- Email: [email protected]
Threshold Detection and Semmes-Weinstein Monofilaments
Threshold Detection and Semmes-Weinstein Monofilaments Judith A. Bell-Krotoski, OTR, FAOTA, CHT CAPT, U.S. Public Health Service, Chief, Hand and Occupational Therapy Department, Clinical Research Therapist, Rehabilitation Research Department, Gillis W. Long Hansen's Disease Center, Carville, Louisiana
Elaine Ewing Fess, MS, OTR, FAOTA, CHT Hand Research, Zionsville, Indiana
John H. Figarola, OTR
LT, U.S. Public Health Service, Acting Deputy Chief, Hand and Occupational Therapy Department, Gillis W. Long Hansen's Disease Center, Carville, Louisiana
Danell Hiltz, MOT, OT/L Alberquerque, New Mexico
he threshold of detection for human tactile T cutaneous receptors has been measured with
as low as a few milligrams of force. 1.2 Under various circumstances, the receptors exhibit even greater sensitivity. Most clinical test instruments used for measurement of cutaneous sensibility have been found to greatly exceed this sensitivity of the tactile cutaneous system by several orders of magnitude. 3A However, vibration produced by the examiner's hand alone greatly exceeds the normal sensitivity theshold of human touch when applied with a handheld instrument lacking control of variables. The amount of hand vibration varies widely among examiners. While seemingly insignificant at first, the normal vibration and movement of the examiner's hand can be seen as highly significant when measured on strain gauges designed to be sensitive enough to measure force applied by an instrument in the normal detection range. It is impossible for a consistent repeatable Correspondence and reprint requests to Judith A. Bell-Krotoski, OTR, FAOTA, CHT; CAPT, U.S. Public Health Service, Chief, Hand and Occupational Therapy Department, Clinical Research Therapist, Gillis W. Long Hansen's Disease Center, 5445 Point Clair Road, Carville, LA 70721-9607.
ABSTRACT: Semmes-Weinstein monofilaments provide a repeatable instrument stimulus with a small standard deviation in contrast to other handheld test instruments, making them an optimum choice for objective sensory testing in a variety of clinics. Normal sensory detection thresholds for the entire body, and the stimulus force for each filament, were determined by Weinstein. He found a nylon filament of 0.005 in wide and 38 mm long (mean force, 68 mg) to be a good predictor of "normal" light touch-deep pressure threshold for the hands and most of the body. However, manufacturers of the nylon used in making the filaments allow an 8-10% tolerance in diameter. This small change in diameter can result in small variations in mean force among filaments of a given size. It has not been previously determined what effect this small variance in force has on the accuracy of the 2.83 (marking number) D.005-in wide filament most often used for normal threshold detection. This study compared the 2.83 filaments available at the Gillis W. Long Hansen's Disease Center, which have a mean force of 62 mg, with those from North Coast Medical, Inc., which have a mean force of 95 mg. The filaments were used by 6 examiners in a standard testing protocol for the hands, arms, faces, legs, and feet of 130 subjects. Heavier and lighter filaments of measured force were also included. Results showed a high correlation in responses for two values for the 2.83 filaments in the range speCified. On detailed analysis between kits there were some differences for site and age. The 2.83 filament was a good predictor of normal in both kits for the hand, the arm, and the leg, as was expected. It was suprathreshold for the face, making use of a lighter filament possible for facial testing, and subthreshold for the plantar surface of the foot, which required a slightly heavier filament (3.61 marking number, mean force, 279 mg) for normal threshold detection as measured in this study. J HAND THER 8:155-162, 1995.
force to be applied by hand in this range, and the vibration of the examiner's hand often overpowers and masks the intended force. The Semmes-Weinstein monofilament test remains the only handheld instrument specifically designed to control application force variables, and to meet sensitivity and repeatability requirements for an objective test instrument when calibrated correctly.S-lO It too, however, needs the elimination of any complicating variables in testing and substantion of its validity in detecting normal and abnormal subjects in order that it can be unquestionably relied on in clinical testing. Normal light touch-deep pressure threshold detection values need to be reconfirmed in currently available test kits. The optimum detection level for normal subjects also needs to be well defined.
BACKGROUND In the pursuit of validity, tests must have certain essential properties. Variables of handheld tests have been described that limit their providing repeatable stimuli, even under the best of conditions. Dyck April-June 1995
155
et al. described the variables inherent to sensibility testing instruments in 1976, whereas Bell-Krotoski and Buford described and reported the variables in commonly used handheld test instruments in 1981.1,3 In the latter study, the instruments commonly used for sensibility testing varied widely in applied stimuli, and therefore lacked the sensitivity, specificity, and repeatability necessary for reliability. Instrument reliability is necessary before an instrument can be determined to have validity in testing with clinical subjects. The Semmes-Weinstein monofilaments were found to be an exception in that they are specifically designed to control force and the vibration of the examiner's hand when applied against the skin. The filaments have been found to produce repeatable stimuli within very small standard deviations (50s) relative to the other tests, as long as they are used in the same way and are calibrated correctlyY Traditionally, the 2.83 filament has been the index for normal in male and female subjects. Optimum detection thresholds for determining normal response versus abnormal response on the entire body were established through extensive testing of normal and abnormal subjects by Sidney Weinstein, coinventor of the filament test with Josephine Semmes in the 1960s.1O,12-15 Many subjects responded to even lighter filaments, and women were found to be slightly more sensitive than men. Weinstein found the mean force produced by the 2.83 filament to be 68 mg. This is slightly lighter than the mean force in kits now produced by the current largest supplier, North Coast Medical, Inc. (NCM). Although the nylon material used as the stimulus has remained the same over the years, kits are now available through a variety of suppliers, The means and 50s for 28 NCM filament kits were reported by Bell-Krotoski and Tomancik in 1987. 11 In this study, the 2.83 filaments had a mean force of 80 mg (SO, 0.007). Bell-Krotoski obtained samples of nylon material from all available manufacturers in an attempt to determine reasons for the slight differences in force produced by the 2.83 and other filaments, and also to determine whether a given size filament could be made even more accurate .1 6 It was found that the only change in the production of the nylon over the years has been that the diameter of the filament material has become laser controlled during manufacture. This should improve control of its size. However, it was also found that even one filament of a given size can vary slightly along its length, accounting for slight differences in its applied force. In addition, filament material manufacturers allow an 810% tolerance on the relative filament diameter when the filament is made. Small differences in diameter mean that separate orders of material can result in slightly different batches within a range. It is not yet known what effect these small differences in applied force have on detection levels, or which end of rangeslightly heavier or lighter-would be optimal. The differences in the Gillis W. Long Hansen's Disease Center (GWLHDC) and the NCM values provide the opportunity to determine whether there is a difference in detection, particularly for the most important 156
JOURNAL OF HAND THERAPY
filament-the 2.83 filament. The SO in applied force discussed here is in milligrams for the 2,83 and lighter filaments, while the SO in force of uncontrolled handheld instruments is considerably higher, measuring in grams. Therefore, this article addresses questions regarding improvements in Semmes-Weinstein monofilament control, not whether the monofilaments are controlled, It is the most controlled of any of the handheld sensory testing instruments used in the clinical setting, Manufacturers of the nylon material have the ability to provide a more stringent control on filament size, but will do so only when it becomes cost-effective. Fortunately, the material in one order is of such quantity that the material for a given supplier is relatively consistent for years. Diameter and force variations make it most important that filament kit sources are reported, calibration is measured, and force values are reported in studies using the filaments, particularly in normative studies. Otherwise, a clinical study cannot be relied on to have used filaments of correct values. Filaments of the wrong diameter or approaching the size of the next heavier or lighter diameter should most certainly be discarded and not used in any study. Bowen et al. tested normal subjects with 3 NCM filament kits in 1988 and found the 2.83 filament (measured mean, 72 mg; SO, 5.50) to be a good predictor of normal for the hands. I? They found the lighter, 2.44 filament (measured mean, 33 mg; SO, 4.16) to be an equally good predictor of normal, and questioned whether this filament would perhaps be even better for screening normal subjects. This suggests that the optimum force detection value for normal may be toward the lighter side, closer to the original 68 mg reported by Weinstein, rather than toward the heavier NCM filaments. The heavier end of the range for normal is closer to the force of the next heavier filament, which has been observed to map areas consistent with known innervations of specific peripheral nerves. The next heavier filament in the standard 20-filament kit, 3.22 marking number, has a measured mean of 172 mg in the standard NCM kits (20 filaments) and 166 mg as reported by Weinstein. Considering that the original mean for the 2.83 filament specified by Weinstein was 68 mg while NCM had slightly higher values (its 2.83 filament was custom manufactured to have a perfect 0.005-in diameter), Bell-Krotoski and Tomancik recommended in 1989 that producers of filament kits at the GWLHDC accept the standard diameter of 0.005 in (not custom manufactured) and the normally occurring small-range means and 50s in kits made for measurement of their patients in the United States and overseas. 11 For the thousands of filaments needed overseas, it was not possible to measure the force of each individual filament; therefore, it was most important to obtain consistent material. Since the force produced is related to the length and diameter, filaments of the same length and diameter should reproduce the same force within a measurable range. Although the material obtained by the GWLHOC was double processed for size, any manipulation of the diameter was
deemed likely to make it relatively impossible for orders to be replicated later and for material to be standard among various kit suppliers. Minimum specifications for standard filaments were needed for an American Leprosy Mission funded international study, "Monitoring of Peripheral Nerve Involvement Underlying Disability of the Hand in Hansen's Disease," with the first author as principal investigator. After testing preproduction samples, filaments for GWLHOC use were obtained in specified sizes (in thousandths of an inch). The lightest, heaviest, and random samples in-between of each filament size were measured for conformity within a certain range, and means and SOs were calculated.
