of Effects Negatively Sloped Keyboard Wedges on User Performance and Perceptions
By Mitchell A. Woods
Thesis submitted to the faculty of Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of Master of Science In Industrial and Systems Engineering
Dr. Kari Babski-Reeves, Chair Dr. Maury A. Nussbaum Dr. Tonya Smith-Jackson
December 20, 2002 Blacksburg, VA
Keywords: CTS, negative slope, wrist posture, electromyography, key strike force, discomfort
Effects of negatively sloped keyboard wedges on user performance and perceptions Mitchell A. Woods (ABSTRACT) Of the studies that considered negatively sloped keyboards, results showed improved comfort and postural effects while typing on keyboards; however, few studies of negatively sloped keyboard angles and their resulting effects on objective physiological measures, psychological measures, and performance have been performed. The objective of this study was to quantify the effects of negative keyboard slopes on forearm muscle activity, wrist posture, key strike force, perceived discomfort, and performance to identify a negative keyboard angle or range of keyboard angles that minimizes exposure to hypothesized risk factors for hand/wrist work related musculoskeletal disorders. Ten experienced typists (4 males and 6 females) participated in a laboratory study to compare keyboard slopes ranging from 7 to 30, at 10 increments from 0 to 30, using an experimental wedge designed for use with QWERTY keyboards. Repeatability was examined by requiring participants to complete the experiment in two test sessions one week apart. Dependent variable data was collected during 10 minute test sessions. Wrist posture data revealed postural benefits for negative angles of 0 or greater compared to 7. Specifically, the percentage of wrist movements within a neutral zone and percentage of wrist movements within 5 and 10 degrees increased as keyboard angle became more negative. EMG results were mixed with some variables supporting negative keyboard angles, while other results favored the standard keyboard configuration. Net typing speed supported the -10 keyboard angle, while other negative typing angles were comparable, if not better, than the standard. These findings showed that there was strong support for improved postural changes associated with negatively sloped keyboard wedges, though user perceptions favored the standard configuration.
Authors Acknowledgements First and foremost, I want to thank Dr. Kari Babski-Reeves for her rigorous help, honesty, and continuing devotion to my thesis project. I want to thank my committee members, Dr. Maury Nussbaum and Dr. Tonya Smith-Jackson, for their advice and contributions in making this a successful and rewarding project. This undertaking would not have been possible without their full support. I am sincerely grateful to all my committee members for deciding to work with me on this project, and I hope they enjoyed it as much as I did. I also would like to thank my family and friends for their constant support throughout my graduate studies. If at first I did not succeed, everyone made me try, try again, and I am indebted to all. This completed work would not have been possible without them.
iii
Table of Contents INTRODUCTION............................................................................................................. 1 BACKGROUND................................................................................................................ 1 STATEMENT OF THE PROBLEM......................................................................................... 3 OBJECTIVE OF THE STUDY .............................................................................................. 3 SCOPE AND LIMITATIONS OF THE STUDY .......................................................................... 4 LITERATURE REVIEW .................................................................................................. 6 EFFECTS OF KEYBOARD ANGLE ON HAND AND WRIST POSTURE DURING TYPING TASKS...... 6 INTERRELATION OF KEYBOARD ANGLE, KEY STRIKE FORCE, AND CARPAL TUNNEL PRESSURE .................................................................................................................... 8 EFFECT OF KEYBOARD ANGLE ON RECORDED MUSCLE ACTIVITY DURING TYPING TASKS... 10 EFFECTS OF KEYBOARD HEIGHT ON OPERATOR DISCOMFORT DURING TYPING TASKS....... 11 EFFECTS OF WRIST REST USAGE DURING TYPING TASKS................................................ 11 TYPING TASK PERFORMANCE ON THE NEGATIVE SLOPED KEYBOARD............................... 13 SUMMARY.................................................................................................................... 13 METHODOLOGY.......................................................................................................... 15 EXPERIMENTAL DESIGN ................................................................................................ 15 DEPENDENT VARIABLES................................................................................................ 15 Electrogoniometric Measurements......................................................................... 16 Electromyography (EMG) Measurements .............................................................. 17 Key Strike Force Measurements ............................................................................ 19 Performance .......................................................................................................... 25 Self-reported Postural Discomfort Questionnaire (SPDQ) ..................................... 25 INDEPENDENT VARIABLE ............................................................................................... 26 TYPING TASK DESCRIPTION .......................................................................................... 28 PARTICIPANTS ............................................................................................................. 28 PROCEDURE ................................................................................................................ 29 STATISTICAL DATA ANALYSIS ........................................................................................ 30 One-way ANOVA ................................................................................................... 31 Linear Correlations by Day..................................................................................... 31 Body Part Groupings.............................................................................................. 32 RESULTS ..................................................................................................................... 33 ELECTROGONIOMETRIC DATA........................................................................................ 36 Day......................................................................................................................... 36 Keyboard Angle...................................................................................................... 36 Left Hand ............................................................................................................ 36 Right Hand ......................................................................................................... 37 KA by Gender Interaction....................................................................................... 40 Side........................................................................................................................ 41 ELECTROMYOGRAPHY DATA ......................................................................................... 41 Day......................................................................................................................... 41 Keyboard Angle and Gender.................................................................................. 42 iv
Side........................................................................................................................ 42 KEY STRIKE FORCE DATA ............................................................................................. 43 Side........................................................................................................................ 43 PERFORMANCE DATA ................................................................................................... 43 Day......................................................................................................................... 43 Keyboard Angle, Gender, and KA by Gender Interaction....................................... 43 SELF-REPORTED POSTURAL DISCOMFORT QUESTIONNAIRE DATA.................................... 44 Self-reported Postural Discomfort Questionnaire (7-point scale) ........................... 44 Day......................................................................................................................... 44 Keyboard Angle, Gender, and KA by Gender Interaction....................................... 44 Exit Questions ........................................................................................................ 46 Self-rating Comparison (5-point scale) ............................................................... 46 Estimated Time for Typing Speed Comparison .................................................. 47 Final Ranking...................................................................................................... 49 Keyboard Angle and Gender .............................................................................. 49 LINEAR CORRELATIONS BY DAY..................................................................................... 50 DISCUSSION................................................................................................................ 52 WRIST POSTURE.......................................................................................................... 52 PERFORMANCE MEASURES COMPARED WITH DISCOMFORT............................................. 56 ELECTROMYOGRAPHY .................................................................................................. 61 KEY STRIKE FORCE...................................................................................................... 63 REPEATABILITY OF RESULTS ......................................................................................... 64 WRIST POSTURE PLOTS RELATED TO RISK .................................................................... 65 CONCLUSION .............................................................................................................. 66 LIMITATIONS ................................................................................................................ 66 Maximum Voluntary Contractions .......................................................................... 66 Electrode Placement .............................................................................................. 67 Typing posture ....................................................................................................... 67 Keyboard Wedge Design.................................................................................... 67 Computer Workstation........................................................................................ 67 Wrist Rest ........................................................................................................... 68 Load Button Placement.......................................................................................... 68 Participants ............................................................................................................ 69 Discomfort Variable................................................................................................ 69 FUTURE RESEARCH...................................................................................................... 69 REFERENCES.............................................................................................................. 71 APPENDIX A ................................................................................................................ 76 PHALENS TEST ............................................................................................................ 76 APPENDIX B ................................................................................................................ 77 TINELS TEST ............................................................................................................... 77 APPENDIX C ................................................................................................................ 78 INFORMED CONSENT FOR PARTICIPANTS IN RESEARCH PROJECTS .................................. 78 v
SIGNATURE PAGE ........................................................................................................ 82 APPENDIX D ................................................................................................................ 83 NORDIC QUESTIONNAIRE .............................................................................................. 83 PAST MEDICAL HISTORY............................................................................................... 84 KEYBOARD USAGE ....................................................................................................... 84 CRITERION QUESTIONNAIRE.......................................................................................... 85 DEMOGRAPHIC INFORMATION........................................................................................ 85 ANTHROPOMETRIC DATA ............................................................................................... 86 APPENDIX E ................................................................................................................ 87 SEAT MEASUREMENTS ................................................................................................. 87 SELF-REPORTED POSTURAL DISCOMFORT QUESTIONNAIRE WITH EXIT QUESTIONS.............. 88 APPENDIX F................................................................................................................. 91 PARTICIPANT DEMOGRAPHIC INFORMATION.................................................................... 91 APPENDIX G ................................................................................................................ 92 SHAPIRO-WILKS NORMALITY TEST RESULTS ................................................................. 92 APPENDIX H ................................................................................................................ 93 TABLE OF F-RATIOS, P-VALUES, AND TUKEY HSD Q-VALUES FOR SIGNIFICANT DIFFERENCES .............................................................................................................. 93 APPENDIX I.................................................................................................................. 97 30-SECOND MEAN %MVC GRAPHS .............................................................................. 97 APPENDIX J............................................................................................................... 101 DESCRIPTIVE STATISTICS FOR SIGNIFICANT AND NON-SIGNIFICANT VARIABLES BY KEYBOARD ANGLE ....................................................................................................................... 101 Electrogoniometric Data....................................................................................... 101 Electromyography Data........................................................................................ 103 Key Strike Force Data .......................................................................................... 104 Performance Data ................................................................................................ 104 VITA ............................................................................................................................ 108
vi
Table of Figures Figure I.1. Number of injury cases reported in the wrist(s) for CTS resulting in lost workdays (data from BLS, 2001) ..................................................................................... 1 Figure I.2. Percentage of reported wrist and CTS cases of the UE resulting in lost work days (data from BLS, 2001) ............................................................................................ 2 Figure II.1. Negatively sloped keyboard with a separate wrist rest................................ 13 Figure II.2. Negatively sloped keyboard with an integrated wrist rest............................ 13 Figure III.1. Load button location inside keyboard......................................................... 20 Figure III.2. Comparison of an E key strike force at 1024 Hz....................................... 23 Figure III.3. Comparison of an E key strike force at 512 Hz......................................... 23 Figure III.4. Comparison of an E key strike force at 256 Hz......................................... 24 Figure III.5. Means of key strike force peaks compared at varying frequencies ............ 24 Figure III.6. Percentage decrease in estimation of mean key strike force peaks........... 25 Figure III.7. Keyboard slope illustration ......................................................................... 27 Figure III.8. Side view of experimental keyboard wedge (0) ........................................ 27 Figure III.9. Top view of experimental keyboard wedge (0) ......................................... 27 Figure III.10. Isometric view of experimental keyboard wedge (0) ............................... 28 Figure IV.1. Mean FE left hand deviation KA by gender interaction .............................. 41 Figure IV.2. Day 1 mean net typing speed KA by gender interaction ............................ 43 Figure IV.3. Normalized right wrist KA by gender interaction ........................................ 45 Figure IV.4. Day 2 normalized right hand KA by gender interaction.............................. 45 Figure IV.5. Day 2 normalized left hand KA by gender interaction ................................ 46 Figure IV.6. Mean self-rating of new keyboard angles compared to 7 keyboard angle (n=20)............................................................................................................................ 47 Figure IV.7. Mean time estimated to reach same typing speed compared to 7 keyboard angle by day (n=10) ...................................................................................................... 48 Figure V.1. Comparison of mean FE deviations within varying zones from neutral position.......................................................................................................................... 53 Figure V.2. Mean net typing speed by keyboard angle ................................................. 58 Figure V.3. Mean values for number of errors by keyboard angle................................. 59
vii
Figure V.4. Mean values for missing words by keyboard angle .................................... 59 Figure V.5. Mean values for misspelled words by keyboard angle................................ 60 Figure V.6. Mean EMG value results comparison by muscle and day .......................... 62 Figure V.7. Mean KSF value results comparison by side .............................................. 64 Figure V.8. Wrist posture related to risk, estimated by CTP (mmHg) ............................ 65
viii
Table of Tables Table II.1. Mean carpal tunnel pressures in mmHg for varying fingertip force of 15 healthy participants (data from Rempel et al., 1997) ....................................................... 9 Table III.1. Balanced Latin square order of exposure to experimental keyboard........... 15 Table IV.1. Table of results ........................................................................................... 34 Table IV.2. Table of results (continued)......................................................................... 35 Table IV.3. Table of results (continued)......................................................................... 35 Table IV.4. Left hand post-hoc differences for keyboard angle ..................................... 38 Table IV.5. Right hand post-hoc differences for keyboard angle................................... 40 Table IV.6. EMG post-hoc differences for keyboard angle ............................................ 42 Table IV.7. Net typing speed post-hoc differences for keyboard angle ......................... 44 Table IV.8. Self-rating post-hoc differences for keyboard angle.................................... 47 Table IV.9. Minimum and maximum values for mean self-rating of new keyboard angles compared to 7 keyboard angle .................................................................................... 47 Table IV.10. Means and standard deviations for estimated time of typing speed.......... 49 Table IV.11. Final ranking means and post-hoc differences for keyboard angle (best to worst) ............................................................................................................................ 50 Table IV.12. Table of linear correlations by day ............................................................ 51 Table V.1. Mean net typing speed ................................................................................ 58
ix
Table of Equations Equation III.1. Equation for normalizing electromyography data .................................. 18 Equation IV.1. Equation for normalizing discomfort rating data.................................... 32
x
CHAPTER I INTRODUCTION Background The estimated costs associated with work related musculoskeletal disorders (WMSDs) range from 13 to 54 billion dollars annually (US Dept. of Health and Human Services, 2001). Research in the area of computer workstation design, most notably keyboard design and placement, has been driven by the growing number of cases of carpal tunnel syndrome (CTS) and similar WMSDs associated with their use. In 1999, there were 27,922 new CTS cases, an increase from 26,266 reported cases in 1998 (Figure I.1) (BLS, 2001). Repetitive motion injuries of the wrist have accounted for over 67% of the total number of repetitive motion injury cases for the upper extremities since 1992 (Figure I.2) (BLS, 2001). CTS cases have consistently accounted for over 50% of all repetitive motion injuries reported in the upper extremities for the same time span (BLS, 2001).
Repetitive motion cases reported in the wrist(s) 60000 50000 No. of Injury 40000 Cases 30000 20000 1992 1993 1994 1995 1996 Year 1997 1998 1999 CTS cases reported in UE
Figure I.1. Number of injury cases reported in the wrist(s) for CTS resulting in lost workdays (data from BLS, 2001)
1
% of wrist(s) exposure for repetitive motion cases in UE 80% 70% % of Injury 60% Cases 50% 40% 1992 1993 1994 1995 Year 1996
% of CTS exposure in UE
1997
1998
1999
Figure I.2. Percentage of reported wrist and CTS cases of the UE resulting in lost work days (data from BLS, 2001) Keyboard design research has focused on ulnar and radial wrist deviation, wrist extension, wrist support, forearm support, and shoulder and neck effects while using conventional keyboards, fixed and adjustable split-angle keyboards, natural or Kinesis keyboards, or more radical keyboard designs, such as the chair-mounted split keyboard and open or TONY keyboard. Wrist extension, or positively angled wrist and hand orientation, has become standard when discussing keyboard angles. This fact is eminent when considering the design of the conventional keyboardeach row, from bottom to top, is consecutively higher, resulting in a positively sloped keyboard face. This design forces keyboard users to conduct typing tasks predominantly with wrists, hands, and fingers extended. Only in the past decade has wrist flexion become a consideration for keyboard design in contrast to the accepted standard, said to be between 0 and a positive 25 slope (ANSI/HFS 100, 1988). Few keyboards or keyboard apparatuses have features that achieve a zero- or negative-degree slope. The apparent advantage of a negatively sloped keyboard is to decrease the positive sloping of conventional keyboards, thus reducing wrist deviation, primarily wrist extension. Reducing wrist extension also minimizes static muscle exertions required to hold the hands over the keyboard during and following typing, since many users float their hands over the keyboard between bursts of typing.
