrected Total 1383.714 27 Dependent Variable: G 5 L Variance Estimates Component Estimate V a r ( y U b i J b u l ) Var(TIME) Var(Error) 41.846 .247 10.824 Dependent Variable: G 5 L Method: A N O V A (Type III Sum of Squares) Variance Components Estimation - 5PGST Position 5 Right ANOVA Type III Sum of Mean Source Squares df Square corrected Model 530.500 14 37.893 Intercept 56430.321 1 56430.321 SUBJECT 526.179 13 40.475 TIME 4.321 1 4.321 Error 86.179 13 6.629 Total 57047.000 28 Corrected Total 616.679 27 Dependent Variable: G5R Variance Estimates Component Estimate Var(SUb!JbCI) Var(TIME) Var(Error) 15.523 -.165a 6.629 Dependent Variable: G5R Method: ANOVA (Type III Sum of Squares) a- For the ANOVA and MINQUE methods, negative variance component estimates may occur. Some possible reasons for their occurrence are: (a) the specified model is not the correct model, or (b) the true value of the variance equals zero. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0099247/manifest

Comment

Related Items