PURPOSE In this study the original detection thresholds determined by Weinstein are revisited: 1. The 2.83 filament available at the GWLHOC was compared with the 2.83 filament available through NCM to determine whether the small differences in their means do not significantly affect the results of testing in normal subjects. 2. The 2.83 filament was compared with heavier and lighter filaments above and below threshold to confirm the 2.83 as an optimum predictor of normal for the hand and the entire arm. 3. Test kits from two different instrument suppliers were used at sites all over the body to confirm the 2.83 filament as a good predictor of normal all over the body, except at the plantar surface of the foot, which requires a slightly heavier filament (3.61).
METHOD Instrument Measurement System Each filament used in the study was tested for calibration using an instrument measurement system
specifically designed by a biomedical engineer to measure the dynamic forces produced by the filaments (Fig. 1).3.4 Means and SOs were determined by direct measurement of filament force. The instrument measurement system used for testing the calibration of the Semmes-Weinstein monofilaments consisted of a strain gauge, a signal filter and an amplifier, and an oscilloscope. The oscilloscope was used to read the signal where it could be frozen and quantified in milligrams or grams per division. Filaments were tested at the beginning of the study, before kits were sent to a new site for testing of patients, and at the end of the study to ascertain whether the filament application force had changed during the study. The filaments were 38 mm long. The diameters of the filaments were measured using a micrometer.
Test Instrument The design of the Semmes-Weinstein filaments produces progressively increasing force stimuli by increasing the diameter of nylon monofilament material. Nylon has optimum viscoelastic properties: it bends when applied in a perpendicular manner to the skin, and bend recovery absorbs the vibration of the examiner's hand. Through its elasticity, nylon maintains a constant force throughout its bend until lifted, and applied force immediately ceases on lifting. In this way, a normal or an absent absolute response to filament weight can be determined, as well as a baseline detection threshold that can be used for subsequent comparison. Six Semmes-Weinstein monofilament style sets were used; 3 produced in the GWLHOC patient sheltered workshop and 3 produced by NCM. The NCM kits contained filaments from the long kit from 1.65 through 4.31. The GWLHOC kit contained 5 filaments in a mini kit. The GWLHOC kits and the NCM kits were numbered 1 to 3, and the test kit used was randomized for each subject. The order of testing (GWLHOC or NCM kit first) was randomized prior to testing. All patients were tested with both the GWLHOC and the NCM filament kits at each site tested.
Filament Bending Forces
FIGURE 1. This instrument measurement system, which was specifically designed by a biomedical engineer to measure the dynamic forces produced by filaments, was used to test the calibration of each filament used ill this study.
The mean force for the GWLHOC 2.83 filament was 62 mg (SO, 2.28 mg, ranges 38-80 mg); the corresponding NCM filament had a mean force of 95 mg (SO, 4.34 mg; range, 90-104 mg). The mean force for the GWLHOC 3.61 filament was 279 mg (SO, 1.9 mg; range, 271-286 mg); the corresponding NCM filament had a mean force of 375 mg (SO, 10.2 mg; range, 336-400 mg). All GWLHOC filaments were measured at 80°F (26.7°C) and all NCM filaments at 78°F (25°C), making their relative measures within 2°F (1. 1°C). April-June 1995
157
Examiners
RESULTS
Testing was done by 6 experienced examiners, in 3 states (Indiana , Texas, and Louisiana). The upper extremities, in particular the hands, were tested for all subjects. When possible, the lower extremities and the face were also tes ted by the same examiner so that a relative comparison could be made for the additional sites tested. A total of 130 upper extremities, 92 lower extremities, and 130 faces were tested. In Indiana, 59 subjects were tested at 1 site by 1 examiner. In Texas, 39 were tested at 2 sites by 1 examiner. In Louisiana, 32 were tested at 2 sites by 4 examiners. (Examiner 1 tested 20 subjects, examiner 2 tested 10, and examiners 3 and 4 tested one each.)
GWLHDC Filament Kits versus NCM Filament Kits
Subjects One hundred thirty subjects were included for testing of the hand and upper extremity (260 upper extremities, with 2 tests for each arm, totaling 520 measurements at each site on the hand and slightly fewer for the upper extremity as a whole). Ninetytwo subjects were included for testing of the foot and lower ex tremity (184 lower extremities, with 2 tests for each leg, totaling 368 measurements at each site on the foot and lower extremity as a whole) . One hundred thirty subjects were included for testing of the face (2 tests for each subject, totaling 260 measurements at each site on the face). Of the 130 subjects, 58 were male and 72 were female. Three males and one female who had been initially included were eliminated due to a prior history of numbness or stroke, or due to incomplete tests. The mean age was 25 years (range, 9-85 years). Seventy-two subjects were younger than the mean age and 56 were older than the mean age (information was not available for one subject) . Children and young adults were included in order to determine the effects of age.
Test Protocol The following sites were included in testing: 6 on the volar surface of the hand, 1 on the dorsum of the hand, and 3 on the arms; 6 on the volar surface of the foot, 1 on the dorsum of the foot, and 2 on the legs; and 4 on the face. These sites were tested in a standard protocol, which included establishing a normal recognition site in close proximity as a comparative reference. IS Each filament was applied 3 times at each site, and each site was returned to 3 times. A "no" response was counted when there was no response to a filament. When there was 1 response out of 3, the response was counted as a "yes."
Statistics For comparison of the GWLHDC filament kits with the NCM filament kits, Pearson's correlation coefficients were used. A two-way analysis of variance (p = 0.05) was used for all other comparisons. 158
JOURNAL OF HAND THERAPY
Using Pearson's correlation coefficients for direct comparison of the GWLHDC filament responses with the NCM filament responses, there was a very high correlation between the GWLHDC and the NCM affirmative response to the 2.83 filament in the population tested (right side of the body, r = 0.92; left side of the body, r = 0.94). Values from the GWLHDC response reflected a tendency for that kit to be slightly more sensitive (NCM will include a few more subjects as normal). There was a high correlation between the GWLHDC and NCM affirmative responses to the 2.83 filament for age (GWLHDC vs NCM < 25, r = 0.92; NCM vs GWLHDC 25 or >, r = 0.92) and sex (GWLHDC male vs NCM female, r = 0.86; NCM male vs GWLHDC female, r = 0.84). There were some significant differences for site . Thus, based on these findings either kit can be used, but the test is most accurate if the same manufacturer is used. Within kits the NCM filament response showed the body's left side to have slightly higher values in the palm and upper extremity (two-way analysis of variance). Right-side and left-side values for the GWLHDC kits were not significantly different. That the left side was slightly more sensitive than the right in response to the NCM filaments is consistent with findings reported by Weinstein.