2
Statement of the Problem Research has been conducted by Hedge (Hedge and Morimoto, 2001; Hedge et al., 1999; Hedge and Powers, 1995; Hedge et al., 1995; Hedge, 1994) using a keyboard tray that achieves a negative slope, called a negatively sloped keyboard support (NSKS) system or a downward-tilting (DT) keyboard tray. Hedge and Powers (1995) found that wrist extension was reduced significantly and none of the participants using the NSKS system extended their hands as severely as participants that used the conventional keyboard configuration. Hedge (1994) found that all workers in a field test of the NSKS system reported less discomfort during keying and 62.5% reported no pain or discomfort at all; thereby concluding the NSKS system reduced postural risks for all workers. Another study by Hedge et al. (1999) compared the DT keyboard tray system to two other conventional positively angled keyboard configurations using electrogoniometric wrist angle measurements. They found that when using the DT system, 67% of the typing movements were made with the wrist in a neutral zone (<15 extension, <30 flexion, <15 ulnar or radial deviation) compared to 42% for the conventional keyboard tray or desk. Gilad and Harel (2000) recommended that a negative tilt of 10 in the sagittal plane might possibly improve the standard keyboard configuration. Although not documented, the negative design Gilad and Harel studied was subjectively evaluated as the most comfortable as compared to the flat design (standard keyboard configuration), Tony TM, and apart (split) designs. These studies provided compelling evidence to support research regarding negatively sloped keyboard systems for reducing musculoskeletal stresses. However, few negative slope angles and their resulting effects on objective physiological measures (such as muscle activity, wrist posture, etc.), psychological measures (such as perceived discomfort), and performance have been studied. Objective of the Study The objective of this study was to quantify the effects of various negative keyboard slopes on forearm muscle activity, wrist posture, key strike force, perceived discomfort, and performance to identify a negative keyboard angle or range of keyboard
3
angles that minimize exposure to hypothesized risk factors for hand and wrist WMSDs. It was also sought to determine if results were repeatable. A laboratory study was conducted to compare keyboard slopes ranging from 7 to 30, at 10 increments from 0 to 30, using an experimental wedge designed for use with current QWERTY keyboards. It was hypothesized that keyboard angle would affect muscle activity, wrist posture, within a neutral zone, key strike forces, and reported discomfort in the forearms, wrists, and hands; while performance would remain consistent. It was also hypothesized that keyboard angles -30 would result in an increase in exposure to hypothesized risk factors for WMSDs of the hand/wrist (i.e. increased muscle activity, deviated wrist postures, etc.), increase reported discomfort and performance would decrease due to inability to view the keyboard and reach keys located on the periphery of the keyboard. Scope and Limitations of the Study The scope of this study was limited to assessing the effects of different negative keyboard angles on muscle activity, wrist posture, key strike force, reported discomfort, and performance as compared to the standard QWERTY keyboard configuration (7 slope). Other factors, such as keyboard height, chair design parameters, and use of arm and wrist rests were not investigated. Research showed that these factors can influence upper extremity fatigue and discomfort (Duncan and Ferguson, 1974; Arndt, 1983; Andersson, 1987; Carter and Banister, 1994; Hedge and Powers, 1995; Albin, 1997; Marklin et al., 1997(a, b); Paul et al., 1996; Smith et al., 1998; Gilad and Harel, 2000). However, few negative keyboard angles have been investigated, and the range of keyboard angles minimizing exposure to WMSD risk factors need to be identified first. Personality type and mental stress have been documented to affect muscle activation (Glasscock et al., 1999). These factors were not considered since the focus of this study was on the biomechanical aspects of typing, not psychophysical assessments. Additionally, studies quantifying the effect of psychophysical factors on the muscles of the forearm and hand were limited.
4
This study also focused on a single typing task. Other peripheral tasks, such as filing, writing, and other activities typically associated with typing positions were not evaluated. Further, the use of other input devices, such as the mouse, was not investigated.
5
CHAPTER II LITERATURE REVIEW Effects of Keyboard Angle on Hand and Wrist Posture during Typing Tasks A review of the literature revealed few articles investigating wrist flexion and the concept of negatively sloped keyboard systems. Stack (1987, 1988(a, b)) reported in an Australian field study that a negatively sloped keyboard designslanting the keyboard away from the user resulting in leveling the angle of the wrists and fingers has been a major improvement in addressing CTS problems in the Tasmanian public service. Hedge (Hedge and Morimoto, 2001; Hedge et al., 1999; Hedge and Powers, 1995; Hedge et al., 1995; Hedge, 1994) has performed multiple studies on the NSKS system and DT keyboard platforms. Hedge and Powers (1995) and Powers and Hedge (1992) found the NSKS system significantly reduced wrist extension and participants did not report any negative reactions to using the NSKS system. Hedge et al. (1995) studied a preset tilt down (PT) system, similar to the NSKS system, and reported that 96% of users reported easy adjustability. Participants also indicated that the system and integrated palm rest resulted in comfortable typing positions. While using the PT system, 62.2% of typing movements were made with less than 20 of wrist extension, where 5.4% of wrist extension movements exceeded 25, compared to 38.8% and 32.6% of wrist movements, respectively, for standard keyboard designs. After three weeks of field use, the PT system produced a number of postural benefits, such as increased neutral zone (10 wrist flexion/extension) hand movements, reduced upper limb musculoskeletal discomfort as seen in self-ratings, and overall body posture improvement for seated work. DeKrom et al. (1990) found that the risk of CTS increased for wrist flexion OR extension, but if the wrist worked in combination of flexion and extension, there was no significant increased risk ratio. Hedge and Powers (1995) concluded that the reduction in wrist extension from use with their NSKS should lessen the risks of developing CTS. Recently, Simoneau and Marklin (2001) analyzed wrist extension at five keyboard angles (+15, +7.5, 0, -7.5, and -15) and at different keyboard heights. As
6
the keyboard slope was changed from +15 to -15, wrist extension decreased approximately 13. It could be said that mean wrist extension decreased approximately 1 for every 2 of change in downward slope. The average negative slope angle participants selected for the keyboard platform was -12, resulting in a keyboard slope close to 0. Even in situations where the wrist is extended, lower wrist extension angles associated with negative keyboard slopes are theoretically beneficial with respect to etiology. Gilad and Harel (2000) found that the negative angle of the keyboard they studied provided a more natural positioning of the hand while keying and decreased muscle strain in the arms as measured by EMG. When using a negatively sloped keyboard at approximately 10o, the hands moved within a neutral zone (<15 extension, <30 flexion, <15 ulnar or radial deviation) 67% of the time as compared to 42% of the time for other keyboard arrangements (Hedge and Morimoto, 2001). Rempel et al. (1995) found that when the angle of the keyboard was sloped -2.6 at the home row, wrist extension, ulnar deviation, and forearm pronation were closer to a neutral position. Consequently, sloping the keyboard negatively also resulted in increased ulnar deviation of both wrists, which may diminish its benefits. However, Marras and Schoenmarklin (1991) and Schoenmarklin and Marras (1993) have stated that wrist extension and flexion pose a greater risk of injury for CTS than does radial or ulnar deviation. Gilad and Harel (2000) also suggested that wrist strain associated with keyboards was primarily due to wrist extension, not ulnar deviation. These improvements in achieving a neutral posture are supported by Rose (1991) who recorded from a compilation of participants a completely relaxed posture of the forearm and hand, which showed the fingers curled greatly inwards toward the palm. In this relaxed state, the finger posture was extremely flexed. Rose found that operating a conventional keyboard required the fingers to be extended rather than flexed, forcing the fingers to work at approximately 75% of their full range of motion Rose described. Typing at such angles prohibits the fingers from operating efficiently. A negatively sloped keyboard may allow the hand and fingers to assume a posture closely resembling the relaxed posture Rose documented. Neutral wrist postures are of
7
benefit also when considering pathophysiological mechanisms of CTS, such as carpal tunnel pressure. Interrelation of Keyboard Angle, Key Strike Force, and Carpal Tunnel Pressure Increased carpal tunnel pressure (CTP) has been implicated as a causal factor for CTS by resulting in compressive forces on the median nerve from surrounding tissues. Gelberman et al. (1984) found that fluids in the palm of the hand flow freely with the hands in neutral-to-moderate extension (<20) or flexion (<20). Weiss et al. (1995) found that extreme extension and flexion of the wrist elevated CTP. It has been shown that brief exposure to a CTP of 30 mm Hg in animals is sufficient to affect nerve functioning for prolonged periods of time (Lundborg et al., 1983). Rempel et al. (1997) found that CTP levels approached this critical value (30 mm Hg) when passively extending the wrist 30 (Table II.1). For discrete, passive wrist movements, mean CTP values were lowest at 15 of flexion, and increased as the wrist was flexed or extended around this value. However, CTP values for wrist flexion were less than corresponding values for wrist extension. This finding supports that of Brain et al. (1947), who found wrist extension raised CTP more so than wrist flexion in cadavers. Further, fingertip forces significantly affect mean CTP values. Under the conditions of Rempel et al.s (1997) experiment, finger force had a greater effect on CTP than did wrist angle. Rempel et al. (1997) found the relationship between CTP and fingertip load could be estimed with a second-order polynomial for the neutral posture.
8
Table II.1. Mean carpal tunnel pressures in mmHg for varying fingertip force of 15 healthy participants (data from Rempel et al., 1997)
Wrist angle 45 of extension 30 of extension 15 of extension Neutral 15 of flexion 30 of flexion 45 of flexion Fingertip force (N) 0 36.0 4.3 27.7 4.9 18.5 3.9 19.7 3.0 16.9 2.8 18.8 4.2 26.6 3.6 6 63.9 8.0 53.5 5.6 41.1 5.6 44.6 6.4 42.4 9.5 32.9 5.5 41.5 5.4 9 70.8 10.4 61.6 7.1 50.8 6.4 53.1 6.1 51.4 9.7 41.4 5.9 44.0 6.8 12 73.7 10.8 62.3 6.8 52.7 6.2 56.5 7.9 54.5 10.0 43.7 7.1 46.3 8.4
Studies have also identified postures (flexion/extension, radial/ulnar deviation, forearm pronation/supination) associated with the lowest CTP values. Weiss et al. (1995) found that the average position of the wrist for the lowest CTP was 2 9 of extension and 2 6 of ulnar deviation. Similarly, Rempel et al. (1992) found CTP to be lowest with the hand in 3.5 5.9 of wrist flexion, 5.0 7.1 ulnar deviation, and 45 of metacarpophalangeal flexion. When considering forearm pronation/supination, Hedge (1994) found CTP was lowest with the hand in 2 of flexion, 3 of ulnar deviation, and pronated. Gilad and Harel (2000) stated that CTP was lowest with the hand in a natural working posture of up to 15 wrist extension, less than 20 wrist flexion, and moderate ulnar deviation. Considering these findings, it would appear that minimizing extension or flexion could reduce CTP during keying. Implications of these findings in relation to typing tasks are meaningful. Average wrist extension angles while typing have been reported to be between 13 and 33 (Hedge and Powers, 1995; Honan et al., 1996; Honan et al., 1995; Sommerich and Marras, 1994). Rempel and Horie (1994) found the lowest CTP occurred at 0 or 15 of wrist extension. They advocated that wrist extension angle was a strong determinant of CTP, and suggested minimizing wrist extension during typing. Additionally, specifications for key activation forces of most keyboards fall between 0.2 N and 0.9 N (Rose, 1991), although the ANSI/HFS standard specified a
9
larger range of key activation force from 0.25 N to 1.5 N (ANSI/HFS 100, 1988). Relaxed finger weights range from 0.3 N to 1.2 N, which often outweigh key activation forces, causing potential accidental activation (Rose, 1991). Force requirements for key activation do not reflect the actual exertions of keyboard operators when keying (Radwin, 1997). Studies have shown that key strike force exceeded the necessary key activation force (Armstrong et al., 1994; Feuerstein et al., 1994). Peak force during keying has been measured to be as much as 2.5 to 4.6 times the required activation force (Armstrong et al., 1994; Feuerstein et al., 1994; Martin et al., 1994). In extreme wrist extension, greater finger forces and higher CTP result for higher key strike forces (Table II.1). CTP values also increased for extreme wrist flexion, although not as severely. Effect of Keyboard Angle on Recorded Muscle Activity during Typing Tasks Muscle activity during typing tasks has also been studied. Martin et al. (1998) studied the reliability of methods to quantify muscle load for keyboard work and concluded surface electromyography (EMG) measurements reliably represented the activity of the extensor carpi ulnaris (ECU), extensor indices proprius (EIP), extensor digitorum communis (EDC), flexor digitorum communis (FDC) and flexor carpi radialis (FCR) muscles. They found the FCU and ECU muscles could be considered the primary muscles used in typing tasks. Gilad and Harel (2000) assessed muscle activity of the forearm and shoulder using EMG measurements for a negatively sloped keyboard system; however, it was unclear what angles were investigated. Surface electrodes were placed over the flexor carpi ulnaris (FCU), extensor carpi ulnaris (ECU), deltoid, and trapezius. They found that EMG measurements for average effort of the FCU when typing on a negative sloped keyboard were 36% less when compared with the average effort on the TonyTM keyboard design and 58% less than the split keyboard design. Results for the average effort of the ECU with the negative design were found to be 28% less than the average effort for the TonyTM keyboard.
10
Effects of Keyboard Height on Operator Discomfort during Typing Tasks Placing the keyboard above elbow height can increase upper-extremity discomfort (Bergqvist et al., 1995; Faucett and Rempel, 1994; Life and Pheasant, 1984; Sauter et al., 1991). Only one study was found that explicitly investigated the influence of keyboard height while using a negatively sloped keyboard system. Simoneau and Marklin (2001) studied the effects of negative keyboard sloping while adjusting the keyboard above, even with, and below elbow height. Results showed that if the elbow and wrist were placed at equal heights with the keyboard negatively sloped 7.5 or 15, wrist extension would average 15 or less. This configuration adheres to the suggested standard (ANSI/HFS 100, 1988). Effects of Wrist Rest Usage during Typing Tasks Wrist and palm rests act as a resting place for the wrist and hands; however, identified benefits associated with wrist rest use are varied. Albin (1997) found that wrist rests promoted a more neutral wrist posture by reducing wrist extension angles. Additionally, participants perceived that conditions with a wrist rest were significantly more comfortable than conditions without a wrist rest. Smith et al. (1998) found that participants using a wrist rest reported feeling more in control of their typing and when not using a wrist rest reported more pain in the front of their shoulders and on the outside of their elbows/forearms. Rose (1991) identified a possible explanation for increased discomfort when not using a wrist rest. He discovered that finger extensor muscles could not be relaxed when typing had ceased because the weight of the hand or fingers normally outweighs key activation forces. Extensor muscles were therefore required to exert constant static contractions to retain fingers in typing-ready postures (Rose, 1991). Hedge and Powers (1995) found that the broad palm rest used with the NSKS supported the hands while typing, allowing the hands to rest and type simultaneously without the wrist being fully pronated. This broad palm rest supported the majority of weight of the hand, substantially decreasing the amount of weight placed directly on the keys and nullifying any accidental activation of keys from resting the hands on the keys. A negatively
11
sloped keyboard may provide improved conditions for resting the hands while typing by facilitating the natural hand and finger posture observed by Rose (1991) and reducing static loading on the extensor muscles. The use of wrist rests has also been implicated as a potential contributor to hand and wrist WMSD development. CTP was found to increase by over 120% when using a wrist rest or resting wrists on a work surface while keying (Hedge, 1994). Paul and Menon (1994) stated that contoured wrist pads actually focused pressure over the carpal tunnel, but floating the hands over a resting surface while typing significantly decreased CTP (Horie et al., 1993; Albin, 1997). Parsons (1991) tested nine different wrist rests and found that none of the forty participants found any wrist rest beneficial, and 10% reported increased discomfort when using a wrist rest with a traditional (positively sloped) keyboard. Wrist rest usage with negatively sloped keyboard systems was found to affect observed wrist extension angles (Simoneau and Marklin, 2001). Wrist rests were either separate from the keyboard system (Figure II.1) or integrated with the keyboard system (Figure II.2). Separate wrist rests did not follow the tilt of the keyboard, and as the keyboard was sloped downward, the hand was kept farther away from the keyboard. Integrated wrist rests remained horizontal with the keyboard platform for every tilt angle, allowing the hand to follow the keyboard tilt. The type of wrist rest, separate or integrated, affected the location of the wrists pivot point (used to measure wrist angles). Simoneau and Marklin (2001) believed the difference in the wrist pivot point was responsible for a discrepancy of 10 in measured wrist extension when compared to a Hedge and Powers (1995) study. Both studies investigated the use of wrist rests with a negatively sloped keyboard system, though maximum negative angles were slightly different [15 in Simoneau and Marklin (2001), 12 0.4 in Hedge and Powers (1995)]. Simoneau and Marklin (2001) suggested that future negative slope keyboard studies should take into consideration the relationship between keyboard heights and wrist rest heights because both affected wrist extension.