Threshold Response at Sites for the 2.83 Filament-GWLHDC Threshold data for the GWLHDC 2.83 filament (62 mg) are summarized in Figure 2A, where percent response is shown for each site. Of 130 subjects, on the dorsum of the hand and arm (not including the palm) the detection was 94% on the right (range, 8598%) and 98% on the left (range, 96-100%). On the volar surface of the palm, the detection was 93% on the right (range, 88-97%) and 94% on the left (range, 88-96%). Of 92 subjects, on the dorsum of the foot and leg (not including the volar aspect of the foot) the detection was 96% on the right (range, 95-97%) and 97% on the left (range, 96-98%). On the volar aspect Qf the foot the detection was 76% on the right (range, 38-89%) and 73% on the left (range, 28-93%). The low end of the range was the heel; the high, the instep. Of 130 subjects, detection was 100% (range, 99100%) on the face. Results of analysis of threshold response to the 2.83 filament are summarized in Table l. As seen in this study, the 2.83 filament can be a useful screen for normal light-touch recognition all over the body, with the exception of the plantar sur-' face of the foot. Here detection falls too low and a slightly heavier filament is needed. The 3.61 filament is optimal for the foot, as detection then improves to 98% on the right (the lowest value being the heel).
TABLE 1.
Analysis of Threshold Response to the Gillis W. Long Hansen's Disease Center (GWLHDC) Monofilament for Extremity, Site, Age, and Sex (Two-way Analysis of Variance, 0.05 Level of Significance)
No significant difference was found between upper-extremity (UE) (arms and hands) affirmative responses for right (R) and left (L) .sides (Figs. 2A and B).
mative responses for Rand L sides. No significant difference was detected between arm and leg affirmative responses for a given site.
There was a significant difference in UE (arms and hands) affirmative responses for sites.
No significant difference was seen between lower extremity (LE) (leg and foot) affirmative responses for Rand L sides (Figs. 2C and D).
No significant difference was detected between arm and dorsum of the hand affirmative responses for Rand L sides (Fig. 2A).
There was a significant difference between LE (legs and feet) affirmative responses for a site.
No significant difference was found between arm and dorsum of the hand affirmative responses for a site.
There was no significant difference between the leg and dorsum of the foot affirmative responses for Rand L sides or site (Fig. 2C).
No Significant difference was found between palmar affirmative responses for Rand L sides (Fig. 2B). There was a significant difference between palmar affirmative responses for a site. The ulnar palm and proximal little finger were significantly different sites; no significant difference was found when these sites were eliminated.
No significant difference was seen between plantar affirmative responses for Rand L sides (Fig. 2D). There was a significant difference between plantar affirmative responses for a site. The heel was the lowest value and instep the highest (Fig. 2D).
There was no significant difference between arm and leg affir-
85199
Upper Extremity
96199
N =130
•
9 If95 •
89/93
Left
Right
Right
A
Left
B
Lower Extremity N=92
FIGURE 2. Threshold data for the Gillis W. Long Hansen's Disease Cen ter (GWLHDC) 2.83 (62 mg) monofilament/ North Coast Medical (NCM) 2.83 (95 mg). The percent response is shown for each site
97
of the upper extremities (n = 130) and the lower extremities (n = 92).
•
) 96/98
•
98/99
•
98
(
•
95/96
Left
Right
D
96/97
Face
Left
Right
N= 130
C
99/100
E April-June 1995
159
Right
198%1
Left
FIGURE 3. Threshold data for the Gillis W. Long Hansen's Disease Center (GWLHDC) 3.61 (279 mg) monofilament for the plantar surface of the foot.
199%1
Threshold Response at Sites for the 3.61 Filament-GWLHDC For the 3.61 filament (279 mg), detection was 99-100% at each of these sites, indicating that the 3.61 filament is heavier than necessary for detecting subjects who have normal responses. Of 130 subjects, on the dorsum of the hand and arm (not including the palm) the detection was 99% on the right (range, 98-100%) and 99% on the left (range, 98-100%). On the volar surface of the palm, the detection was 99% on the right (range, 99-100%) and 99% on the left (range, 99-100%). Of 92 subjects, on the dorsum of the foot and leg (not including the volar aspect of the foot) the detection was 100% on the right and 100% on the left. On the volar aspect of the foot the detection was 98% on the right (range, 94-100%) and 99% on the left (range, 91-100%) (Fig. 3). The low end of the range was the heel; the high, the instep. Of 130 subjects, detection was 100% (100% range) on the face.
DISCUSSION Detection Threshold In regard to detection sensitivity, the reader should realize that 100% detection is not the optimum is psychometric testing. 19 •2o Detection threshold can be set at any level, but the optimum is the greatest detection with the fewest false-positions. The 100% level could be used, for instance, but the result would be to lose sensitivity and eliminate subjects who have early change. A detection around 90% is good in that while it may include a few subjects for follow-up who are "within normal limits," it will catch a greater number of subjects who are in early stages of change from normal. It means that usually a normal subject will feel the filament if a few attempts are made, not that every subject will feel the filament 100% of the time. Subjects can be mapped on the face using the 2.83 filament; however, it yields 100% detection here and, depending on whether there is a history of a problem, the use of lighter filaments should be considered. That the 2.83 filament is optimal for the screening of normal subjects for the arm and most of the body is further substantiated by the fact that a threshold difference between the 2.83 and 3.61 filaments has been seen to map out early areas of nerve 160
JOURNAL OF HAND THERAPY
change. 18 •21 If the heavier, 3.61 filament were used as the normal detection level, patients who had early change would be eliminated. For patients who have no history of problems and are within normal limits in threshold testing with the 2.83 filament, some examiners go one step further and use the lighter abovethreshold filaments to determine whether they can detect a differential in peripheral nerve innervation areas, e.g., an ulnar nerve versus a median nerve. This can be helpful in specific cases, if it can be established that a subject can generally detect lighter filaments everywhere except in the problem area. For most clinical testing, however, it is more important to establish whether a patient is normal, not whether a patient can feel better than normal. The 2.83 filament (62 mg in this study), when calibrated correctly, is the most helpful for this purpose. A filament greater than 3.61 (279 mg) is not needed anywhere on the body. If we are careful to differentiate a false-positive response from a true detection, we have to make sure the subject actually has a chance to feel the filament. The lightest filament can sometimes be applied too lightly. If misapplied, unless bounced against the skin, their force values are too light, not too heavy, because they bend at a peak force. For this reason, the lighter filaments are applied three times (and Weinstein now recommends as many as five times) at a site unresponsive to a filament in order to increase test accuracy. Patients generally respond to the first application; however, fewer than three times for an unresponsive area has been found to produce less accurate data and to increase the number of false-positives. Three applications at an unresponsive site was used in this study, and is recommended as a standard in test protocols. The results would have been different if only one application were made, and perhaps heavier filaments would have been better predictors of normal. Three applications for the lighter filaments (lighter than 4.07) becomes increasingly important when examiners are screening specific sites for each nerve, as in this study. When a full mapping is done, if one response is missed, another is given in close proximity for the same nerve or area. The only argument against applying the filaments three times at an unresponsive site is the consideration regarding summation of touch. There is absolutely no evidence to support the concept that detection threshold can be changed if the filaments are applied more than once, i.e., that summation in light touch-deep pressure testing is a problem. LaMotte tell us that summation is a problem for pain and temperature testing, not for light touch-deep pressure testing. 22 Conversely, the test would not be a good one if learning occurred, and to deliver an inappropriate threshold for testing with a threshold test would be foolish. Weinstein also investigated detection thresholds for other types of sensibility tests and instruments; including localization and two-point discrimination. He designed the filament test to improve on the recognized limitations of available testing instruments. Weinstein found the 2.83 (68 mg) filament to be a
good predictor of normal male subjects over most of the body. Females were found to be even slightly more sensitive on the hand. For testing females alone, one lighter filament, 2.44 (28 mg), could be used. Von Prince and Butler, Werner and Orner, and Bell-Krotoski all have extensively investigated use of the filaments with peripheral nerve injuries and are in agreement regarding the 2.83 filament as a predictable indicator of normal. 18,23,24 Clinical mapping with the 2.83 filament and both heavier and lighter filaments (in an ordinal ranking according to the force they produce) confirms that the 2.83 filament is a good reference for normal. If this filament is not detected, then heavier filaments detected most often, if not always, map out specifiC areas consistent with nerve branch innervation. 18 A differential mapping of abnormal areas is a most important value of the test, second only to determining a normal or abnormal threshold response, and supports the validity of the test in defining clinical morphology. In fact, if this mapping at different force levels were not able to be done, the test would be of little use. The differential mapping based on application force detection also supports the importance of controlled force with any sensory testing instrument. That variation in force can affect the results of sensibility testing is clearly shown and proven by differential mapping of levels of response and function at specific force levels. 18 ,21,23,24 Dellon, long an advocate of two-point discrimination, and Levin et al. have criticized the SemmesWeinstein monofilaments for being reported in force, rather than in pressure. 25.26 Force is much easier for the examiner to understand and appreciate. Pressure is simply force per unit area. While it is true that the area of each filament increases with the size of the filament (and the force it provides), it is actually less accurate to refer to the filaments in pressure units rather than in force units. Once the filament is bent in contact with the skin (which must be done to deliver a constant force threshold), it is a crescentshaped edge through which the force is applied, not a specified diameter area. Therefore, calculations of area to determine the pressure are then inaccurate, as this edge in contact cannot be directly determined. One filament of a given size is always the same diameter; therefore, the area of a given filament size is a constant. While a constant tip on all of the filaments theoretically might improve the instrument, it is relatively impossible to measure the exact area in contact with the skin when doing clinical testing at this time. The possibility of placing one constant tip on the filaments has been investigated. Von Frey27 did put a constant tip on his horsehair filaments, which precedes the Semmes-Weinstein filaments. However, von Frey worked within a narrow range of light filaments with normal subjects. Today the test must produce a range of forces sufficient in sensitivity to measure patients as they first fall off from normal, and sufficient in strength to measure deep-pressure sensation to indicate whether there is residual nerve function perhaps worth treating. The same range needed
to measure diminishing nerve function is needed to measure function as it returns. It has been impossible to reproduce the whole range necessary with a constant tip. A tip that would fit a larger filament would not produce light forces if placed on the lighter filaments, and a tip light enough to fit the lighter filaments would increase the pressure of the larger filaments by its small area. A broad tip is too wide to achieve the lightest thresholds needed, and a small tip is too thin to achieve the heavier stimulus without eliciting a painful response. The new West instrument developed by Weinstein was designed as an improvement in the filaments. It has a rounded tip for each of the "minikit" filaments. Weinstein has reported slightly more repeatable data with this instrument. In addition to the rounded tip and the placement of five-in-one filaments in the same handle as suggested by Bell, each filament is certified for specific calibration. 16 This is not done by any other manufacturer at this time (most companies specify the length and the diameter and depend on them for a repeatable force). Given the small range in variation of the force produced by variations in diameter, this may not make much difference in determining abnormal patients in clinical testing, and other kits are good for monitoring. Due to small changes that can occur in any instrument, it is always better when the same kit is used for follow-up measurements. The change in tip of the West instrument really makes it another instrument, and is necessary to conduct clinical trials to determine whether the interpretation scales based on it are identical to those of the original style. Nonetheless, it offers the promise of improvement in calibration and may be a welcome addition to a clinical evaluation of sensibility. Based on findings of this study with statistically significant differences for age and site with small changes in a monofilament so sensitive, it is advisable to stay within the same kit manufacturer for testing the same subjects. Differences need further investigation to ensure an optimumly calibrated instrument. Filament forces must be measured and reported for clinical studies. Once needed limits and optimum stimuli levels have been defined, it is unknown whether computerized instruments will outdo the usefulness, portability, and practicality available with the Semmes-Weinstein monofilament test. 4 ,22,28,29 Additionally, the computerized stimulus is a different stimulus, and it will need to be shown to have an advantage over the simple monofilament test in demonstrating changes in clinical morphology. Today there is a need for a handheld test to do this objectively and cost effectively. The Semmes-Weinstein monofilaments can achieve the sensitivity and repeatability necessary for a reliable instrument when calibrated correctly and are currently available for clinical use. Acknowledgments The authors thank Lillian B. Browder, MEd, for help with statistical analysis and Jerry Simmons for help with illustrations.
April-June 1995
161
REFERENCES 1. Dyck pJ, O'Brien pc, Bushek W, e t al: Clinical vs quantitative evaluation of cutaneous sensa tion. Arch Neurol 33:651 -656, 1976. 2. Mountcastle VB, LaMotte RH , Carli C: Detection thresholds for stimuli in humans and monkeys: Comparison with threshold events in mechanoreceptive afferent nerve fibers innervating the monkey hand. J N europhysiol 35:122-136, 1972. 3. Bell-Krotoski JA, Buford WL Jr: The force/time relationship of clinically used sensory testin g instruments. J Hand Ther 1:7685, 1988. 4. Buford WL, Bell JA: Dynamic properties of the hand held tactile assessment stimuli. In Proceedings of the 34th Annual Conference on Engineering in Medicine and Biology, Houston , Texas, September 21-23, 1981. 5. Bell-Krotoski J: A study of peripheral nerve involvement underlying physical disability of the hand in Hansen's disease. J Hand Ther 5:133-142, 1992. 6. Bell-Krotoski J, Weinstein S, Weinstein C: Testing sensibility, including touch-pressure, two-point discrimination, point localization, and vibration. J H and Ther 6:114-123, 1993. 7. Fess E: Evaluating the hand by objective measures. In Hunter JM, Schneider LH, Mackin EJ, Bell JA (eds): Rehabilitation of th e Hand. St. Louis, C. V. Mosby, 1978, p. 5. 8. Fess E: The need for reliability and validity in hand assessm ent instruments. J Hand Surg lAm] 11 :621-623, 1986. 9. Johnston MV, Keith RA, Hinderer SR, Gonnella C: Measurement standards for interdisciplina ry medical rehabilitation . In Archives of Physical Medicine and Rehabilitation. Philadelphia, W. B. Saunders, 1992. 10. Weinstein S: Fifty years of somatosensory research: From the Semmes-Weinstein monofilaments to the Weinstein Enhanced Sensory Test. J Hand Ther 6:1, 11-28, 1993. 11.. Bell-Krotoski JA, Tomancik E: Repeatability of the SemmesWeinstein monofilaments. J H and Surg [Am] 12:155-161, 1987. 12. Semmes J, Weinstein S, Ghent L, Tueber HL: Somatosensory Changes after Penetrating Brain Wounds in Man. Cambridge, MA, Harvard University Pres, 1960. 13. Weinstein S: Intensive and extensive aspects of tactile sensitivity as a function of bod y part, sex, and laterality. In Ke nshalo DR (ed): The Skin Senses. New York, Plenum Press, 1968, pp. 195-218.
162
JOURNAL OF HAND THERAPY
14. Weinstein S: Tactile sensitivity of the phalanges. Perce pt Mot Skills 14:354-354, 1962. 15. Weinstein S, Seren E: Tactile sensitivity as a function of ha ndedness and laterality. J Comp Physiol Psychol 54:665 - 669, 1961. 16. Bell-Krotoski JA: Pocket filaments and specifications for the Semmes-Weinstein monofilaments. III Amadio PC, Hentz VR (eds): Year Book of H and Surgery. St. Louis, C. V. Mosby, 1992, pp. 293-294. 17. Bowen VL, Griener JS, Jones SV: Threshold of sensation: Interrater reliability and establishment of normal using the SemmesWeinstein monofilament (abstract). J Hand Ther 3:36-37, 1990. 18. Bell-Krotoski JA: Sensibility testing: Current concepts. III Hunter JM, Schneider LH, Mackin EJ, Callahan AD (eds): Rehabilitation of the Hand , 4th ed. St. Louis, C. V. Mosby, 1995, pp. 1- 19. 19. Eiijkman E, Vendrick AJH: Detection theory applied to the absolute sensitivity of sensory systems. Biophys J 3:65-78, 1963. 20. Weinstein SW, Drozdenko R, Weinstein C: Some Considerations of Clinical Tests, Presentation and Discussion. Danbury, CT, NeuroCommunications Research Laboratories, 1992. 21. Waylett-Rendall J: Sensibility evaluation and rehabilitation. Orthop Clin North Am 9:43-56, 1988. 22. LaMotte RH: Personal communication, 1992. 23. Von Prince K, Butler B: Measuring sensory function of th e hand in peripheral nerv e injuries. Am J Occup Ther 21:385396, 1967. 24. Werner JL, Orner GE: Evalu ating cutaneous pressure sensitivity of the hand. Am J Occup Ther 24:347 -356, 1970. 25. Levin S, Pearsall G, Ruderman RJ: Von Frey's method of measuring pressure sensitivity in the hand: An engineering analysis of the Semmes-Weinstein pressure aesthiometer. J Hand Surg [Am] 3:211 , 1978. 26. Dellon AL: The sensational contributions of Erik Moberg. J Hand Surg [Br] 15:14- 24, 1990. 27. Von Frey M: Zur Physiologie der Juckemptin dung. Arch Neurol Physiol 7:142, 1922. 28. Dyck PJ, Zimmerman lR, O'Brien Pc, et al: Introduction of automated systems to evaluate touch-pressure, vibration, and thermal cutaneous sensation in man. Ann Neurol 4:502 - 510, 1978. 29. Horch K, Hardy M, Jimenez S, Jabaley J: An automated tactile tes ter for evaluation of cutaneous sensibility. J Hand Surg [Am] 17:5, 829- 837, 1992.