12
Figure II.1. Negatively sloped keyboard with a separate wrist rest
Figure II.2. Negatively sloped keyboard with an integrated wrist rest
Typing Task Performance on the Negative Sloped Keyboard Varying results were also found between negative and standard keyboard designs when considering typing performance measures. Hedge (1995) found that 91% of users reported improved work throughput when using a negative sloped keyboard. Typing quality was also found to improve 27% when using the negative design versus the standard configuration (Gilad and Harel, 2000). Simoneau and Marklin (2001) found no significant differences in mean typing speed, or in mean typing accuracy across the five slope conditions (+15, +7.5, 0, -7.5, and -15) they studied. Similarly, Hedge and Powers (1995) found no significant difference in errors or in cumulative typing time between the standard keyboard configuration and the negative sloped keyboard. Summary Negatively sloped keyboard systems have come under light scrutiny within the past decade as a possible improvement upon the standard of a positively sloped keyboard. They have been documented to reduce the percentage of wrist postures outside of a neutral zone, produced full-body postural benefits, received positive reactions from participants using them, and have improved work output (Hedge et al., 1995; Hedge, 1994). Literature revealed a slowly growing recognition for use of negatively sloped keyboards; however, additional research was still needed. Few field and laboratory studies have been conducted primarily exploring negative sloped keyboards, and only one study compared electromyography results across four types of keyboard designs, not negative keyboard angles. No studies have investigated key strike force with the
13
use of negatively sloped keyboards, and few negative keyboard angles have been studied.
14
CHAPTER III METHODOLOGY Experimental Design A laboratory study was conducted to attain the stated objectives. A withinparticipant design was used to test for the effects of keyboard angle, as well as day, order, subject, gender, and side on wrist postures, forearm muscle activity, key strike force, self-reporting of discomfort for specific areas of the upper extremity, and typing task performance. Order of exposure to keyboard angle was balanced using a Latin square design (Table III.1). Table III.1. Balanced Latin square order of exposure to experimental keyboard angle Keyboard angle Participant 1 2 3 4 5 6 7 8 9 10 7o 3 4 5 1 2 1 2 3 4 5 0o 4 5 1 2 3 2 3 4 5 1 -10o 5 1 2 3 4 5 1 2 3 4 -20o 2 3 4 5 1 4 5 1 2 3 -30o 1 2 3 4 5 3 4 5 1 2
Dependent Variables Five dependent variablesfour objective measures and one subjective measurewere investigated. Objective measures included electrogoniometric measured wrist postures, EMG of the forearm muscles, key strike force measurements, and performance (i.e. typing speed, accuracy, types and numbers of errors made). The subjective measure was a self-reported postural discomfort questionnaire (SPDQ) with exit questions completed after each experimental condition. Discomfort was quantified using the Corlett postural discomfort scale (Corlett and Bishop, 1976).
15
Electrogoniometric Measurements A bi-axial electrogoniometer (SG65, Biometrics, Gwent, UK) was used to measure wrist flexion/extension (FE) and radioulnar (RU) deviation. The SG65 outputted signals of 2.3 mV to a battery-operated data logger outputting signals of 01.28 V. A straight line was drawn from the third metacarpal along the forearm to facilitate goniometer fitting. Goniometers were attached while the hand and forearm were in a neutral position [arm resting at the participants side, elbow flexed 90, wrist straight, and hand pronated to minimize crosstalk (Bucholz and Wellman, 1997)]. The distal end of the goniometer was affixed along the drawn line using double-sided tape. The participant extended the wrists completely, with help from the experimenter, and the proximal end of the goniometer was affixed to the forearm along the straight line. Cloth tape was placed over the ends of the goniometer to ensure no displacement occurred during typing tasks. Loose wiring was taped and clipped to clothing, allowing for slack, to avoid any shifting of the goniometers during typing. Calibration of the goniometers occurred before data collection. Zero angles were set to represent neutral posture, and positive angles denoted wrist extension and ulnar deviation, while negative angles denoted wrist flexion and radial deviation. All goniometric data was analyzed using DataLOGTM Software (Biometrics, Gwent, UK). Typing task data was software filtered and smoothed. Data unassociated with the typing task (i.e. reaching to turn the page, floating hands over the keys for an extended period of time without typing, etc.) was removed from analysis by using an IDENT switch to mark the times of such instances. These markers were easily noted in the program during data analysis, and time lengths were extracted from the data. Mean FE and RU angles were calculated for the entire ten-minute testing period, minus the first and last ten seconds of the experimental condition. Additionally, the percentage of time spent within a neutral zone (<15 extension, <30 flexion, <15 ulnar or radial deviation) was calculated and values were compared across conditions (i.e. day, keyboard angle, order, subject, gender, side). Other intervals were also investigated for FE and RU deviations, including 5 and 10.
16
Electromyography (EMG) Measurements EMG measurements of the flexor carpi ulnaris (FCU) and extensor carpi ulnaris (ECU) of both arms were obtained using 10-millimeter, circular Ag/AgCl pregelled bipolar disposable electrodes. Prior to electrode application, the skin surface area was shaved, slightly abraded, and cleansed with alcohol to ensure minimal impedance. Electrodes for the FCU were located 3 cm medial to the ulna at a point one-third of the length of the forearm from the elbow crease (Perotto, 1994). Electrodes for the ECU were located one-third of the distance between the lateral epicondyle of the humerus and the olecranon process and the styloid process of the ulna (Soderberg, 1992). Interelectrode distance was set to 2.5 cm. Signals were transmitted through short (less than 30 cm) leads to preamplifiers (100 gain). The leads were secured to the arm with tape to reduce noise and minimize displacement during typing. EMG signals were hardware amplified, band-pass filtered (10-500 Hz), RMS converted (110 ms time constant), and AD converted. If electrodes lost adhesion or were moved during the experiment, the trial was re-run (Marras, 1990). The gain was set such that the RMS signal did not exceed 2-3 volts. Input impedance was measured using a standard voltmeter to ensure impedance was within acceptable levels (0-10 ohms). After stabilization of the electrodes (15 minutes), resting and maximum voluntary contractions were obtained. Participants were instructed to sit with their arms in a posture representative of typing (elbows flexed at approximately 90, palm downward, and wrists straight). Resting EMG measurements were recorded at 256 Hz for ten seconds with the participants hands in his or her lap or resting on a table. Three, five second maximum voluntary contractions (MVC) were collected for each muscle in both forearms independently while seated, with a one-minute rest period between each trial. Three types of MVC tests were performed: (1) handgrip strength test (overall forearm muscle activation test); (2) wrist flexion/extension tests (individual forearm muscle activation tests); and (3) hand-twisting test. A number of studies have used handgrip strength tests to obtain MVC data for the forearm muscles. However, a study by Juul-Kristensen et al. (2002) showed that traditional handgrip strength tests using dynamometers underestimated MVC by an average of 34% for the participants tested. Twelve of the thirteen participants obtained
17
maximum ECU EMG measurements during maximum wrist extension, and nine of thirteen participants obtained maximum FCU measurements during the handgrip strength test. Therefore, wrist flexion/extension tests isolate the FCU and ECU muscles better in some cases than conducting a handgrip strength test alone. For this study, all MVC tests were performed while participants were positioned with arms resting at their sides, elbows flexed at 90, hands pronated, and wrists straight (representative of a typing posture). Handgrip strength was assessed using a dynamometer (Jamar Adjustable Dynamometer, Asimow Engineering Co.). The wrist flexion/extension MVC tests were performed using a stationary handle connected to an adjustable chain in the same posture described above. MVC for the FCU was obtained by requiring participants to perform a wrist flexion exertion, and for the ECU, during a wrist extension exertion. During each exertion, the experimenter held the upper limb in a consistent posture to reduce contributions from the upper arm and other supporting muscles. In the hand-twisting test, the participant was asked to twist a rod while seated, in the same posture as described above. This test was performed twice, alternating the direction of twisting for each hand. MVC tests were randomized across participants for the first test session, and the same sequence was used in the second test session to test repeatability. The peak RMS EMG signal was identified for each trial using LabView Software (Barr et al., 2001), and the maximum value for all three trials was taken as the MVC for that test. Later, MVCs from each test were compared, and the maximum value was used as the EMGmaximal value for that muscle for normalization of task EMG. After completing all MVC tests, participants then performed a five-minute practice-typing task during which the EMG signal was monitored to ensure equipment was working properly. Task RMS EMG was sampled at 256 Hz using National Instruments AD bit card, and LabView software, which smoothed (10 Hz low pass filter) and stored data. This smoothed data was used to estimate normalized force levels using Equation III.1. EMGobserved EMGresting EMGmaximal EMGresting
(Equation III.1)
18
Data unassociated with the typing task (i.e. reaching to turn the page, floating hands over the keys for an extended period of time without typing, etc.) was considered irrelevant and were removed from EMG data using a LabView program at the same times when irrelevant electrogoniometric data was removed. Means were calculated for the entire ten-minute testing period and at 30-second increments, minus the first and last ten seconds of the experimental condition. Processed data was expressed in terms of percent of maximum voluntary contraction and compared across conditions. Key Strike Force Measurements Two 22.2 N (5 lbf) Model LBS Series load buttons (Interface, Scottsdale, AZ) were used to measure key strike forces. The load button was small and cylindrical, 3.05 mm (0.12 in.) in height and 9.65 mm (0.38 in.) in diameter. It had an accuracy of 0.25% and its rated output was 2.0 mV/V. A 1.5 m (5 ft) integral cable connected the load button to a DMA Signal Conditioner/Amplifier. The manufacturer calibrated the amplifier to measure 0 VDC as 0 N (0 lbf) and 10 VDC as 22.2 N (5 lbf). The amplifier was connected to a National Instruments terminal block (SCB 100) for analysis with LabView Software. A Dell QuietKeyTM keyboard was engineered to allow the load buttons to monitor one key for each hand without damaging the integrity of the keyboards performance (Figure III.1). Participants were unaware which keys were being investigated. A metal plate inside the plastic keyboard casing held the circuit board against a plastic sheet of supporting rubber cones, connected to each key of the keyboard. The rubber cones forced the keys back to its resting state after a key was depressed and the circuit board, which is thin (<1 mm) and lightweight, had been activated. Along with a foam material, load buttons were housed underneath the key inside the foam and sandwiched between the metal plate and the circuit board. The foam kept the load button in place under the correct key and also allowed for a consistent and normal feel to the user for all the keys.
19
Figure III.1. Load button location inside keyboard The load buttons measured any depressing of a keys rubber cone through the thin circuit board, thus measured the key strike force for the key. The keys that were considered for study were the E key for the left hand and the N key for the right hand. The E key is a heavily used vowel, and appears in the Q row above home row. The N key is a commonly used consonant, and is located in the Z row below the home row. Choosing these keys allowed for measurement of key strike forces where finger extension varied between the rows above and below the home row. A LabView program was written to calculate peak values of each individual key strike force (KSF), and then averaged the peak values. Data from the load buttons was deciphered continuously for the ten-minute experiment. Three criteria were used to determine if a key strike force was made: (1) baseline, (2) threshold, and (3) width. The baseline value was determined so that any value below it was not considered part of a key strike force, or noise, and any value above the baseline was potentially a key strike force. To set the baseline, a pilot test was run at 1024 Hz, twice the frequency of the experiment. Data was collected continuously for four minutes, and E key and N key data was analyzed to find a maximum value, defined to be the maximum background noise in the absence of any key strikes. These values were manually inspected and adjusted to ensure all noise was removed and only potentially completed KSFs registered; moreover, 0.146 N for the E key, and 0.093 N for the N key, were used as the baseline for KSF consideration. Any data collected below these values was automatically discarded as noise.
20
The threshold value was used as a value that the key strike force must exceed in order to be considered a key strike force that activated the key successfully. This value was determined from another pilot test, run at 1024 Hz, where actual typing was done on the E and N keys. Light and hard tapping was done in rapid and slow motions, to test the range of the load button readings. Keys surrounding the E and N keys were also struck to see if there was any effect on the load buttons. This data was analyzed for peak values of successful key strike forces, which were found to be at 0.5-0.6 N. The threshold was set to 0.2 N for both keys, about half of these successful peak values. Any peak below this threshold value was assumed as noise or a partial key strike force that may not have activated the key successfully. The width value was the number of consecutive data points above the baseline and the threshold that contributed a key strike force. This value was set at 50 samples, about one-half a full key strike force or 0.1 sec, for data collection at 512 Hz. Larger widths might reduce the apparent amplitudes by combining separate key strike forces together. Consequently, the number of data points from above the baseline and past the threshold, and back to the baseline, had to exceed 50 in order for the program to consider the strike as a key strike force, and accurately measure the amplitude. Data was rechecked visually after the LabView program had analyzed the data. Data was collected at 512 Hz; contrary to the frequency of 250 Hz used in previous literature (Gerard et al., 1996; Gerard, Armstrong, Martin, Rempel, 2002; Gerard, Armstrong, Rempel, Woolley, 2002), and based on pilot study data. It was assumed that a peak value could never be accurately attained, but 1024 Hz, or three times the values used to collect key strike force data in previous literature, was sufficiently large to test for estimating peak values. Also, since capacity in the computer was limited, it was not conceivable, nor practical for data analysis purposes, to record data above 1024 Hz. A single key strike force for the E key (N key not shown) was graphed for 1024 Hz, 512 Hz, and 256 Hz (Figure III.2, Figure III.3, and Figure III.4, respectively). Comparing the same key strike force at different frequencies showed that the larger frequencies more accurately represented the key strike force than did lower frequencies. The same key strike force peak for 1024 Hz and 512 Hz was estimated to be above 1.8 N, but was approximately 1.58 N for 256 Hz. These larger frequencies
21
more accurately estimated peak values. Judging from this initial result, 1024 Hz and 512 Hz better estimated peak key strike forces. To further determine the best frequency regarding accurate key strike force representation, storage capacity, and time, means of key strike force peaks were calculated and graphed (Figure III.5). If key strike force peaks were accurately estimated, then mean values would not change significantly as frequency decreased. Figure III.5 shows a significant drop in this estimation as frequency decreased for both keys. Figure III.6 shows the percentage decrease in the estimation of the mean peak key strike forces for each incremental decrease in frequency. The decrease was slight for the first incremental decrease to 512 Hz, but was much greater for lower frequencies. When the sampling rate decreased from 1024 Hz to 512 Hz, there was a 2.3% and 1.7% decrease in the mean estimation of the peak values for key strike forces for the N key and E key, respectively. When the sampling rate decreased from 512 Hz to 256 Hz, there was a 6.5% and 3.6% decrease, or a tripling of percentage decrease from 1024 Hz, in the estimation of mean key strike force peaks for the N key and E key, respectively. This trend validated reasons to choose 512 Hz for sampling key strike force data. It was determined that to accurately estimate peak values for key strike forces in this experiment, 512 Hz was required. The first and last ten seconds of data for the experimental condition were removed from analysis. Means of key strike force peaks were presented in Newtons, and were calculated for the entire testing period for each key studied and compared across conditions.
22
1.9 1.7 1.5 Force (N) 1.3 1.1 0.9 0.7 0.5 0.3 0.000
0.010
0.020
0.029
0.039
0.049
0.059
0.068
0.078
Time (sec)
Figure III.2. Comparison of an E key strike force at 1024 Hz
1.9 1.7 1.5 Force (N) 1.3 1.1 0.9 0.7 0.5 0.3 0.000 0.010 0.020 0.029 0.039 0.049 0.059 0.068 0.078 0.088 Time (sec)
Figure III.3. Comparison of an E key strike force at 512 Hz
23
1.9 1.7 1.5 Force (N) 1.3 1.1 0.9 0.7 0.5 0.3 0.000 0.012 0.023 0.035 0.047 0.059 0.070 0.082 Time (sec)
Figure III.4. Comparison of an E key strike force at 256 Hz
Mean KSF Peaks (N) vs. Frequency (Hz)
Mean KSF Peaks ('N' key) 2.2
Key Strike Force (N)
Mean KSF Peaks ('E' key)
2.1 2.0 1.9 1.8 1.7 1.6 1.5 64 128 256 Frequency (Hz) 512 1024
Figure III.5. Means of key strike force peaks compared at varying frequencies
24
% decrease ('N' key)
% decrease ('E' key)
12
Percentage decrease
10 8 6 4 2 0 1.7 1024 Hz - 512 Hz 512 Hz - 256 Hz 6.5
9.7 7.0 6.0 2.3 3.6 3.7
256 Hz - 128 Hz
128 Hz - 64 Hz
Frequency comparison
Figure III.6. Percentage decrease in estimation of mean key strike force peaks Performance SkillCheck typing test software (SkillCheck, Inc., Burlington, MA) was used to administer a pre-test typing task, practice-typing tasks, and test-typing tasks for all experimental conditions. The program provided text passages for participants to recreate. The task was limited to the computer screen. Performance measures calculated by the software (i.e. test duration, gross typing speed, number of errors, errors per minute, net typing speed, missing words, extra words, joined words, split words, and misspelled words) were automatically recorded for each condition in the form of reports. These reports were printed and specific performance measures were compared across conditions. Self-reported Postural Discomfort Questionnaire (SPDQ) The SPDQ with added exit questions (Appendix E) was completed after each experimental condition. The Corlett postural discomfort scale was used to assess a participants subjective feelings of discomfort following exposure to each keyboard angle. Participants indicated on a 7-point scale, ranging from 0 = extremely comfortable to 7 = extremely uncomfortable, the amount of discomfort they experienced for the forearm, wrist, palm, and fingers independently. Exit questions were asked, such as: Compared to the standard keyboard configuration, the new typing angle was much easier, easier, about the same, harder, or much harder to use?; Did you reach the
25
same level of typing speed and accuracy compared to the standard keyboard configuration?; and at the end of the experiment, Please rank each keyboard angle from best to worst. Participants were forced to rank order the keyboard angles from 1 = best to 5 = worst, where each keyboard angle received a different ranking. Means, frequency counts, and other descriptive statistics were calculated for responses to the SPDQ and added exit questions, and were compared across day and keyboard angle. Friedman repeated measures ANOVA along with the SAS contrast method for post-hoc analysis was used to analyze the final ranking data. Independent Variable The independent variable was keyboard angle. The natural positive sloping of the keyboard was estimated to be 7 between the keyboard support surface and the plane crossing the center of the keys in the Q row and Z row, defined as keyboard slope in ANSI/HFS 100-1988 (Figure III.7). Keyboard wedges were used to position the keyboard at 0, -10, -20, and -30; but the angle of 7 was presented naturally (i.e. no wedge present). For 0 keyboard angle (Figure III.8, Figure III.9, and Figure III.10), the wedge was designed at 7 to counteract the natural sloping of the keyboard for a net effect of 0. The keyboard wedge was a triangular prism spanning the entire width of a standard QWERTY keyboard. The wedge was constructed of four wood piecestwo triangular wedges for the sides, one flat board for the keyboard surface, and a wood block to secure the keyboard from slipping. Keyboard wedges achieved a negative slope in a different manner than did the DT, NSKS, or PT keyboard tray systems. In general, the wedge can be used on a tabletop desk or conventional keyboard tray. The wedge design allowed users to place the keyboard anywhere on its surface (forward, backward, lateral) to accommodate user preferences, varying hand sizes, and comfort zones. To prevent possible confounding effects of a perceived wrist rest, a fixed stop was used to prevent slipping while typing (Figure III.8) for all experimental conditions, thus minimizing the exposure of the top edge of the wedge.
26
Figure III.7. Keyboard slope illustration
Figure III.8. Side view of experimental keyboard wedge (0)
Figure III.9. Top view of experimental keyboard wedge (0)
27
Figure III.10. Isometric view of experimental keyboard wedge (0) Typing Task Description A 17 monitor, standard QWERTY keyboard and Generation IV fully adjustable bi-level table (SIS Human Factor Technologies, Londonderry, NH) was used for the typing task. Typing tasks varied for each experimental condition. Text passages in the software package were adequate for the pre-test typing task, but another sourcea technical communication textbookwas used to create practice- and test-typing tasks. Using varying text passages prevented participants from establishing a learning curve during the experiment. Passages were designed to last the duration of each experiment. No passage was ever completed prior to the completion of a test session. No numeric keypad and mouse activity was required for the task. Participants Participants included ten experienced typists with a mean age of 22.1 years who spent at least 20 hours at a standard keyboard each week. Nine of the ten participants were right-handed. Participants used the ten-digit touch-typing method and typed a minimum of 45 words per minute. Potential participants were screened using the Nordic Questionnaire (Kuorinka et al., 1987) for current hand/wrist MSDs and other potential confounding disorders. Participants included in the study were those individuals responding negatively to, Have you at any time, during the last 12 months, had trouble (ache, pain, discomfort) in the wrists or hands? (Kuorinka et al., 1987). Participants also needed to have negative findings for at least one of Phalens and Tinels tests (Kuschner et al., 1992; Phalen, 1966) and report no symptomology (pain, numbness, or
28
tingling) in the median nerve distribution of the hand (thumb, index, middle, and ring fingers, and palm) (Rempel et al., 1988). Detailed descriptions of Phalens test and Tinels test can be found in Appendix A and Appendix B, respectively. Participants exhibiting any signs of WMSDs, severe hand/wrist injury, arthritis, use of antiinflammatory medication, or those that did not meet any of the criteria were excluded. Demographic information and anthropometric data were also collected (Van Cott and Kinkade, 1972), and are summarized in Appendix F. Procedure Participants were required to attend a qualifying session, and upon qualification, complete two test sessions. In the qualifying session, potential participants received a verbal and written description of the project, its objectives, and the procedures used, and were asked to complete informed consent documents approved through the Institutional Review Board for research involving human participants prior to any data collection (Appendix C). The Nordic questionnaire (Appendix D) and Phalens and Tinels tests were conducted to establish the absence of any WMSDs associated with the hands and wrists, thus determining whether or not the participants health at the present moment qualified him/her for further participation in the experiment. A threeminute pre-test typing task was completed for assessing typing proficiency. Participants were required to have a net typing speed of 45 words per minute. For those participants meeting all inclusion criteria, demographic information (i.e. age, current employment, keyboarding habits, etc.) and anthropometric data (i.e. height, weight, etc.) were collected (Appendix F). Once inclusion in the experiment was determined, participants were scheduled for two test sessions at a later date. The two test sessions were completed at the same time of day on the same day of the week, one week apart, to assess repeatability of the experimental procedure and results. While seated, EMG surface electrodes and electrogoniometers were fastened to the participants forearm, wrists, and hands. Following calibration of the equipment and a fifteen-minute stabilization period, resting and MVC EMG data were collected. Participants sat at the experimental computer workstation, which was adjusted so that the typing posture closely resembled the posture used by Hedge et al. (1999). The
29
participant was seated in a chair with a backrest in an upright position. Rempel et al. (1997), Simoneau et al. (1999), and Marklin et al. (1999) advocated that placing the keyboard above elbow height increased upper-extremity discomfort; therefore, for the purposes of this study, the forearm was kept parallel to the floor, keeping an elbow angle of approximately 80 to 110. Keyboard tray height was adjusted such that the elbow point was aligned with the wrist point, parallel to the floor. Chair height was adjusted so that the knees formed a 90 angle and the feet were flat on the floor. A foot rest was used, if necessary. Workstation parameters (keyboard height and chair height) were recorded. The appropriate keyboard wedge was placed under the keyboard and a fiveminute practice-typing task ensued to allow participants to become accustomed to the typing angle of the experimental condition. A five-minute rest period elapsed, and then participants completed a ten-minute typing task, followed by another five-minute rest period. During this second rest period, goniometer data was downloaded to the computer and the next keyboard wedge was set up. The SPDQ, with added exit questions, was then administered to assess the participants perceived discomfort for the forearm, wrist, hand, and fingers for that keyboard angle. This procedure repeated for each experimental condition until all five had been completed. When the participant had finished all experimental conditions, electrodes and goniometers were removed, and the participant was compensated for their time. Protocols for the second test session were identical to the first. Participants followed the same balanced Latin square design for both test sessions. Total testing session time was 8.5 hours per participant. Statistical Data Analysis Descriptive statistics (i.e. means, medians, standard deviations, frequency counts, etc.) were calculated for each dependent variable where appropriate. The Shapiro-Wilks normality test was performed on all continuous variables to ensure assumptions pertaining to normality were met for subsequent analyses (Appendix G).
30
One-way ANOVA Three one-way repeated measures ANOVA models were used for data analysis. The first ANOVA model determined if differences between days existed for each dependent variable. For those dependent variables where no significance by day was identified, the data was pooled, and the second ANOVA model testing for differences in order was conducted on the aggregate data. For dependent variables where significance was found by day, data was not pooled, and the second ANOVA model was conducted. The third and final model tested for keyboard angle, gender, and subject differences, as well as the keyboard angle and gender interaction. ANOVA was also used to assess differences for each dependent variable by side. Tukeys HSD was used for post-hoc analysis for significant findings. The Friedman repeated measures ANOVA along with the SAS contrast method for post-hoc analysis was used to analyze final ranking data. Significant values were reported at p < 0.05. Linear Correlations by Day Linear correlations (Pearsons or Spearmans Rho) were used to assess repeatability of each dependent variable depending upon normality of the continuous dependent variables. Critical values for Pearsons and Spearmans Rho correlation coefficients were used to determine significance, and was calculated based on sample size. All variables except MVC variables had samples of N = 50, thus critical values were rPCV = 0.288 (Fisher, 1973) for Pearsons correlations for normal data and rSCV = 0.279 (Zar, 1972) for Spearmans Rho correlations for non-normal data for df = 48 and p < 0.05. MVC variables had samples of N = 10 and data was normal, thus critical values were rPCV-MVC = 0.632 (Fisher, 1973) for df = 8 at the same p level. Significance was found if correlations for either case exceeded critical values. Since the sample size differed for MVC variables, strong correlations were defined as rPearsonMVC > 0.632. Weak correlations were rPearsonMVC < 0.632. Strong correlations for remaining parametric and non-parametric variables were defined as values above or equal to 0.6, moderate correlations were defined as values from 0.279 to 0.599, and values below 0.279 were considered weak correlations. Correlation ranges were chosen based on
31
where significance was found, while staying close to values used in common practice. Corresponding p-values were used to indicate the strength of these correlations. Body Part Groupings The SPDQ data consisted of ratings for various body parts. A one-way repeated measures ANOVA was conducted on body parts to determine significance among the ratings, and to assess if body parts could be grouped into logical body systems. Data was collected for ring and pinky fingers (RPs); the thumb, index, and middle fingers (TIMs); hand; wrist; and forearm for the left and right sides. Body parts were combined logically if there was no significance in the ratings found as a result of an ANOVA. For example, RPs and TIMs were compared first, since both were sets of fingers. If significance was not found for the left side and the right side for all keyboard angles and both days, the RPs and TIMs were combined into a body group. The logical bodygrouping process continued until: (1) ratings showed significance between each body groups or parts; (2) ratings showed significance by keyboard angle; (3) body groups or parts showed significance by side; or (4) ratings showed significance by day. Keeping the same body groups allowed for day, keyboard angle, and side analyses, as well as linear correlations to test repeatability. After logical body systems were identified, ratings were normalized according to Equation IV.1. Sum score Maximum sum score
(Equation IV.1)
Normalizing was necessary since summed body group ratings were likely to be greater than summed ratings of original body parts for the ten participants. The maximum response (MR) normalizing technique accounted for this discrepancy with a larger denominator when two body parts were grouped (14), or three body parts were grouped (21), and so forth. One-way repeated measures ANOVA were performed on these normalized values by day, keyboard angle, and side analyses. All discomfort data was treated as interval data, then analyzed as continuous data if summing and normalizing was necessary.
32
CHAPTER IV RESULTS Descriptive statistics for significant and non-significant variables by keyboard angle appear in Appendix J. Other descriptive statistics for significant findings are presented in the following sections (i.e., gender, side, etc.). A summary of the results for variables analyzed through repeated measures ANOVA are summarized in Table IV.1, Table IV.2, and Table IV.3. Only statistically significant results with descriptive statistics are presented in this section. Tables for post-hoc differences are presented in this section and all data are arranged by descending means. All F-ratios and p-values with Tukey HSD post-hoc q-values are tabulated and presented in Appendix H. In general, most dependent variables were consistent across days, no order effects were found, and differences between participants were found for most variables. Significant day effects were found for normalized EMG values for the left FCU, and left and right ECU; left hand mean RU deviations, mean %RU deviations in the neutral zone, and mean %RU deviations within 10; left hand and right hand mean %RU deviations within 5; net typing speed, errors; missing words; misspelled words; and normalized left and right hand and forearm discomfort ratings. Insignificant results between participants were found for mean %FE deviations within 5; day 2 left hand mean %RU deviations within 5; right hand mean %FE deviations within 5; day 2 missing words, combined extra-, joined-, and split-word (EJS) data; and self rating of keyboard usage difficulty as compared to the standard keyboard arrangement.