Elaine Ewing Fess, MS, OTR, FAOTA, CHT Hand Research, Zionsville, Indiana
John H. Figarola, OTR
LT, U.S. Public Health Service, Acting Deputy Chief, Hand and Occupational Therapy Department, Gillis W. Long Hansen's Disease Center, Carville, Louisiana
Danell Hiltz, MOT, OT/L Alberquerque, New Mexico
he threshold of detection for human tactile T cutaneous receptors has been measured with
as low as a few milligrams of force. 1.2 Under various circumstances, the receptors exhibit even greater sensitivity. Most clinical test instruments used for measurement of cutaneous sensibility have been found to greatly exceed this sensitivity of the tactile cutaneous system by several orders of magnitude. 3A However, vibration produced by the examiner's hand alone greatly exceeds the normal sensitivity theshold of human touch when applied with a handheld instrument lacking control of variables. The amount of hand vibration varies widely among examiners. While seemingly insignificant at first, the normal vibration and movement of the examiner's hand can be seen as highly significant when measured on strain gauges designed to be sensitive enough to measure force applied by an instrument in the normal detection range. It is impossible for a consistent repeatable Correspondence and reprint requests to Judith A. Bell-Krotoski, OTR, FAOTA, CHT; CAPT, U.S. Public Health Service, Chief, Hand and Occupational Therapy Department, Clinical Research Therapist, Gillis W. Long Hansen's Disease Center, 5445 Point Clair Road, Carville, LA 70721-9607.
ABSTRACT: Semmes-Weinstein monofilaments provide a repeatable instrument stimulus with a small standard deviation in contrast to other handheld test instruments, making them an optimum choice for objective sensory testing in a variety of clinics. Normal sensory detection thresholds for the entire body, and the stimulus force for each filament, were determined by Weinstein. He found a nylon filament of 0.005 in wide and 38 mm long (mean force, 68 mg) to be a good predictor of "normal" light touch-deep pressure threshold for the hands and most of the body. However, manufacturers of the nylon used in making the filaments allow an 8-10% tolerance in diameter. This small change in diameter can result in small variations in mean force among filaments of a given size. It has not been previously determined what effect this small variance in force has on the accuracy of the 2.83 (marking number) D.005-in wide filament most often used for normal threshold detection. This study compared the 2.83 filaments available at the Gillis W. Long Hansen's Disease Center, which have a mean force of 62 mg, with those from North Coast Medical, Inc., which have a mean force of 95 mg. The filaments were used by 6 examiners in a standard testing protocol for the hands, arms, faces, legs, and feet of 130 subjects. Heavier and lighter filaments of measured force were also included. Results showed a high correlation in responses for two values for the 2.83 filaments in the range speCified. On detailed analysis between kits there were some differences for site and age. The 2.83 filament was a good predictor of normal in both kits for the hand, the arm, and the leg, as was expected. It was suprathreshold for the face, making use of a lighter filament possible for facial testing, and subthreshold for the plantar surface of the foot, which required a slightly heavier filament (3.61 marking number, mean force, 279 mg) for normal threshold detection as measured in this study. J HAND THER 8:155-162, 1995.
force to be applied by hand in this range, and the vibration of the examiner's hand often overpowers and masks the intended force. The Semmes-Weinstein monofilament test remains the only handheld instrument specifically designed to control application force variables, and to meet sensitivity and repeatability requirements for an objective test instrument when calibrated correctly.S-lO It too, however, needs the elimination of any complicating variables in testing and substantion of its validity in detecting normal and abnormal subjects in order that it can be unquestionably relied on in clinical testing. Normal light touch-deep pressure threshold detection values need to be reconfirmed in currently available test kits. The optimum detection level for normal subjects also needs to be well defined.
BACKGROUND In the pursuit of validity, tests must have certain essential properties. Variables of handheld tests have been described that limit their providing repeatable stimuli, even under the best of conditions. Dyck April-June 1995
155
et al. described the variables inherent to sensibility testing instruments in 1976, whereas Bell-Krotoski and Buford described and reported the variables in commonly used handheld test instruments in 1981.1,3 In the latter study, the instruments commonly used for sensibility testing varied widely in applied stimuli, and therefore lacked the sensitivity, specificity, and repeatability necessary for reliability. Instrument reliability is necessary before an instrument can be determined to have validity in testing with clinical subjects. The Semmes-Weinstein monofilaments were found to be an exception in that they are specifically designed to control force and the vibration of the examiner's hand when applied against the skin. The filaments have been found to produce repeatable stimuli within very small standard deviations (50s) relative to the other tests, as long as they are used in the same way and are calibrated correctlyY Traditionally, the 2.83 filament has been the index for normal in male and female subjects. Optimum detection thresholds for determining normal response versus abnormal response on the entire body were established through extensive testing of normal and abnormal subjects by Sidney Weinstein, coinventor of the filament test with Josephine Semmes in the 1960s.1O,12-15 Many subjects responded to even lighter filaments, and women were found to be slightly more sensitive than men. Weinstein found the mean force produced by the 2.83 filament to be 68 mg. This is slightly lighter than the mean force in kits now produced by the current largest supplier, North Coast Medical, Inc. (NCM). Although the nylon material used as the stimulus has remained the same over the years, kits are now available through a variety of suppliers, The means and 50s for 28 NCM filament kits were reported by Bell-Krotoski and Tomancik in 1987. 11 In this study, the 2.83 filaments had a mean force of 80 mg (SO, 0.007). Bell-Krotoski obtained samples of nylon material from all available manufacturers in an attempt to determine reasons for the slight differences in force produced by the 2.83 and other filaments, and also to determine whether a given size filament could be made even more accurate .1 6 It was found that the only change in the production of the nylon over the years has been that the diameter of the filament material has become laser controlled during manufacture. This should improve control of its size. However, it was also found that even one filament of a given size can vary slightly along its length, accounting for slight differences in its applied force. In addition, filament material manufacturers allow an 810% tolerance on the relative filament diameter when the filament is made. Small differences in diameter mean that separate orders of material can result in slightly different batches within a range. It is not yet known what effect these small differences in applied force have on detection levels, or which end of rangeslightly heavier or lighter-would be optimal. The differences in the Gillis W. Long Hansen's Disease Center (GWLHDC) and the NCM values provide the opportunity to determine whether there is a difference in detection, particularly for the most important 156
JOURNAL OF HAND THERAPY
filament-the 2.83 filament. The SO in applied force discussed here is in milligrams for the 2,83 and lighter filaments, while the SO in force of uncontrolled handheld instruments is considerably higher, measuring in grams. Therefore, this article addresses questions regarding improvements in Semmes-Weinstein monofilament control, not whether the monofilaments are controlled, It is the most controlled of any of the handheld sensory testing instruments used in the clinical setting, Manufacturers of the nylon material have the ability to provide a more stringent control on filament size, but will do so only when it becomes cost-effective. Fortunately, the material in one order is of such quantity that the material for a given supplier is relatively consistent for years. Diameter and force variations make it most important that filament kit sources are reported, calibration is measured, and force values are reported in studies using the filaments, particularly in normative studies. Otherwise, a clinical study cannot be relied on to have used filaments of correct values. Filaments of the wrong diameter or approaching the size of the next heavier or lighter diameter should most certainly be discarded and not used in any study. Bowen et al. tested normal subjects with 3 NCM filament kits in 1988 and found the 2.83 filament (measured mean, 72 mg; SO, 5.50) to be a good predictor of normal for the hands. I? They found the lighter, 2.44 filament (measured mean, 33 mg; SO, 4.16) to be an equally good predictor of normal, and questioned whether this filament would perhaps be even better for screening normal subjects. This suggests that the optimum force detection value for normal may be toward the lighter side, closer to the original 68 mg reported by Weinstein, rather than toward the heavier NCM filaments. The heavier end of the range for normal is closer to the force of the next heavier filament, which has been observed to map areas consistent with known innervations of specific peripheral nerves. The next heavier filament in the standard 20-filament kit, 3.22 marking number, has a measured mean of 172 mg in the standard NCM kits (20 filaments) and 166 mg as reported by Weinstein. Considering that the original mean for the 2.83 filament specified by Weinstein was 68 mg while NCM had slightly higher values (its 2.83 filament was custom manufactured to have a perfect 0.005-in diameter), Bell-Krotoski and Tomancik recommended in 1989 that producers of filament kits at the GWLHDC accept the standard diameter of 0.005 in (not custom manufactured) and the normally occurring small-range means and 50s in kits made for measurement of their patients in the United States and overseas. 11 For the thousands of filaments needed overseas, it was not possible to measure the force of each individual filament; therefore, it was most important to obtain consistent material. Since the force produced is related to the length and diameter, filaments of the same length and diameter should reproduce the same force within a measurable range. Although the material obtained by the GWLHOC was double processed for size, any manipulation of the diameter was
deemed likely to make it relatively impossible for orders to be replicated later and for material to be standard among various kit suppliers. Minimum specifications for standard filaments were needed for an American Leprosy Mission funded international study, "Monitoring of Peripheral Nerve Involvement Underlying Disability of the Hand in Hansen's Disease," with the first author as principal investigator. After testing preproduction samples, filaments for GWLHOC use were obtained in specified sizes (in thousandths of an inch). The lightest, heaviest, and random samples in-between of each filament size were measured for conformity within a certain range, and means and SOs were calculated.