33
Table IV.1. Table of results for objective dependent variables Variable Normalized EMG FCU - RH Normalized EMG ECU - RH Normalized EMG FCU - LH Normalized EMG ECU - LH MVC FCU - RH MVC ECU - RH MVC FCU - LH MVC ECU - LH Mean of KSF Peaks - E key Mean of KSF Peaks - N key Mean FE - LH Mean RU - LH % FE in Neutral Zone - LH % RU in Neutral Zone - LH Mean FE - RH Mean RU - RH % FE in Neutral Zone - RH % RU in Neutral Zone - RH % FE < 5 degrees - LH %RU < 5 degrees - LH % FE < 5 degrees - RH %RU < 5 degrees - RH %FE < 10 degrees - LH %RU < 10 degrees - LH %FE < 10 degrees - RH %RU < 10 degrees - RH Net Typing Speed No. of Errors Missing Words Misspelled Words Combined EJS data Day N Y Y Y N N N N N N N Y N Y N N N N N Y N Y N Y N N Y Y Y Y N 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 1 2 Order N N N N N N N N/A N/A N/A N/A N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N KA N Y Y N Y N N N/A N/A N/A N/A N N Y Y Y Y Y Y Y Y Y Y Y Y Y Y N N Y Y Y Y N Y Y N N N N N N N Subject Y Y Y Y Y Y Y N/A N/A N/A N/A Y Y Y Y Y Y Y Y Y Y Y Y N Y N N Y Y Y Y Y Y Y Y Y Y Y Y N Y Y N Gender Y N N N N N N N/A N/A N/A N/A N N N N N N N N N N N N N N N N N N N N N N N N N N N N Y N N N
NOTE: EMG = electromyography, MVC = maximum voluntary contraction, FCU = flexor carpi ulnaris, ECU = extensor carpi ulnaris, RH = right hand, LH = left hand, KSF = key strike force, FE = flexion/extension, RU = radial/ulnar deviation, EJS = extra-, jointed-, and split-words
34
Table IV.2. Table of results for subjective dependent variables Variable Normalized Left Hand Rating (SPDQ) (TIMs, RPs, Hand) Normalized Left Wrist Rating (SPDQ) Normalized Left Forearm Rating (SPDQ) Normalized Right Hand Rating (SPDQ) (TIMs, RPs, Hand) Normalized Right Wrist Rating (SPDQ) Normalized Right Forearm Rating (SPDQ) Self-rating Day 1 Y 2 N Y Y N Y N 1 2 1 2 1 2 Order N N N N N N N N N N N KA N N N N N N N N N N Y Subject Y Y Y Y Y Y Y Y Y Y N Gender N N Y N N N N Y N N Y
NOTE: SPDQ = Self-reported postural discomfort questionnaire; TIM = thumb, index, and middle fingers; RP = ring and pinky fingers
Table IV.3. Table of results by side Variable Normalized EMG - FCU Normalized EMG - ECU MVC FCU MVC ECU Mean of KSF Peaks Mean FE Mean RU % FE in Neutral Zone % RU in Neutral Zone % FE < 5 degrees %RU < 5 degrees %FE < 10 degrees %RU < 10 degrees Normalized Hand Rating (SPDQ) (TIMs, RPs, Hand) Normalized Wrist Rating (SPDQ) Normalized Forearm Rating (SPDQ) 1 2 1 2 1 1 2 1 2 1 2 1 2 Day 1 2 1 2 Side N Y N Y N N Y N Y Y N N Y N N Y N N Y N N N N N
NOTE: EMG = electromyography, MVC = maximum voluntary contraction, FCU = flexor carpi ulnaris, ECU = extensor carpi ulnaris, KSF = key strike force, FE = flexion/extension, RU = radial/ulnar deviation, SPDQ = Self-reported postural discomfort questionnaire
35
Electrogoniometric Data Day Left hand mean RU deviations, mean %RU deviations in the neutral zone, mean %RU deviations within 5, and mean %RU deviations within 10 were significant by day. Left hand values for day 2 were significantly greater than day 1 for mean RU deviations (18.6 versus 15.6). Day 1 was significantly greater than day 2 for mean left hand %RU deviations in the neutral zone (47.3 versus 32.7), mean %RU deviations within 10 (25.9 versus 12.0), and mean %RU deviations within 5 (8.9 versus 2.4). Right hand mean %RU deviations within 5 were significantly greater on day 2 than day 1 (23.5 versus 12.9). Keyboard Angle Left Hand All dependent variables were significant by keyboard angle (Table IV.4). The -30 keyboard angle showed the lowest mean FE deviations for the left hand (=3.2), which differed significantly from -20, -10, 0, and 7, though no difference was found between the 0 and 7 keyboard angles. The greatest mean percentage of left hand FE deviations in the neutral zone ( =91.2%, =13.24%) was found for the 30o keyboard angle, which differed significantly from -20, -10, 0, and 7, though no difference was found between the 0 and 7 keyboard angles. Similarly, -30 showed the greatest mean percentage of left hand FE deviations within 10 ( =71.1%, =23.57%) and within 5o ( =40.5%, =23.51%), and in both cases differed significantly from -20, -10, 0, and 7, though no difference was found between the -10, 0, and 7 keyboard angles. The 7 keyboard angle showed the lowest mean RU deviations for both day 1 ( =13.3, =7.81o) and day 2 ( =14.8, =6.62o). On day 1, the 7 keyboard angle was different from -20 and -30; and for day 2, it was different from -10, -20, and -30. The greatest mean percentage of left hand RU deviations in the neutral zone was found for the 7o keyboard angle on day 1 ( =59.5%, =47.35%) and day 2 ( =51.3%,
36
=44.24%). On day 1, the 7 keyboard angle was surprisingly similar to -10 ( =57.2%,
=41.49%), but was different from 0, -20, and -30. On day 2, the 7 keyboard angle
was different from -10, -20, and -30. The 7 keyboard angle resulted in the greatest mean percentage of left hand RU deviations within 10 on day 1 ( =39.56%, =38.79%) and day 2 ( =27.4%, =34.61%). For both days, the 7 keyboard angle was different from -10, -20, and -30. The 7 keyboard angle also produced the greatest mean percentage of left hand RU deviations within 5 on day 1 ( =14.7%, =20.77%) and day 2 ( =7.2%, =12.34%). On day 1, the 7 keyboard angle was different from 10, -20, and -30. On day 2, Tukeys HSD post-hoc analysis was inconclusive. Right Hand All dependent variables were significant by keyboard angle except mean %RU deviations within 10, and day 1 and day 2 mean %RU deviations within 5 (Table IV.5). The -30 keyboard angle resulted in the lowest mean FE deviations for the right hand ( =2.4, =6.22), which differed significantly from -20, -10, 0, and 7. All keyboard angles were significantly different from each other. The -30 keyboard resulted in the greatest mean percentage of right hand FE deviations in the neutral zone ( =94.4%, =10.4%), mean percentage of right hand FE deviations within 10 ( =77.0%, =17.6%), and mean percentage of right hand FE deviations within 5 ( =44.0%, =20.86%), which differed significantly from -20, -10, 0, and 7, though no difference was found between the -10, 0, and 7 keyboard angles. The 7 keyboard angle produced the lowest right hand RU deviations ( =10.8,
=8.19), and was similar to 0( =12.2, =8.47), but both keyboard angles differed
significantly from -10, -20, and -30. The greatest mean percentage of right hand RU deviations in the neutral zone was found for the 7o keyboard angle ( =68.6%, =38.48%), which was similar to 0 ( =62.9%, =39.84%), but was different from -10, 20, and -30.
37
Table IV.4. Left hand post-hoc differences for keyboard angle Left Hand Mean (Standard Deviation) 7: 30.8 (6.84) 0: 28.8 (6.87) -10: 21.5 (7.19) -20: 10.8 (6.66) -30: 3.2 (7.18) -30: 91.2% (13.24%) -20: 70.3% (28.94%) -10: 25.3% (30.02%) 0: 6.3% (13.06%) 7: 3.0% (7.88%) -30: 71.1% (23.57%) -20: 45.4% (32.05%) -10: 11.0% (20.36%) 0: 1.2% (3.14%) 7: 0.4% (1.29%) -30: 40.5% (23.51%) -20: 21.4% (25.06%) -10: 3.0% (8.35%) 0: 0.1% (0.21%) 7: 0.01% (0.04%) Day 1 and 2 D D C B A A B C D D A B C C C A B C C C D D D D
Mean FE
%FE in NZ
%FE < 10
%FE < 5
38
Table IV.4. Left hand post-hoc differences for keyboard angle (continued) Left Hand Mean (Standard deviation) -30: 19.7 (7.28) -20: 17.3 (6.92) -10: 14.4 (7.65) 0: 13.4 (7.94) 7: 13.3 (7.81) 7: 59.5% (47.35%) -10: 57.2% (41.49%) 0: 48.6% (44.22%) -20: 39.4% (40.0%) -30: 31.8% (37.7%) 7: 39.6% (38.79%) 0: 34.2% (38.09%) -10: 28.6% (28.72%) -20: 18.2% (26.09%) -30: 9.1% (20.89% 7: 14.7% (20.77%) 0: 13.3% (19.86%) -10: 8.7% (16.41%) -20: 4.9% (11.66%) -30: 3.2% (9.48%) Mean (Standard deviation) -30: 22.8 (6.21) -20: 21.4 (6.78) -10: 18.0 (5.58) 0: 16.1 (6.26) 7: 14.8 (6.62) 7: 51.3% (44.24%) 0: 47.9% (42.33%) -10: 34.3% (35.44%) -20: 22.1% (31.14%) -30: 8.2% (10.19%) 7: 27.4% (34.61%) 0: 20.4% (26.06%) -10: 7.5% (9.28%) -20: 4.2% (6.98%) -30: 0.6% (1.16%) 7: 7.2% (12.34%) 0: 3.9% (6.12%) -10: 0.9% (1.84%) -20: 0.2% (0.36%) -30: 0.04% (0.07%) Day 1 A A B B
Mean RU
C C C C C C
D D D D
%RU in NZ
A A A
B B B B B B B B B B B B
%RU < 10
A A A
C C C
D D
%RU < 5
A A A A A
C C C
D D
Left Hand
Day 2 B C C C C C D D D D
Mean RU
%RU in NZ
A A
B B
%RU < 10
A A A
B B B
C C
D D
%RU < 5
Tukey HSD post-hoc analysis inconclusive
39
Table IV.5. Right hand post-hoc differences for keyboard angle Right Hand Mean (Standard deviation) 7: 34.0 (6.22) 0: 28.7 (6.25) -10: 20.8 (5.32) -20: 11.3 (5.99) -30: 2.4 (6.22) -30: 94.4% (10.4%) -20: 70.0% (29.2%) -10: 22.1% (18.61%) 0: 3.5% (5.07%) 7: 0.7% (1.41%) -30: 77.0% (17.6%) -20: 45.0% (33.61%) -10: 5.4% (6.06%) 0: 0.6% (1.35%) 7: 0.07% (0.18%) -30: 44.0% (20.86%) -20: 20.4% (21.89%) -10: 1.0% (1.40%) 0: 0.1% (0.38%) 7: 0.02% (0.07%) -30: 17.0 (9.54) -20: 14.7 (9.17) -10: 13.1 (9.10) 0: 12.2 (8.47) 7: 10.8 (8.19) 7: 68.6% (38.48%) 0: 62.9% (39.84%) -10: 54.5% (44.27%) -20: 45.3% (43.44%) -30: 37.4% (43.89%) Day 1 and 2 E D C B A A B C D D A B C C C A B C C C A A A B B B B B B B B D D D D
Mean FE
%FE in NZ
%FE < 10
%FE < 5
Mean RU
C C C C C C
D D D D
%RU in NZ
A A A
KA by Gender Interaction The keyboard angle by gender interaction was significant only for mean FE left hand deviations (F(4,8) = 3.0585, p < 0.05). Mean values decreased as keyboard angle decreased, but gender differences were noted at the -30 keyboard angle, where males ( =7.69, =4.95) were higher than females ( =0.29, =7.05) (Figure IV.1).
40
Female
Male
Mean FE Left Hand Deviations (deg)
35 30 25 20 15 10 5 0 -30 -20 -10 Keyboard Angle (deg) 0 7
Figure IV.1. Mean FE left hand deviation KA by gender interaction Side Day 1 and day 2 mean RU deviations, and day 2 mean %RU deviations in the neutral zone, mean %RU deviations within 5, and mean %RU deviations within 10 were significant by side. The left side was significantly greater than the right side for day 1 mean RU deviations (15.6 versus 12.8) and day 2 mean RU deviations (18.6 versus 14.4). The right side was significantly greater than the left side for day 2 mean %RU deviations in the neutral zone (49.1 versus 32.7), day 2 mean %RU deviations within 10 (39.2 versus 12.0), and mean %RU deviations within 5 (23.5 versus 2.4). Electromyography Data Graphs showing mean EMG values at 30-second intervals versus time for each of the four muscles for one participant appear in Appendix I for illustrative purposes only. The graphs depict fluctuations in muscle activity throughout the entire ten-minute test session; which were not seen in mean values. Day Normalized EMG values for the left arm FCU and ECU were significant by day. Day 1 values were significantly greater than day 2 values for the left arm FCU muscle (6.4% MVC versus 3.9% MVC) and left arm ECU muscle (11.9% MVC versus 9.1% MVC). Day 2 values were significantly greater than day 1 values for the right arm ECU muscle (12.0% MVC versus 11.1% MVC).
41
Keyboard Angle and Gender Day 2 normalized left arm FCU values and day 1 and day 2 normalized right arm ECU values were significant by keyboard angle (Table IV.6). The 7 keyboard angle required significantly less left arm FCU muscle activity than any other keyboard angle on day 2 of testing. On day 1, the 0 keyboard angle required significantly less right arm ECU muscle activity than the other keyboard angles. On day 2, each of the three negative keyboard angles required significantly less right arm ECU muscle activity than the 0 or the 7 keyboard angle, though no differences were found for muscle activity requirements among the negative angles. Right arm normalized FCU values were significant by gender. FCU values for females (7.3% MVC) were significantly greater than males (4.5% MVC).
Table IV.6. EMG post-hoc differences for keyboard angle EMG Variable Mean (Standard deviation) (%MVC) 7: 12.57% (3.54%) -10: 11.32% (3.02%) -30: 10.94% (3.45%) -20: 10.78% (3.93%) 0: 10.11% (4.28%) Mean (Standard deviation) (%MVC) 7: 13.38% (4.9%) 0: 12.40% (4.49%) -30: 11.56% (5.01%) -10: 11.46% (5.17%) -20: 11.39% (4.77%) -20: 4.21% (2.88%) -30: 4.08% (2.78%) -10: 4.05% (2.85%) 0: 3.73% (2.49%) 7: 3.59% (2.17%) Day 1 A A A A B B B B
Right arm ECU
C C C Day 2
D D
EMG Variable
A A
Right arm ECU
B B B A A A A
C C C B B B B
D D
Left arm FCU
-20: -30: -10: 0: 7:
C C C
D D
Side Day 2 normalized FCU and ECU values were significant by side, where the right side was significantly greater than the left side for both values (FCU: 6.34% MVC versus 3.93% MVC; ECU: 12.04% MVC versus 9.07% MVC).
42
Key Strike Force Data Mean Side key strike force was significantly greater for the N key than the E key (1.0682 N versus 1.0212 N). Performance Data Day Mean net typing speed, errors, missing words, and misspelled words were significant by day. Day 2 mean net typing speed was greater than day 1 (55.20 wpm versus 56.68 wpm). Day 1 mean values were greater than day 2 for errors (18.30 versus 14.56), missing words (1.24 versus 0.7), and misspelled words (15.82 versus 12.94). These findings suggest that participants traded off accuracy of the task for increased speed. Keyboard Angle, Gender, and KA by Gender Interaction Day 1 and day 2 mean net typing speed was significant by keyboard angle (Table IV.7). Day 2 missing words was significant by gender, where males (1.3) showed higher values than females (0.3). Day 1 mean net typing speed (F(4,8) = 3.6, p = 0.0156) showed significance in the KA by gender interaction (Figure IV.2).
Female 59
Male
Day 1 Mean NTS (Wpm)
58 57 56 55 54 53 52 7 0 -10 Keyboard Angle (deg) -20 -30
Figure IV.2. Day 1 mean net typing speed KA by gender interaction
43
Table IV.7. Net typing speed post-hoc differences for keyboard angle Performance Variable Mean (Standard deviation) (Wpm) -10: 57.29 (8.03) 7: 56.09 (9.22) -20: 55.10 (7.39) 0: 54.24 (9.35) -30: 53.30 (8.25) Mean (Standard deviation) (Wpm) -10: 58.36 (8.47) -20: 58.00 (7.23) 7: 56.68 (8.10) 0: 55.92 (9.05) -30: 54.44 (6.55) Day 1 -10: 7: -20: 0: -30: A A A B B B B Day 2 -10: -20: 7: 0: -30: A A A A B B B
Net Typing Speed
C C C
D D
Performance Variable
Net Typing Speed
C C C
D D
Self-reported Postural Discomfort Questionnaire Data Self-reported Postural Discomfort Questionnaire (7-point scale) After grouping body parts, one-way repeated measures ANOVA was conducted for discomfort ratings and self-ratings, and some variables showed significance by day, keyboard angle, subject, gender, and the KA by gender interaction. Day Normalized discomfort values for the left hand (day 1: =0.23 MR, =0.2 MR; day 2: =0.19 MR, =0.17 MR), left forearm (day 1: =0.32 MR, =0.27 MR; day 2:
=0.24 MR, =0.22 MR), right hand (day 1: =0.23 MR, =0.19 MR ; day 2: =0.18
MR, =0.18 MR), and right forearm (day 1: =0.30 MR, = 0.28 MR; day 2: =0.24 MR, =0.25 MR) showed significance by day. It is interesting to note that for each body part, discomfort rating decreased from day 1 to day 2. Keyboard Angle, Gender, and KA by Gender Interaction Only self-rating of keyboard usage difficulty (compared to the standard 7) was significant by keyboard angle. Self-rating, and normalized discomfort values for the left wrist and right wrist showed significance by gender. Males rated more discomfort for both the left wrist (males: =0.41 MR, =0.21 MR; females: =0.21 MR, =0.18 MR) 44
and right wrist (males: =0.41 MR, =0.21 MR; females: =0.15 MR, =0.18 MR). Day 2 normalized discomfort values for both the left hand (F(4,8) = 3.132, p = 0.0278) and right hand (F(4,8) = 4.6, p = 0.0048), and normalized discomfort values for the right wrist (F(4,8) = 4.739, p = 0.0017) showed significance in KA by gender interaction, with males reporting more discomfort across keyboard angles (Figures IV.3, IV.4, and IV.5).
Female 55
Normalized Right Wrist (%MR)
Male
45 35 25 15 5 7 0 -10 Keyboard Angle (deg) -20 -30
Figure IV.3. Normalized right wrist KA by gender interaction
Female 40
Day 2 Normalized Right Hand (%MR)
Male
35 30 25 20 15 10 5 0 7 0 -10 Keyboard Angle (deg) -20 -30
Figure IV.4. Day 2 normalized right hand KA by gender interaction
45
Female
Day 2 Normalized Left Hand (%MR)
Male
35 30 25 20 15 10 5 7 0 -10 Keyboard Angle (deg) -20 -30
Figure IV.5. Day 2 normalized left hand KA by gender interaction Exit Questions Exit questions were added to the self reported postural discomfort questionnaire (SPDQ) to compare the non-standard keyboard angles to the standard including: (1) self-rating comparison; (2) estimated time for typing speed comparison; and (3) final ranking. Self-rating Comparison (5-point scale) Self-ratings of keyboard angle usage difficulty compared to the standard configuration were assessed using a 5-point scale in which participants rated each new keyboard angle against the standard 7 keyboard angle. Ratings ranged from 1= much harder to use to 5 = much easier to use. Self-ratings did not vary by day, indicating that participants self-rated the keyboard angles consistently for each day. As can be seen from Figure IV.6, mean self-ratings decreased (or the level of difficulty for use increased) as the keyboard angle became more negative, and these ratings were found to be significant (Table IV.8). Minimum and maximum self-ratings received for each keyboard angle were approximately the same (Table IV.9), indicating that some participants perceived the usage of these keyboard angles to be comparable to the standard keyboard angle.
46
Self-rating was significant by gender. Males ( =2.71, =0.82) self-rated higher than females ( =2.19, =0.81), meaning males felt negative keyboard angles were easier to type on compared to the standard than females did.
Much easier to use
5
Easier to use
4
Rating
About the same as the standard Harder to use Much harder to use
3
2
1 0 -10 -20 -30
Keyboard Angle (Degrees)
Figure IV.6. Mean self-rating of new keyboard angles compared to 7 keyboard angle (n=20)
Table IV.8. Self-rating post-hoc differences for keyboard angle Discomfort Variable Self-rating Mean (Standard deviation) 0: 3.00 (0.73) -10: 2.65 (0.49) -20: 2.30 (0.57) -30: 1.65 (0.93) Day 1 and 2 A A B B
C D
Table IV.9. Minimum and maximum values for mean self-rating of new keyboard angles compared to 7 keyboard angle Keyboard Angle 0 -10 -20 -30 Minimum Value 2 2 1.5 1 Maximum Value 4 3 3 4
Estimated Time for Typing Speed Comparison Sixty percent of the populations responses indicated that they could reach the same level of typing speed and accuracy on any new keyboard angle when compared to their performance on the 7 keyboard angle. Participants were asked to estimate the 47
amount of time they believed would be necessary to reach this performance level (Figure IV.7). Mean time estimates for new keyboard angles on day 1 fell roughly between 30-50 hours. Day 2, however, showed on average that less than 10 hours is needed to acclimate to -20, but over 70 hours is needed to reach the same level at the -10 keyboard angle. From day 1 to day 2, estimated learning times decreased on average by 33% for 0, 85% for -20, and 55% for the -30 keyboard angle. For -10, the learning time increased on average by 37%. In general, most participants felt confident they could learn to type on the new keyboard angles in a relatively short period of time (Table IV.10).
Day 1 140 120 100
No. of Hours
76.16 n=7
Day 2
80 60 40 20 0 0
29.39 n=9
48.03 n=7
40.11 n=7
37.00 n=2
19.84 n=10 6.05 n=4
16.69 n=3
-10
-20
-30
Keyboard Angle (Degrees)
Figure IV.7. Mean time estimated to reach same typing speed compared to 7 keyboard angle by day (n=10)
48
Table IV.10. Means and standard deviations for estimated time of typing speed Day Day 1 Keyboard Angle 0 -10 -20 -30 0 -10 -20 -30 Mean (Standard deviation) (Hrs) 29.39 (42.85) 48.03 (61.55) 40.11 (65.50) 37.00 (49.50) 19.84 (41.71) 76.16 (129.81) 6.05 (11.96) 16.69 (27.13)
Day 2
Final Ranking Participants were forced to rank order the keyboard angles from 1 = best to 5 = worst. The Friedman test was generated for this repeated measures comparison. For both day 1 and day 2, there was a significant difference between the mean rankings (day 1: F(4,36) = 7.48, p < 0.001; day 2: F(4,36) = 4.16, p < 0.01), with most preferring the standard or 7 configuration. This result was expected since the final ranking was a forced rankingonly one ranking could be assigned to each keyboard angle. Keyboard Angle and Gender Post-hoc differences by day were observed, while mean rankings were practically identical (Table IV.11). Day 1 mean rankings showed more differences between the final rankings among the keyboard angles, while day 2 rankings indicated that only the -30 keyboard angle was ranked significantly lower than all the other keyboard angles. Gender differences were observed for day 2 rankings. The time by gender interaction for day 2 showed significance at F(4, 32) = 3.28, p < 0.05. The univariate test revealed a significant difference for the 7 keyboard angle between males and females (F(1,9) = 10.82, p < 0.05). Females ( = 1.333) showed a higher mean ranking than males ( = 3.5).
49
Table IV.11. Final ranking post-hoc differences for keyboard angle (best to worst) Day Mean (Standard deviation) 7: 1.8 (1.03) 0: 2.4 (1.35) -10: 2.8 (0.92) -20: 3.6 (0.97) -30: 4.4 (1.35) 7: 2.2 (1.48) 0: 2.3 (1.16) -10: 2.7 (0.95) -20: 3.5 (1.18) -30: 4.3 (1.34) Post-hoc A A B B
Day 1
C D D
Day 2
A A A A B
Linear Correlations by Day Linear correlations were computed to assess repeatability across testing days (Table IV.12). Not counting MVC variables, Pearsons correlations were above 0.59, indicating a strong correlation between days. All MVC correlations were below 0.5. All nonparametric correlations were above 0.33 except for the left arm FCU muscle, which correlated at a value of 0.07. These variables were the only variables that did not show repeatability from day 1 to day 2. Thirteen percent of the correlations were weak, 19% were moderate, and 68% were strong.
50
Table IV.12. Table of linear correlations by day Parametric Correlations: Pearsons
Variable Normalized EMG RH FCU Mean RU - LH Mean FE - RH Net Typing Speed (NTS) MVC RH FCU MVC RH ECU MVC LH FCU MVC LH ECU Correlation (R) 0.5899 0.6272 0.9257 0.8978 0.1974 0.4448 0.2803 0.4933 P-value 0.0000 0.0000 0.0000 0.0000 0.5846 0.1978 0.4328 0.1474
Nonparametric Correlations: Spearman's Rho
Variable Spearmans Rho 0.5751 Normalized EMG RH ECU 0.0684 Normalized EMG LH FCU 0.4262 Normalized EMG LH ECU 0.9323 Mean of KSF Peaks E key 0.8898 Mean of KSF Peaks N key 0.8740 %FE in Neutral Zone - LH 0.5633 %RU in Neutral Zone - LH 0.8652 Mean FE - LH 0.6354 Mean RU - RH 0.9335 %FE in Neutral Zone - RH 0.7211 %RU in Neutral Zone - RH 0.8500 %FE < 5 degrees - LH 0.6345 %RU < 5 degrees - LH 0.9000 %FE < 5 degrees - RH 0.4764 %RU < 5 degrees - RH 0.8755 %FE < 10 degrees - LH 0.6380 %RU < 10 degrees - LH 0.9101 %FE < 10 degrees - RH 0.7303 %RU < 10 degrees - RH 0.8300 Final Ranking 0.6396 No. of Errors 0.4438 Missing Words 0.3333 Extra, Joined, Split words (EJS) 0.6635 Misspelled Words Normalized Left Hand Rating (SPDQ) 0.8540 (TIMs, RPs, Palm) 0.5937 Normalized Left Wrist Rating (SPDQ) 0.7994 Normalized Left Forearm Rating (SPDQ) Normalized Right Hand Rating (SPDQ) 0.7917 (TIMs, RPs, Palm) 0.8408 Normalized Right Wrist Rating (SPDQ) 0.8443 Normalized Right Forearm Rating (SPDQ) 0.7159 Self-rating weak correlation (rS, rP 0.279; rPMVC < 0.632) moderate correlation (0.279 < rP, rS < 0.632) strong correlation (rS, rP, rPMVC 0.632) Prob>|Rho| <0.0001 0.6370 0.0020 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0005 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.0012 0.0180 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001
51
CHAPTER V DISCUSSION The objective of this study was to identify a negative keyboard angle or range of keyboard angles that minimized exposure to hypothesized risk factors for hand/wrist work related musculoskeletal disorders. It was hypothesized that keyboard angle would have significant effects on muscle activity of the flexor and extensor carpi ulnaris muscles in the forearm, key strike force, wrist posture, and reported discomfort while performance remained consistent. It was further hypothesized that at a keyboard angle of 30, the effects on all dependent variables would be negative. In general, the only hypotheses that were supported by this study for negative angles decreasing from 7 to -20 were: (1) wrist posture was within a neutral zone a greater percentage of the time with respect to FE deviations; but not RU deviations; and (2) performance remained consistent. Increased reported discomfort was the only hypothesis that was supported for the 30 keyboard angle. Wrist Posture This study supported findings from Hedge et al. (1999) that found 67% of typing movements (both FE and RU deviations) made while using negatively sloped keyboards were within the neutral zone (<15 extension, <30 flexion, <15 ulnar or radial deviation). However, for both hands, this study found mean FE typing movements exceeded 70% and 90% within the same neutral zone for -20 and -30, respectively, angles not studied in previous literature. The 7 and 0 keyboard angles yielded less than 7% of FE typing movements within the neutral zone. Mean FE percentages within tighter intervals than the neutral zone (<5 and <10) were also noticeably greater for extreme negative keyboard angles. For mean percentages of FE deviations within 10, -20 and -30 keyboard angles produced mean FE percentages above 45%, where the other keyboard angles were below 12%. For the tighter interval of within 5, -20 and -30 keyboard angles produced values above 20%, far superior to the other keyboard angles (<3%). These results show benefit for the -20 and -30 keyboard angles
52
regarding wrist posture, and are summarized in Figure V.1. Ideally, percentages of typing movements should be as close to the neutral position as possible to reduce risk of MSD development. The relationship of risk to typing posture is explained further in this section.