PURPOSE In this study the original detection thresholds determined by Weinstein are revisited: 1. The 2.83 filament available at the GWLHOC was compared with the 2.83 filament available through NCM to determine whether the small differences in their means do not significantly affect the results of testing in normal subjects. 2. The 2.83 filament was compared with heavier and lighter filaments above and below threshold to confirm the 2.83 as an optimum predictor of normal for the hand and the entire arm. 3. Test kits from two different instrument suppliers were used at sites all over the body to confirm the 2.83 filament as a good predictor of normal all over the body, except at the plantar surface of the foot, which requires a slightly heavier filament (3.61).
METHOD Instrument Measurement System Each filament used in the study was tested for calibration using an instrument measurement system
specifically designed by a biomedical engineer to measure the dynamic forces produced by the filaments (Fig. 1).3.4 Means and SOs were determined by direct measurement of filament force. The instrument measurement system used for testing the calibration of the Semmes-Weinstein monofilaments consisted of a strain gauge, a signal filter and an amplifier, and an oscilloscope. The oscilloscope was used to read the signal where it could be frozen and quantified in milligrams or grams per division. Filaments were tested at the beginning of the study, before kits were sent to a new site for testing of patients, and at the end of the study to ascertain whether the filament application force had changed during the study. The filaments were 38 mm long. The diameters of the filaments were measured using a micrometer.
Test Instrument The design of the Semmes-Weinstein filaments produces progressively increasing force stimuli by increasing the diameter of nylon monofilament material. Nylon has optimum viscoelastic properties: it bends when applied in a perpendicular manner to the skin, and bend recovery absorbs the vibration of the examiner's hand. Through its elasticity, nylon maintains a constant force throughout its bend until lifted, and applied force immediately ceases on lifting. In this way, a normal or an absent absolute response to filament weight can be determined, as well as a baseline detection threshold that can be used for subsequent comparison. Six Semmes-Weinstein monofilament style sets were used; 3 produced in the GWLHOC patient sheltered workshop and 3 produced by NCM. The NCM kits contained filaments from the long kit from 1.65 through 4.31. The GWLHOC kit contained 5 filaments in a mini kit. The GWLHOC kits and the NCM kits were numbered 1 to 3, and the test kit used was randomized for each subject. The order of testing (GWLHOC or NCM kit first) was randomized prior to testing. All patients were tested with both the GWLHOC and the NCM filament kits at each site tested.
Filament Bending Forces
FIGURE 1. This instrument measurement system, which was specifically designed by a biomedical engineer to measure the dynamic forces produced by filaments, was used to test the calibration of each filament used ill this study.
The mean force for the GWLHOC 2.83 filament was 62 mg (SO, 2.28 mg, ranges 38-80 mg); the corresponding NCM filament had a mean force of 95 mg (SO, 4.34 mg; range, 90-104 mg). The mean force for the GWLHOC 3.61 filament was 279 mg (SO, 1.9 mg; range, 271-286 mg); the corresponding NCM filament had a mean force of 375 mg (SO, 10.2 mg; range, 336-400 mg). All GWLHOC filaments were measured at 80°F (26.7°C) and all NCM filaments at 78°F (25°C), making their relative measures within 2°F (1. 1°C). April-June 1995
157
Examiners
RESULTS
Testing was done by 6 experienced examiners, in 3 states (Indiana , Texas, and Louisiana). The upper extremities, in particular the hands, were tested for all subjects. When possible, the lower extremities and the face were also tes ted by the same examiner so that a relative comparison could be made for the additional sites tested. A total of 130 upper extremities, 92 lower extremities, and 130 faces were tested. In Indiana, 59 subjects were tested at 1 site by 1 examiner. In Texas, 39 were tested at 2 sites by 1 examiner. In Louisiana, 32 were tested at 2 sites by 4 examiners. (Examiner 1 tested 20 subjects, examiner 2 tested 10, and examiners 3 and 4 tested one each.)
GWLHDC Filament Kits versus NCM Filament Kits
Subjects One hundred thirty subjects were included for testing of the hand and upper extremity (260 upper extremities, with 2 tests for each arm, totaling 520 measurements at each site on the hand and slightly fewer for the upper extremity as a whole). Ninetytwo subjects were included for testing of the foot and lower ex tremity (184 lower extremities, with 2 tests for each leg, totaling 368 measurements at each site on the foot and lower extremity as a whole) . One hundred thirty subjects were included for testing of the face (2 tests for each subject, totaling 260 measurements at each site on the face). Of the 130 subjects, 58 were male and 72 were female. Three males and one female who had been initially included were eliminated due to a prior history of numbness or stroke, or due to incomplete tests. The mean age was 25 years (range, 9-85 years). Seventy-two subjects were younger than the mean age and 56 were older than the mean age (information was not available for one subject) . Children and young adults were included in order to determine the effects of age.
Test Protocol The following sites were included in testing: 6 on the volar surface of the hand, 1 on the dorsum of the hand, and 3 on the arms; 6 on the volar surface of the foot, 1 on the dorsum of the foot, and 2 on the legs; and 4 on the face. These sites were tested in a standard protocol, which included establishing a normal recognition site in close proximity as a comparative reference. IS Each filament was applied 3 times at each site, and each site was returned to 3 times. A "no" response was counted when there was no response to a filament. When there was 1 response out of 3, the response was counted as a "yes."
Statistics For comparison of the GWLHDC filament kits with the NCM filament kits, Pearson's correlation coefficients were used. A two-way analysis of variance (p = 0.05) was used for all other comparisons. 158
JOURNAL OF HAND THERAPY
Using Pearson's correlation coefficients for direct comparison of the GWLHDC filament responses with the NCM filament responses, there was a very high correlation between the GWLHDC and the NCM affirmative response to the 2.83 filament in the population tested (right side of the body, r = 0.92; left side of the body, r = 0.94). Values from the GWLHDC response reflected a tendency for that kit to be slightly more sensitive (NCM will include a few more subjects as normal). There was a high correlation between the GWLHDC and NCM affirmative responses to the 2.83 filament for age (GWLHDC vs NCM < 25, r = 0.92; NCM vs GWLHDC 25 or >, r = 0.92) and sex (GWLHDC male vs NCM female, r = 0.86; NCM male vs GWLHDC female, r = 0.84). There were some significant differences for site . Thus, based on these findings either kit can be used, but the test is most accurate if the same manufacturer is used. Within kits the NCM filament response showed the body's left side to have slightly higher values in the palm and upper extremity (two-way analysis of variance). Right-side and left-side values for the GWLHDC kits were not significantly different. That the left side was slightly more sensitive than the right in response to the NCM filaments is consistent with findings reported by Weinstein.
Threshold Response at Sites for the 2.83 Filament-GWLHDC Threshold data for the GWLHDC 2.83 filament (62 mg) are summarized in Figure 2A, where percent response is shown for each site. Of 130 subjects, on the dorsum of the hand and arm (not including the palm) the detection was 94% on the right (range, 8598%) and 98% on the left (range, 96-100%). On the volar surface of the palm, the detection was 93% on the right (range, 88-97%) and 94% on the left (range, 88-96%). Of 92 subjects, on the dorsum of the foot and leg (not including the volar aspect of the foot) the detection was 96% on the right (range, 95-97%) and 97% on the left (range, 96-98%). On the volar aspect Qf the foot the detection was 76% on the right (range, 38-89%) and 73% on the left (range, 28-93%). The low end of the range was the heel; the high, the instep. Of 130 subjects, detection was 100% (range, 99100%) on the face. Results of analysis of threshold response to the 2.83 filament are summarized in Table l. As seen in this study, the 2.83 filament can be a useful screen for normal light-touch recognition all over the body, with the exception of the plantar sur-' face of the foot. Here detection falls too low and a slightly heavier filament is needed. The 3.61 filament is optimal for the foot, as detection then improves to 98% on the right (the lowest value being the heel).