7
0
-10
-20
-30
Ideal case
100 90
Percentage of typing movements
80 70 60 50 40 30 20 10 0 Neutral +/- 5 deg +/-10 deg +/- 15 deg Degrees from Neutral
Figure V.1. Comparison of mean FE deviations within varying zones from neutral position Maximum negative slopes that were studied herein (-20 and -30) greatly exceeded slopes studied by Hedge and Powers (1995) (-12 0.4) and Simoneau and Marklin (2001) (-15). Surprisingly, results from this study for the extreme negative angles were similar to results found in these two studies. Hedge and Powers (1995) found mean wrist extension at a keyboard slope of -12 0.4 to be 1.2 2.0. Simoneau and Marklin (2001) found lowest mean wrist extension to be between 1-2 at a keyboard slope of -15 with wrists located above the elbows; but wrist extension for this same keyboard angle with the wrists below the elbows was between 15-17. Mean wrist extension in this study was similar, between 2-3, for the -30 keyboard angle. In
53
addition to these negative slope contrasts, comparing the standard keyboard angle amongst the studies showed noticeable differences as well. Mean wrist extension for the 7 keyboard angle studied herein was between 3134. These values agreed with aforementioned studies for average wrist extension for use with the standard keyboard configuration, although the values were on the higher end of the range. Hedge and Powers (1995) found wrist extension for a similar configuration to be 13.0 2.2, and Simoneau and Marklin (2001) found values between 8-10 for wrists located above the elbows and between 24-25 for wrists below the elbows on their 7.5 standard keyboard configuration. These values found by Hedge and Powers (1995) and Simoneau and Marklin (2001) where wrists were below the elbows were on the lower and middle ends, respectively, of the average range for wrist extension with the standard keyboard configuration. Although it was clear these studies showed benefits for using negatively sloped keyboards since mean wrist extension decreased as keyboard angle decreased, it was possible they underestimated potential wrist posture benefits. This finding was more obvious in this study when considering the percentage of FE typing movements within specified zones. These comparisons illustrate how dependent results were on typing posture, and these discrepancies may have been manifested in the use, or lack thereof, of a wrist rest. The Hedge and Powers and Simoneau and Marklin studies used wrist rests, either integrated or separate. Simoneau and Marklin (2001) found that wrist extension decreased approximately 13 for a net change of 30 in keyboard slope, or a 1 decrease in wrist extension for 2 of change in downward slope. This study tested negative angles that were double what Simoneau and Marklin (2001) tested without a wrist rest, and found wrist extension decreased on average 30 resulting from a net change of 37, or practically a 1 decrease in wrist extension for 1 of change in downward slope. Mean FE deviations at all five keyboard angles for both hands obtained in this study may be accredited to the absence of a wrist rest, because the lack of a wrist rest might allow upper extremity typing posture to vary. In hindsight, the absence of a wrist rest may have been a limitation. Mean FE deviations, however, still showed that greater negative angles benefit persons as deviated wrist postures are reduced while keying. Similar results were not as apparent for RU deviations. 54
Mean RU deviations in this study were found to be between 17-23 for the -30 keyboard angle. Values found in the studies by Hedge and Powers and Simoneau and Marklin were slightly lower for their respective maximum negative angles; but, results showed dependence on typing posture in the Simoneau and Marklin study. Simoneau and Marklin (2001) found mean RU deviations were lowest when wrists were below the elbows (as opposed to FE postures, presented earlier, which were lowest when wrists were located above the elbow). A separate wrist rest was used in that study. It is possible that different results may have been obtained if a wrist rest, separate or integrated, were used in the current study, to help keep the forearm parallel to the floor. These alternatives are a consideration for future research. The impact of decreasing keyboard angle on radioulnar deviations was unexpected. Right hand mean RU deviations and mean %RU deviations within the neutral zone for negative angles were significantly lower compared to the flat and standard keyboard angles. Interestingly, side effects were noticed for all %RU deviations, with the right side greatly exceeding the left side. Higher right-hand values may have been caused by a primarily right-handed population or by more static postures associated with loading differences between the hands. The difference in right hand mean %RU deviations within the neutral zone between the 7 and -30 keyboard angles was 69% and 37%, respectively. Similar values were found for day 1 left hand mean %RU deviations within the neutral zone, but day 2 values were greatly varied (50% versus 8%). These results might be attributed to studying more extreme negative angles, unlike previous literature, or because of varying typing posture between the days. Left side values were greater than right side values for mean RU deviations. Mean %RU deviations for the left hand within tighter intervals than the neutral zone (<5 and <10) were significant by day and keyboard angle, and decreased from day 1 to day 2. This result across days contradicted results for the right hand, where no differences by day were noted, and both variables were not significant by keyboard angle. Mean left-hand RU deviations increased from day 1 to day 2 also, which is a basis for increased mean %RU deviations. Reasons for significant differences in left hand activity may be that the right-hand dominant population did not achieve the same
55
level of comfort with the left hand at the typing angles, or, again, due to loading differences across the hands. Discomfort ratings varied by day for left and right hands and forearms, but were not significant by side. All day 1 means were greater than day 2 means, indicating less discomfort was experienced as time elapsed. It was possible a higher level of discomfort was tolerated for the right hand and yielded fewer effects by day, unlike the left hand, which varied greatly in RU typing movements. The left hand is also known to be overloaded because of the layout of keys (Dvorak, 1936), and may have contributed to the significant findings for all left hand variables and this decrease within tighter intervals across days. There were keyboard angle by gender interactions for day 2 normalized discomfort values for both the left and right hand, and normalized values for the right wrist. In general, normalized discomfort values for males were higher than females, and gradually converged as keyboard angle decreased. It would appear males were most comfortable at negative typing angles, where females were most comfortable at neutral or positive typing angles. It is possible that this finding is due to hand anthropometric differences, though this hypothesis was not tested. Performance Measures Compared with Discomfort Most performance measures remained consistent across keyboard angle, but most were significant by day. All performance measures except extra-, jointed-, and split word (EJS) data were significant by day, and may have resulted from participant differences (Figures V.2 - V.5). For all performance measures other than typing speed, values decreased from day 1 to day 2. Participants may have felt more comfortable with the study after completing the first trial, and their familiarity with the text passages may have unavoidably led to better performance, even with a week in between experiments. On the other hand, this result could have established that negative keyboard angles were easily learned. The -10 keyboard angle yielded the best mean net typing speeds, where 7 was second best on day 1 and -20 was second best on day 2 (Table V.I). Surprisingly, the 10 keyboard angle showed the largest mean values for all errors studied except
56
missing words, despite having the largest mean net typing speed. All keyboard angles resulted in lower missing words than the standard on either day. In general, typing at negative angles yielded better or comparable results to the standard regarding number of errors, missing words, misspelled words, or extra-, joined-, and split-word errors, while significant differences were only found for net typing speed across keyboard angles. Gilad and Harel (2000), Simoneau and Marklin (2001), and Hedge and Powers (1995) found no significant differences in respective typing quality measures (i.e. mean typing accuracy, mean errors, cumulative typing time) for the negative angles they studied in comparison to the 7 keyboard angle. Contrary to the stated hypothesis, the extreme keyboard angles did not hinder participants from seeing or reaching keys consistently and accurately with two of the three top performing keying angles resulting at -10 and -20. Although participants were very familiar with the standard keyboard configuration, these results showed evidence that participants could unknowingly succeed at typing on negative keyboard angles and were more accustomed to typing at these angles than they felt. The differences in net typing speed should invariably be a result of differences in errors, since errors are directly related to a participants net typing speed. This study produced results to show otherwise for keyboard angle differences. Day 1 and day 2 mean net typing speed showed the only significance by keyboard angle, and the KA by gender interaction showed significance on day 1. Females were more accurate than males, and participants typed their best at the -10 keyboard angle. While gender differences showed females outperforming males at the -20 keyboard angle, males outperformed females at the -30 keyboard angle. Participant differences, along with gender differences, may have contributed to these performance measure differences in keyboard angle because of varying hand size and differences in typing habits. Varying hand sizes could have lead to accidental key activations when a participant attempted to type certain words. Also, errors may originate from varying cognitive behaviors that lead to mistakes exclusive to that participant, which would result in lower typing speed.
57
Table V.1. Mean net typing speed Keyboard Angle 7 0 -10 -20 -30 Minimum Value (Wpm) 42.1 38.7 45.2 44.6 44.5 Median Value (Wpm) 55.7 55.2 57.3 56.7 54.1 Maximum Value (Wpm) 79.0 74.4 76.4 71.2 71.0
7
66 62
Mean Values (Wpm)
0
-10
-20
-30
58 54 50 46 42
56.1 54.2
57.3 55.1 53.3
58.4 56.7 55.9
58.0 54.4
NTS -- Day 1
Performance Measures
NTS -- Day 2
Figure V.2. Mean net typing speed by keyboard angle
58
7
35 30
Mean Values (No. of Errors)
0
-10
-20
-30
25
20.7
20 15 10 5 0
19.8 16.8 15.5 14.6 15.8 13.2 13.7
17.2
17.0
No. of errors -- Day 1
No. of errors -- Day 2
Performance Measures
Figure V.3. Mean values for number of errors by keyboard angle
7
3.5 3.0
Mean Values (No. of Words)
0
-10
-20
-30
2.5 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0
2.1
1.1 0.8
1.3 0.9
1.3 0.7 0.2 0.9 0.4
Missing words -- Day 1
Missing words -- Day 2
Performance Measures
Figure V.4. Mean values for missing words by keyboard angle
59
7
30 25 20 15 10 5 0
14.5 14.3 17.9 14.8
0
-10
-20
-30
Mean Values (No. of Words)
17.6 13.6 13.3 13.8 11.6
12.4
Misspelled words -- Day 1
Misspelled words -- Day 2
Performance Measures
Figure V.5. Mean values for misspelled words by keyboard angle Although performance measures varied to some extent, usability difficulty increased as keyboard angle decreased, though no other differences were found across keyboard angles for other subjective measures. Unlike previous studies, studying more extreme negative angles and more types of typing errors accounted for significant findings in performance measures. The usability self-rating showed participants preferred the 7 and 0 to the -10, -20, or -30 successively, and there was no significance in the ratings by keyboard angle according to the 7-point discomfort rating; however, it was possible these ratings were confounded. All participants in this study were accustomed to the standard 7 keyboard configuration, and may have had a natural negative response to the negative typing angles, simply because they were different. This finding needs to be investigated following prolonged exposure to the negative angles to support or refute this confounding hypothesis. Time estimates for reaching the same level of typing speed and performance as the standard configuration varied widely, and can be partially attributed to the varying number of responses. As was the case for both days, all or a majority of the participants were able to make time estimates for the 0 and -10 keyboard angles, 60
while only three to four participants were able to make such estimates for the -20 or 30 keyboard angles. These results showed that a great amount of personal preference was associated with this measure. Some participants felt comfortable with the -30 keyboard angle, and gave short time estimates. Other participants felt extreme discomfort, thus gave longer time estimates to acclimate to the typing angle. Final rankings were more widely varied on day 1 than day 2. Even though support was great for the flat and standard keyboard angles, final ranking comparison by day showed participants were more accustomed to the negative typing angles on day 2 than day 1, since rankings were only significant for the -30 keyboard angle on day 2. This result shows that given enough exposure, negative typing angles may be better received by users. Although there was a contradiction in objective and subjective results, discomfort decreased across days in this study, and may lead to general support for objective measure findings. The findings in this study did not support results found by Hedge (1994) that a majority of workers reported less discomfort using negatively sloped keyboards. Participants in this study, however, were not professional typists, as they were in the Hedge study. This fact may have exemplified increases for discomfort in this study because typing was not a full-time task, and participants may not have adapted to the rigorousness of the typing tasks. Males gave a higher mean self-rating than females, so negative angles were less disruptive to their typing ability. Males also reported more discomfort in the left and right wrists than females, even though performance did not yield any gender differences except day 2 missing words, where males made more errors than females. These results may be been attributed to differences in typing habits or anthropometry. Overall, as keyboard slope decreased, participants rated more discomfort while typing. Electromyography Muscle activity significantly increased in this study for right arm ECU means for the 7 keyboard angle and for left arm FCU means for the -20 keyboard angle (Figure V.6). For the right arm, the 7 keyboard angle yielded the highest ECU means of all the keyboard angles tested on both days. On day 1, it differed significantly only when 61
compared with the lowest mean value at the 0 keyboard angle. On day 2, mean values were higher for the 0 keyboard angle, but the lowest means occurred at -20, followed by the -10 and -30 keyboard angles. On day 1, the left arm FCU values yielded different results. First, the FCU mean values were much lower in comparison to the mean ECU values. The 7 keyboard angle had the lowest mean, and differed only from -20, which had the highest mean. These contradicting results were interesting. On the one hand, these results showed benefits for the use of negative keyboard angles over the standard when considering the right arm; however, when considering the left arm, the standard yielded benefits over negative keyboard angles for one of the days.
Right Arm ECU -- Day 1 14 Mean EMG values (%MVC) 12 10 8 6 4 2 7 0
Right Arm ECU -- Day 2
Left Arm FCU -- Day 1
-10 Keyboard Angle (deg)
-20
-30
Figure V.6. Mean EMG value results comparison by muscle and day It is known there is increased loading on the left hand due to the layout of keys, thus more muscle activity is presumed for the left arm; however, this study showed overall more loading on the right hand. One possible explanation is the fact the population was dominantly right-handed. As discussed later, key strike forces were significantly greater for the right hand than the left. These results may be possible causes for higher right side muscle activity. However, these explanations should result in higher side values for both days. Day 1 yielded no side differences, so discrepancies over days may be possibly explained by electrode placement. Moderately strong correlations of almost 0.6 resulted for right hand mean FCU and ECU values. Right hand and left hand MVCs may also have contributed to these significant differences for
62
day 2 since these values correlated weakly or moderately at best, and were used to obtain normalized values for EMG variables. Another possible explanation is that static loading associated with loading differences between the hands may result in localized muscle fatigue of these muscles, increasing the number of muscle fibers recruited during the test session, thereby increasing the amplitude of the EMG signal. Fatigue was not specifically measured in this study, though it may be useful to look at this variable in future studies. Right arm mean FCU values were significant by gender. FCU values for females were significantly greater than values for males. This result may be due to posture or anthropometry of females, but might be explained by other factors. First, three of four females were right-handed. More arm movement is required for more animated key strikes, where momentum of the hand striking the keys is heightened with help from the lower arm. For a largely right-handed population, the right hand could have been used more forcefully than usual, which resulted in higher key strike forces. It was possible that this increased loading of the right arm contributed to greater right arm muscle activity. Simply, though, this result could be due to an uneven gender population through speculation. Gilad and Harel (2000) studied females only while testing EMG for a negative keyboard angle, thus no comparisons for gender population with EMG results can be made with previous studies. Key Strike Force Key strike force was not affected by keyboard angle. Figure V.7 shows relatively close trends in mean values for both keys by keyboard angle; moreover, N key strike forces were significantly greater than E key strike forces. There was a greater difference in the right index finger striking the N; key below home row than the left middle finger striking the E; key above the home row. These results could be attributed to the largely right-handed population, but it was more likely due to the location of the key. The right index finger is forced to curl a greater percentage of its range as Rose documented when striking the N key below home row than any other finger extending for a key above home row. Negative keyboard angles yielded lower mean values than the standard for the N key, though this finding was not significant. These results may
63
be attributed to the gradual decreasing of theta since the N key became easier to strike as keyboard angle decreased. Interestingly, E key strike forces increased almost linearly except for a decrease at the -30 keyboard angle. These results may be due to the greater extension required for the left hand to strike the E key, since it was above home row, and the possible subsequent lifting of the hand and arm to reach it, thus contributing heightened momentum to strike the key.
Left Hand 1.10 1.09 1.08 1.07 1.06 1.05 1.04 1.03 1.02 1.01 1.00 7 0 -10
Right Hand
Mean KSF values (N)
-20
-30
Keyboard Angle (deg)
Figure V.7. Mean KSF value results comparison by side Repeatability of Results For the most part, variables correlated extremely well by day. Most of the electrogoniometric data correlated strongly, as did key strike force, discomfort ratings, and most performance measures. However, repeatability was hard to achieve for certain electromyography variables. There is a great deal of variability in collecting electromyography data. One source of variability was MVC, which are discussed later as a limitation in this study. Aside from MVC, only the left hand mean FCU value did not correlate by day. All other EMG values showed strong or moderate correlations by day. The FCU muscle was the harder of the two muscles to mark for electrode placement. It was possible that this, among other limitations discussed later, resulted in these correlations. This study reinforced that results obtained from experiments using electromyography protocol
64
conducted once may not be completely accurate, and that a second trial, if possible, should be run to increase validity of results. Wrist Posture Plots Related to Risk It is known that brief exposure to a CTP of 30 mmHg in animals is sufficient to affect nerve functioning for prolonged periods of time (Lundborg et al., 1983). CTP values could be considered predictors of risk, and are estimated as such by the plot in Figure V.8. CTP values for wrist flexion and extension up to 45 were calculated by Rempel et al. (1997), and a relationship between CTP and fingertip force was found to fit a second-order polynomial for the neutral posture. Mean FE deviations were averaged for the two hands, and are plotted with values shown for each keyboard angle against the risk curve estimated by CTP according to Rempel et al. (1997). As can be seen from the figure, higher potential risk associated with increased CTPs due to deviated wrist postures are associated with less negative keyboard angles. No true risk curve exists for deviated wrist postures, though this figure does illustrate meaningful potential benefits associated with using negatively sloped keyboard angles, as opposed to the standard configuration.