TABLE 1.
Analysis of Threshold Response to the Gillis W. Long Hansen's Disease Center (GWLHDC) Monofilament for Extremity, Site, Age, and Sex (Two-way Analysis of Variance, 0.05 Level of Significance)
No significant difference was found between upper-extremity (UE) (arms and hands) affirmative responses for right (R) and left (L) .sides (Figs. 2A and B).
mative responses for Rand L sides. No significant difference was detected between arm and leg affirmative responses for a given site.
There was a significant difference in UE (arms and hands) affirmative responses for sites.
No significant difference was seen between lower extremity (LE) (leg and foot) affirmative responses for Rand L sides (Figs. 2C and D).
No significant difference was detected between arm and dorsum of the hand affirmative responses for Rand L sides (Fig. 2A).
There was a significant difference between LE (legs and feet) affirmative responses for a site.
No significant difference was found between arm and dorsum of the hand affirmative responses for a site.
There was no significant difference between the leg and dorsum of the foot affirmative responses for Rand L sides or site (Fig. 2C).
No Significant difference was found between palmar affirmative responses for Rand L sides (Fig. 2B). There was a significant difference between palmar affirmative responses for a site. The ulnar palm and proximal little finger were significantly different sites; no significant difference was found when these sites were eliminated.
No significant difference was seen between plantar affirmative responses for Rand L sides (Fig. 2D). There was a significant difference between plantar affirmative responses for a site. The heel was the lowest value and instep the highest (Fig. 2D).
There was no significant difference between arm and leg affir-
85199
Upper Extremity
96199
N =130
•
9 If95 •
89/93
Left
Right
Right
A
Left
B
Lower Extremity N=92
FIGURE 2. Threshold data for the Gillis W. Long Hansen's Disease Cen ter (GWLHDC) 2.83 (62 mg) monofilament/ North Coast Medical (NCM) 2.83 (95 mg). The percent response is shown for each site
97
of the upper extremities (n = 130) and the lower extremities (n = 92).
•
) 96/98
•
98/99
•
98
(
•
95/96
Left
Right
D
96/97
Face
Left
Right
N= 130
C
99/100
E April-June 1995
159
Right
198%1
Left
FIGURE 3. Threshold data for the Gillis W. Long Hansen's Disease Center (GWLHDC) 3.61 (279 mg) monofilament for the plantar surface of the foot.
199%1
Threshold Response at Sites for the 3.61 Filament-GWLHDC For the 3.61 filament (279 mg), detection was 99-100% at each of these sites, indicating that the 3.61 filament is heavier than necessary for detecting subjects who have normal responses. Of 130 subjects, on the dorsum of the hand and arm (not including the palm) the detection was 99% on the right (range, 98-100%) and 99% on the left (range, 98-100%). On the volar surface of the palm, the detection was 99% on the right (range, 99-100%) and 99% on the left (range, 99-100%). Of 92 subjects, on the dorsum of the foot and leg (not including the volar aspect of the foot) the detection was 100% on the right and 100% on the left. On the volar aspect of the foot the detection was 98% on the right (range, 94-100%) and 99% on the left (range, 91-100%) (Fig. 3). The low end of the range was the heel; the high, the instep. Of 130 subjects, detection was 100% (100% range) on the face.
DISCUSSION Detection Threshold In regard to detection sensitivity, the reader should realize that 100% detection is not the optimum is psychometric testing. 19 •2o Detection threshold can be set at any level, but the optimum is the greatest detection with the fewest false-positions. The 100% level could be used, for instance, but the result would be to lose sensitivity and eliminate subjects who have early change. A detection around 90% is good in that while it may include a few subjects for follow-up who are "within normal limits," it will catch a greater number of subjects who are in early stages of change from normal. It means that usually a normal subject will feel the filament if a few attempts are made, not that every subject will feel the filament 100% of the time. Subjects can be mapped on the face using the 2.83 filament; however, it yields 100% detection here and, depending on whether there is a history of a problem, the use of lighter filaments should be considered. That the 2.83 filament is optimal for the screening of normal subjects for the arm and most of the body is further substantiated by the fact that a threshold difference between the 2.83 and 3.61 filaments has been seen to map out early areas of nerve 160
JOURNAL OF HAND THERAPY
change. 18 •21 If the heavier, 3.61 filament were used as the normal detection level, patients who had early change would be eliminated. For patients who have no history of problems and are within normal limits in threshold testing with the 2.83 filament, some examiners go one step further and use the lighter abovethreshold filaments to determine whether they can detect a differential in peripheral nerve innervation areas, e.g., an ulnar nerve versus a median nerve. This can be helpful in specific cases, if it can be established that a subject can generally detect lighter filaments everywhere except in the problem area. For most clinical testing, however, it is more important to establish whether a patient is normal, not whether a patient can feel better than normal. The 2.83 filament (62 mg in this study), when calibrated correctly, is the most helpful for this purpose. A filament greater than 3.61 (279 mg) is not needed anywhere on the body. If we are careful to differentiate a false-positive response from a true detection, we have to make sure the subject actually has a chance to feel the filament. The lightest filament can sometimes be applied too lightly. If misapplied, unless bounced against the skin, their force values are too light, not too heavy, because they bend at a peak force. For this reason, the lighter filaments are applied three times (and Weinstein now recommends as many as five times) at a site unresponsive to a filament in order to increase test accuracy. Patients generally respond to the first application; however, fewer than three times for an unresponsive area has been found to produce less accurate data and to increase the number of false-positives. Three applications at an unresponsive site was used in this study, and is recommended as a standard in test protocols. The results would have been different if only one application were made, and perhaps heavier filaments would have been better predictors of normal. Three applications for the lighter filaments (lighter than 4.07) becomes increasingly important when examiners are screening specific sites for each nerve, as in this study. When a full mapping is done, if one response is missed, another is given in close proximity for the same nerve or area. The only argument against applying the filaments three times at an unresponsive site is the consideration regarding summation of touch. There is absolutely no evidence to support the concept that detection threshold can be changed if the filaments are applied more than once, i.e., that summation in light touch-deep pressure testing is a problem. LaMotte tell us that summation is a problem for pain and temperature testing, not for light touch-deep pressure testing. 22 Conversely, the test would not be a good one if learning occurred, and to deliver an inappropriate threshold for testing with a threshold test would be foolish. Weinstein also investigated detection thresholds for other types of sensibility tests and instruments; including localization and two-point discrimination. He designed the filament test to improve on the recognized limitations of available testing instruments. Weinstein found the 2.83 (68 mg) filament to be a
good predictor of normal male subjects over most of the body. Females were found to be even slightly more sensitive on the hand. For testing females alone, one lighter filament, 2.44 (28 mg), could be used. Von Prince and Butler, Werner and Orner, and Bell-Krotoski all have extensively investigated use of the filaments with peripheral nerve injuries and are in agreement regarding the 2.83 filament as a predictable indicator of normal. 18,23,24 Clinical mapping with the 2.83 filament and both heavier and lighter filaments (in an ordinal ranking according to the force they produce) confirms that the 2.83 filament is a good reference for normal. If this filament is not detected, then heavier filaments detected most often, if not always, map out specifiC areas consistent with nerve branch innervation. 18 A differential mapping of abnormal areas is a most important value of the test, second only to determining a normal or abnormal threshold response, and supports the validity of the test in defining clinical morphology. In fact, if this mapping at different force levels were not able to be done, the test would be of little use. The differential mapping based on application force detection also supports the importance of controlled force with any sensory testing instrument. That variation in force can affect the results of sensibility testing is clearly shown and proven by differential mapping of levels of response and function at specific force levels. 18 ,21,23,24 Dellon, long an advocate of two-point discrimination, and Levin et al. have criticized the SemmesWeinstein monofilaments for being reported in force, rather than in pressure. 25.26 Force is much easier for the examiner to understand and appreciate. Pressure is simply force per unit area. While it is true that the area of each filament increases with the size of the filament (and the force it provides), it is actually less accurate to refer to the filaments in pressure units rather than in force units. Once the filament is bent in contact with the skin (which must be done to deliver a constant force threshold), it is a crescentshaped edge through which the force is applied, not a specified diameter area. Therefore, calculations of area to determine the pressure are then inaccurate, as this edge in contact cannot be directly determined. One filament of a given size is always the same diameter; therefore, the area of a given filament size is a constant. While a constant tip on all of the filaments theoretically might improve the instrument, it is relatively impossible to measure the exact area in contact with the skin when doing clinical testing at this time. The possibility of placing one constant tip on the filaments has been investigated. Von Frey27 did put a constant tip on his horsehair filaments, which precedes the Semmes-Weinstein filaments. However, von Frey worked within a narrow range of light filaments with normal subjects. Today the test must produce a range of forces sufficient in sensitivity to measure patients as they first fall off from normal, and sufficient in strength to measure deep-pressure sensation to indicate whether there is residual nerve function perhaps worth treating. The same range needed
to measure diminishing nerve function is needed to measure function as it returns. It has been impossible to reproduce the whole range necessary with a constant tip. A tip that would fit a larger filament would not produce light forces if placed on the lighter filaments, and a tip light enough to fit the lighter filaments would increase the pressure of the larger filaments by its small area. A broad tip is too wide to achieve the lightest thresholds needed, and a small tip is too thin to achieve the heavier stimulus without eliciting a painful response. The new West instrument developed by Weinstein was designed as an improvement in the filaments. It has a rounded tip for each of the "minikit" filaments. Weinstein has reported slightly more repeatable data with this instrument. In addition to the rounded tip and the placement of five-in-one filaments in the same handle as suggested by Bell, each filament is certified for specific calibration. 16 This is not done by any other manufacturer at this time (most companies specify the length and the diameter and depend on them for a repeatable force). Given the small range in variation of the force produced by variations in diameter, this may not make much difference in determining abnormal patients in clinical testing, and other kits are good for monitoring. Due to small changes that can occur in any instrument, it is always better when the same kit is used for follow-up measurements. The change in tip of the West instrument really makes it another instrument, and is necessary to conduct clinical trials to determine whether the interpretation scales based on it are identical to those of the original style. Nonetheless, it offers the promise of improvement in calibration and may be a welcome addition to a clinical evaluation of sensibility. Based on findings of this study with statistically significant differences for age and site with small changes in a monofilament so sensitive, it is advisable to stay within the same kit manufacturer for testing the same subjects. Differences need further investigation to ensure an optimumly calibrated instrument. Filament forces must be measured and reported for clinical studies. Once needed limits and optimum stimuli levels have been defined, it is unknown whether computerized instruments will outdo the usefulness, portability, and practicality available with the Semmes-Weinstein monofilament test. 4 ,22,28,29 Additionally, the computerized stimulus is a different stimulus, and it will need to be shown to have an advantage over the simple monofilament test in demonstrating changes in clinical morphology. Today there is a need for a handheld test to do this objectively and cost effectively. The Semmes-Weinstein monofilaments can achieve the sensitivity and repeatability necessary for a reliable instrument when calibrated correctly and are currently available for clinical use. Acknowledgments The authors thank Lillian B. Browder, MEd, for help with statistical analysis and Jerry Simmons for help with illustrations.
April-June 1995
161
REFERENCES 1. Dyck pJ, O'Brien pc, Bushek W, e t al: Clinical vs quantitative evaluation of cutaneous sensa tion. Arch Neurol 33:651 -656, 1976. 2. Mountcastle VB, LaMotte RH , Carli C: Detection thresholds for stimuli in humans and monkeys: Comparison with threshold events in mechanoreceptive afferent nerve fibers innervating the monkey hand. J N europhysiol 35:122-136, 1972. 3. Bell-Krotoski JA, Buford WL Jr: The force/time relationship of clinically used sensory testin g instruments. J Hand Ther 1:7685, 1988. 4. Buford WL, Bell JA: Dynamic properties of the hand held tactile assessment stimuli. In Proceedings of the 34th Annual Conference on Engineering in Medicine and Biology, Houston , Texas, September 21-23, 1981. 5. Bell-Krotoski J: A study of peripheral nerve involvement underlying physical disability of the hand in Hansen's disease. J Hand Ther 5:133-142, 1992. 6. Bell-Krotoski J, Weinstein S, Weinstein C: Testing sensibility, including touch-pressure, two-point discrimination, point localization, and vibration. J H and Ther 6:114-123, 1993. 7. Fess E: Evaluating the hand by objective measures. In Hunter JM, Schneider LH, Mackin EJ, Bell JA (eds): Rehabilitation of th e Hand. St. Louis, C. V. Mosby, 1978, p. 5. 8. Fess E: The need for reliability and validity in hand assessm ent instruments. J Hand Surg lAm] 11 :621-623, 1986. 9. Johnston MV, Keith RA, Hinderer SR, Gonnella C: Measurement standards for interdisciplina ry medical rehabilitation . In Archives of Physical Medicine and Rehabilitation. Philadelphia, W. B. Saunders, 1992. 10. Weinstein S: Fifty years of somatosensory research: From the Semmes-Weinstein monofilaments to the Weinstein Enhanced Sensory Test. J Hand Ther 6:1, 11-28, 1993. 11.. Bell-Krotoski JA, Tomancik E: Repeatability of the SemmesWeinstein monofilaments. J H and Surg [Am] 12:155-161, 1987. 12. Semmes J, Weinstein S, Ghent L, Tueber HL: Somatosensory Changes after Penetrating Brain Wounds in Man. Cambridge, MA, Harvard University Pres, 1960. 13. Weinstein S: Intensive and extensive aspects of tactile sensitivity as a function of bod y part, sex, and laterality. In Ke nshalo DR (ed): The Skin Senses. New York, Plenum Press, 1968, pp. 195-218.
162
JOURNAL OF HAND THERAPY
14. Weinstein S: Tactile sensitivity of the phalanges. Perce pt Mot Skills 14:354-354, 1962. 15. Weinstein S, Seren E: Tactile sensitivity as a function of ha ndedness and laterality. J Comp Physiol Psychol 54:665 - 669, 1961. 16. Bell-Krotoski JA: Pocket filaments and specifications for the Semmes-Weinstein monofilaments. III Amadio PC, Hentz VR (eds): Year Book of H and Surgery. St. Louis, C. V. Mosby, 1992, pp. 293-294. 17. Bowen VL, Griener JS, Jones SV: Threshold of sensation: Interrater reliability and establishment of normal using the SemmesWeinstein monofilament (abstract). J Hand Ther 3:36-37, 1990. 18. Bell-Krotoski JA: Sensibility testing: Current concepts. III Hunter JM, Schneider LH, Mackin EJ, Callahan AD (eds): Rehabilitation of the Hand , 4th ed. St. Louis, C. V. Mosby, 1995, pp. 1- 19. 19. Eiijkman E, Vendrick AJH: Detection theory applied to the absolute sensitivity of sensory systems. Biophys J 3:65-78, 1963. 20. Weinstein SW, Drozdenko R, Weinstein C: Some Considerations of Clinical Tests, Presentation and Discussion. Danbury, CT, NeuroCommunications Research Laboratories, 1992. 21. Waylett-Rendall J: Sensibility evaluation and rehabilitation. Orthop Clin North Am 9:43-56, 1988. 22. LaMotte RH: Personal communication, 1992. 23. Von Prince K, Butler B: Measuring sensory function of th e hand in peripheral nerv e injuries. Am J Occup Ther 21:385396, 1967. 24. Werner JL, Orner GE: Evalu ating cutaneous pressure sensitivity of the hand. Am J Occup Ther 24:347 -356, 1970. 25. Levin S, Pearsall G, Ruderman RJ: Von Frey's method of measuring pressure sensitivity in the hand: An engineering analysis of the Semmes-Weinstein pressure aesthiometer. J Hand Surg [Am] 3:211 , 1978. 26. Dellon AL: The sensational contributions of Erik Moberg. J Hand Surg [Br] 15:14- 24, 1990. 27. Von Frey M: Zur Physiologie der Juckemptin dung. Arch Neurol Physiol 7:142, 1922. 28. Dyck PJ, Zimmerman lR, O'Brien Pc, et al: Introduction of automated systems to evaluate touch-pressure, vibration, and thermal cutaneous sensation in man. Ann Neurol 4:502 - 510, 1978. 29. Horch K, Hardy M, Jimenez S, Jabaley J: An automated tactile tes ter for evaluation of cutaneous sensibility. J Hand Surg [Am] 17:5, 829- 837, 1992.