Risk (Estimated by CTP (mmHg))
yrisk = 0.0068x + 0.1133x + 17.3619
2
Risk 32.41 28.75 21.14 7 0 -10 -20 -30 Poly. (Risk)
17.7
2.81
11.05
0
-45
-30
-15
0
15
30
45
(Flexion) Wrist Angle (deg) (Extension)
Figure V.8. Wrist posture related to risk, estimated by CTP (mmHg)
65
CHAPTER VI CONCLUSION This study sought to identify a negative keyboard angle, or range of angles, that minimized exposure to hypothesized hand and wrist WMSDs. Surprisingly, multiple objective and subjective measures revealed contradictory findings. Objective measures, such as wrist posture, favored negative angles of 0 or greater compared to 7, and in some cases, heavily supported -20 and -30. Certain EMG results supported negative angles, while other results favored the standard keyboard configuration. Net typing speed supported the -10 keyboard angle, while other negative typing angles were comparable, if not better, than the standard. However, other objective measures (i.e., key strike force) and the 7-point discomfort subjective rating showed no difference in the keyboard angles studied, whether positive or negative. However, it should be noted that discomfort ratings and all types of errors did decrease across two days of the study. Other subjective measures such as the 5-point self-rating, time estimates, and final rankings supported either 7 or a flat angle of 0, but could have been a confounded result based on user preferences. It was evident from these results that one keyboard angle could not be identified, based on the multiple measures used and varying results. Personal preference could decide which angle was best based on subjective measures; however, this study concludes that a typing angle within the range of 0 to -30 provides objective postural benefits, reduced muscle activity in some cases, with improved or analogous typing performance when compared to the standard. Limitations Maximum Voluntary Contractions Maximum voluntary contractions required for normalization of EMG data were a limitation in this study. This experiment was conducted twice on different days, and MVC values for the FCU muscle for both arms showed poor repeatability. The participant, experimenter, time of day, equipment, or electrode placement and application could have influenced these values. Electromyography data collection is
66
affected by any of these factors. It was possible that any of these factors contributed to less accurate readings than could have been obtained. Electrode Placement Electrode placement for the FCU muscle may have varied because markings made a week earlier were not readily apparent, possibly leading to weak correlations for some EMG variables. Over the weeks time, it was possible the permanent marks had worn away from the usual flaking and washing of the skin. However, shaved areas were obvious in most cases, which facilitated similar electrode placement according to Perotto (1994) and Soderberg (1992). Typing posture Many factors affected typing posture, making it a possible limitation in this study: (1) keyboard wedge design; (2) computer workstation; and (3) wrist rest. Keyboard Wedge Design Typing posture varied because of keyboard wedge design. The wedges used were made from wood to withstand multiple uses during the experiment over two weeks. Using this material made the wedges larger than if other materials were used, such as plastic or neoprene closed cell sponge or foam material. The larger wedges created some problems when the chair and the computer workstation needed to be adjusted for different participants. These effects were mainly postural, and can be attributed to limitations in the wedge design, as well as the computer workstation, which is explained hereafter. Computer Workstation In tandem with the keyboard wedge design, another limitation resulting in postural effects emanated from use of the computer workstation. The typing posture did not completely resemble the one that was intended for some extreme cases. When seating participants, the wrist point and elbow points at the ends of the ulna were used as reference marks to decipher if the forearm was parallel to the floor. Sometimes the
67
wrist point was slightly above the elbow point. These cases where the forearm was not completely parallel with the floor can be attributed mainly to the limitations in adjusting seat height and keyboard tray height, but were magnified if the anthropometrics of the participant tested its limits. Chair, workstation, and postural values were recorded in day 1 experiments to accurately repeat experiments on day 2. Trying to recreate the same typing posture was difficult because of these limitations in adjusting the posture, and varying keyboard height, seat height, and other measures may have contributed to the varying results by day. Wrist Rest In conjunction with keyboard wedge design and the computer workstation, lack of a wrist rest may have also affected typing posture. It was feared that a wrist rest would interfere with the natural movement of the hands while typing, and potential hazards with its use were forewarned by Simoneau and Marklin (2001). Hence, a wrist rest was removed from use in this study. Unknowingly, removal of a wrist rest may have contributed to a varied typing posture. Keeping the forearm parallel was difficult because of keyboard wedge design and adjusting chair and keyboard tray height, and the lack of a wrist rest may have exacerbated these limitations. Results showed discrepancies when compared with other studies, thus the absence of a wrist rest may have been a cause. Load Button Placement Placement of the load button inside the keyboard may have been a limitation. Ideally, to most accurately measure a key strike force, a load button should be placed in direct contact with the finger. Accomplishing this task was impossible due to the size of the load button used, and because the anonymity of the load buttons placement would have been compromised. To ensure the keyboard functioned properly, yet the participants were unaware what keys were being studied, the load buttons were placed according to Figure III.1. Doing so placed the load button underneath a circuit board, a plastic-domed sheet, and the plastic keycap. Each of these components possibly experienced deformations during key strike forces, or absorbed part of the key strike
68
forces, leading to potentially less accurate load button readings. However, since this same system was used for all conditions across day, keyboard angle, side, subject, and gender, the comparisons of the readings were consistent. Participants Participants in this study were not professional typists. The sample population was also skewed, due to scheduling conflicts, where six participants were female and the remaining four were male. This inequality may have resulted in gender differences that may not have existed with equal gender populations. These may have been limitations in the study. However, studies by Gilad and Harel (2000); Simoneau and Marklin (2001); Hedge and Morimoto (2001); Hedge et al. (1999); Hedge and Powers (1995); Hedge et al. (1995); Hedge (1994); Powers and Hedge (1992); Gerard et al. (1996); Gerard, Armstrong, Martin, Rempel (2002); and Gerard, Armstrong, Rempel, Woolley, (2002) studied all or primarily female participants. These studies either made no mention of gender differences or were unable to identify gender differences due to study populations of mostly females. This study provided interesting gender differences unlike previous literature. Discomfort Variable The discomfort variable in this study was potentially a confounding variable. The standard keyboard configuration has been widely used and accepted since the ANSI/HFES 100 1988 standard. All participants in this study were accustomed to this standard configuration. It was possible there was a natural negative response by the participants to typing at angles different than the standard, simply because these angles were different. A longitudinal or longer terms study is needed to investigate this finding. Future Research Use of a wrist rest with extreme negative angles should be explored further. It has been shown to affect typing posture, and may be a factor for finding an ideal negative typing angle. Simoneau and Marklin (2001) mentioned wrist extension angles may differ as much as 10 when a wrist rest is used. It would be beneficial to conduct
69
similar studies using keyboard wedges testing angles up to -30, perhaps at smaller increments, but with and without the use of a wrist rest. In addition, future studies should be aware that limitations in wedge design, computer workstation design, and wrist rests will affect typing posture, since participant anthropometry will be unknown; therefore, measures should be taken to ensure typing posture is consistent. It is possible to constrain the anthropometry in the study, but results may not be generalized to populations outside such constraints. Results for this study only applied to use of a standard QWERTY keyboard. Future research could be conducted to determine if results for the same negative angles span different types of keyboards, such as for natural keyboards. Future projects may be able to determine full body postural effects for one or varying keyboard angle(s) on the QWERTY keyboard, but also for different types of keyboards.
70
REFERENCES 1. Albin, T. (1997). Effect of wrist rest use and keyboard tilt on wrist angle while keying. Proceedings of the 13th Triennial Congress of the International Ergonomics Association, Tempere, Finland, 16-18. 2. Armstrong, T., Foulke, J., Martin, B., Gerson, J., and Rempel, D. (1994). Investigation of applied forces in alphanumeric keyboard work. American Industrial Hygiene Association Journal, 55, 30-35. 3. Arndt, R. (1983). Working posture and musculoskeletal problems of video display terminal operatorsreview and reappraisal. American Industrial Hygiene Association Journal, 44(6), 437-446. 4. Andersson, G. (1987). Biomechanical aspects of sitting: An application to VDT terminals. Behaviour and Information Technology, 6(3), 257-269. 5. Barr, A., Goldsheyder, D., zkaya, N., and Nordin, M. (2001). Testing apparatus and experimental procedure for position specific normalization of electromyographic measurements of distal upper extremity musculature. Clinical Biomechanics, 16, 576-585. 6. Bergqvist, U., Wolgast, E., Nilsson, B., and Voss, M. (1995). Musculoskeletal disorders among VDT workers: Individual, ergonomic, and work organizational factors. Ergonomics, 38, 763-776. 7. Brain, W., Wright, A., and Wilkinson, M. (1947). Spontaneous compression of both median nerves in the carpal tunnel. The Lancet, March, 277-282. 8. Buchholz, B. and Wellman, H. (1997). Practical operation of a biaxial goniometer at the wrist joint. Human Factors, 39(1), 119-129. 9. Carter, J. and Banister, E. (1994). Musculoskeletal problems in VDT work: a review. Ergonomics, 37(10), 1623-1648. 10. Corlett, E. and Bishop, R. (1972). A technique for assessing postural discomfort. Ergonomics, 19(2), 175-182. 11. DeKrom, M., Kester, A., Knipschild, P., and Spaans, F. (1990). Risk factors for carpal tunnel syndrome. American Journal of Epidemiology, 132(6), 1102-1110. 12. Duncan, J. and Ferguson, D. (1974). Keyboard operating posture and symptoms in operating. Ergonomics, 17(5), 651-662. 13. Dvorak, A. (1936). Typewriting behavior. New York: American Book. 14. Faucett, J. and Rempel, D. (1994). VDT-related musculoskeletal symptoms: Interaction between work posture and psychosocial work factors. American Journal of Industrial Medicine, 26, 597-612. 15. Feuerstein, M., Armstrong, T., and Hickey, P. (1994). Keyboard force, upper extremity symptoms, perceived effort and mood in symptomatic and asymptomatic word processors. Proceedings of the 12th Triennial Congress of the International Ergonomics Association, Toronto, Ontario, Canada. 71
16. Fisher, R. (1973). Statistical methods for research workers (14th ed.). New York, NY. Hafner Press. 17. Gelberman, R., Szabo, R., and Mortenson, W. (1984). Carpal tunnel pressures and wrist position in patients with Colles fractures. Journal of Trauma, 24, 747-749. 18. Gerard, M., Armstrong, T., Rempel, D., and Woolley, C. (2002). Short term and long term effects of enhanced auditory feedback on typing force, EMG, and comfort while typing. Applied Ergonomics, 33, 129-138. 19. Gerard, M., Armstrong, T., Martin, B., and Rempel, D. (2002). The effects of work pace on within-participant and between-participant keying force, electromyography, and fatigue. Human Factors, 44(1), 51-61. 20. Gerard, M., Armstrong, T., Foulke, J., and Martin, B. (1996). Effects of key stiffness on force and the development of fatigue while typing. American Industrial Hygiene Association Journal, 57, 849-854. 21. Gilad, I. and Harel, S. (2000). Muscular effort in four keyboard designs. International Journal of Industrial Ergonomics, 26(1), 1-7. 22. Glasscock, N., Turville, K., Joines, S., and Mirka, G. (1999). The effect of personality type on muscle coactivation during elbow flexion. Human Factors, 41(1), 51-60. 23. Hedge, A. and Morimoto, S. (2001). Intervention for prevention/treatment of carpal tunnel syndrome. CUErgoCornell University Ergonomics Web. http://ergo.human.cornell.edu/hedge.html (18 Sep. 2001). 24. Hedge, A., Morimoto, S., and McCrobie, D. (1999). Effects of keyboard tray geometry on upper body posture and comfort. Ergonomics, 42(10), 1333-1349. 25. Hedge, A. and Powers, J. (1995). Wrist postures while keyboarding: effects of a negative slope keyboard system and full motion forearm supports. Ergonomics, 38(3), 508-517. 26. Hedge, A., McCrobie, D., Land, B., Morimoto, S., and Rodriguez, S. (1995). Healthy keyboarding: Effects of wrist rests, keyboard trays, and a preset tiltdown system on wrist posture, seated posture, and musculoskeletal discomfort. Proceedings of the Human Factors and Ergonomics Society 39th Annual Meeting, Santa Monica, CA, 630-634. 27. Hedge, A. (1994). Effects of a negative slope keyboard system on musculoskeletal complaints among computer workers [Abstract]. International Ergonomics Conference '94, Vol. 4, Aug. 15-19, Toronto, Canada, p. 462. 28. Honan, M., Jacobson, M., Tal, R., and Rempel, D. (1996). Changes in wrist postures during a prolonged typing task. Proceedings of the Human Factors and Ergonomics Society 40th Annual Meeting, Santa Monica, CA, 629-631. 29. Honan, M., Serina, E., Tal, R., and Rempel, D. (1995). Wrist postures while typing on a standard and split keyboard. Proceedings of the Human Factors and Ergonomics Society 39th Annual Meeting, Santa Monica, CA, 366-368.
72
30. Horie, S., Hargens, A., and Rempel, D. (1993). Effect of keyboard wrist rest in preventing carpal tunnel syndrome [Abstract]. Proceedings of the American Public Health Association Annual Meeting, San Francisco, CA, p. 319. 31. Human Factors Society (1988). The American National Standard for Human Factors Engineering Requirements for Video Display Terminal Workstations, ANSI/HFES 100-1988. Santa Monica, CA: Human Factors Society. 32. Juul-Kristensen, B., Fallentin, N., Hansson, G., Madeleine, P., Andersen, J., and Ekdahl, C. (2002). Physical workload during manual and mechanical deboning of poultry. International Journal of Industrial Ergonomics, 29, 107-115. 33. Kuorinka, I., Jonsson, B., Kilbom, A., Vinterberg, H., Biering-Sorensen, F., Andersson, G., et al. (1987). Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Applied Ergonomics, 18(3), 233-237. 34. Kuschner, S., Ebramzadeh, E., Johnson, D., Brien, W., and Sherman, R. (1992). Tinels sign and Phalens test in carpal tunnel syndrome. Orthopedics, 15(11), 12971302. 35. Life, M. and Pheasant, S. (1984). An integrated approach to the study of posture in keyboard operation. Applied Ergonomics, 15, 83-90. 36. Lundborg, G., Myers, R., and Powell, H. (1983). Nerve compression injury and increased endoneurial fluid pressure: A miniature compartment syndrome. Journal of Neurol Neurosurg Psychiatry, 46, 1119-1124. 37. Marklin, R., Simoneau, G., and Monroe, J. (1999). Wrist and forearm posture from typing on split and vertically inclined keyboards. Human Factors, 41(4), 559-569. 38. Marklin, R., Simoneau, G., and Monroe, J. (1997a). The effect of split and verticallyinclined computer keyboards on wrist and forearm posture. Proceedings of the Human Factors and Ergonomics Society 41st Annual Meeting, Santa Monica, CA, 642-646. 39. Marklin, R., Simoneau, G., and Monroe, J. (1997b). An ergonomics study of alternative keyboard designs. National Institute of Occupational Safety and Health, Center for Disease Control, National Institute of Health. Doc No. 5 R03 OH0318402. 40. Marras, W. and Schoenmarklin, R. (1991). Wrist motions and CTD risk in industrial and service environments. In Y. Quinnec, F. Daniellou, A. Chapon, P. Falzon, J. Leplat, B. Metz, H. Monod, J. C. Sperandio, and A. Wisner (Eds.), Designing for everyone: Vol. 1. Proceedings of the 11th Congress of the International Ergonomics Association (pp. 36-38). New York: Taylor & Francis. 41. Marras, W. (1990). Guidelines: Industrial electromyography (EMG). International Journal of Industrial Ergonomics, 6, 89-93. 42. Marshall, M., Mozrall, J., and Shealy, J. (1999). The effects of complex wrist and forearm posture on wrist range of motion. Human Factors, 41(2), 205-213.
73
43. Martin, B., Rempel, D., Sudarsan, P., Dennerlein, J., Jacobson, M., Gerard, M., et al. (1998). Reliability and sensitivity of methods to quantify muscle load during keyboard work. Advances in Occupational Ergonomics and Safety, 495-498. 44. Martin, B., Armstrong, T., Foulke, J., Natarajan, S., Klinenberg, E., Serina, E., et al. (1994). Finger force during computer keyboard work: Part I. Relation of keyboard reaction force to finger flexor muscles surface EMG. Proceedings of the 12th Triennial Congress of the International Ergonomics Association, Toronto, Ontario, Canada, 198-200. 45. Parsons, C. (1991). Use of wrist rests by data input VDU operators. In E. J. Lovesey (Ed.), Contemporary Ergonomics (pp. 319-321). London: Taylor & Francis. 46. Paul, R., Lueder, R., Selner, A., and Limaye, J. (1996). Impact of new input technology on design of chair armrests: Investigation on keyboard and mouse. Proceedings of the Human Factors and Ergonomics Society 40th Annual Meeting, Santa Monica, CA, 380-384. 47. Paul, R. and Menon, K. (1994). Ergonomic evaluation of wrist pads. Proceedings of the 12th Triennial Congress of the International Ergonomics Association, Toronto, Ontario, Canada, 204-207. 48. Perotto, A. (Ed.). (1994). Anatomical guide for the electromyographer: The limbs and trunk (3rd ed.). Charles C. Thomas, Springfield, IL. 49. Phalen, G. (1966). The carpal-tunnel syndrome. The Journal of Bone and Joint Surgery, 48A(2), 211-228. 50. Powers, J. and Hedge, A. (1992). Effects of full motion forearm supports and a negative slope keyboard system on hand-wrist posture while keyboarding. Proceedings of the Human Factors Society 36th Annual Meeting, Santa Monica, CA, 796-800. 51. Radwin, R. (1997). Activation force and travel effects on overexertion in re...