A study of the relative responsiveness of five sensibility tests for assessment of recovery after median nerve injury and repair

A STUDY OF THE RELATIVE RESPONSIVENESS OF FIVE SENSIBILITY TESTS FOR ASSESSMENT OF RECOVERY AFTER MEDIAN NERVE INJURY AND REPAIR CHRISTINA JEROSCH-HEROLD From the School of Occupational Therapy and Physiotherapy, University of East Anglia, Norwich, UK

A longitudinal dynamic cohort study was conducted on patients with median nerve injuries to evaluate the relative responsiveness of five sensibility tests: touch threshold using the WEST (monofilaments), static two-point discrimination, locognosia, a pick-up test and an object recognition test. Repeated assessments were performed starting at 6 months after surgery. In order to compare the relative responsiveness of each test, effect size and the standard response mean were calculated for sensibility changes occurring between 6 and 18 months after repair. Large effect sizes (>0.8) and standard response means (>0.8) were obtained for the WEST, locognosia, pick-up and object recognition tests. Two-point discrimination was hardly measurable at any time point and exhibited strong flooring effects. Further analysis of all time points was undertaken to assess the strength of the monotonic relationship between test scores and time elapsed since surgery. Comparison of monotonicity between the five tests indicated that the WEST performed best, whereas two-point discrimination performed worst. These results suggest that the monofilament test (WEST), locognosia test, Moberg pick-up test and tactile gnosis test capture sensibility changes over time well and should be considered for inclusion in the outcome assessment of patients with median nerve injury. Journal of Hand Surgery (British and European Volume, 2003) 28B: 3: 255–260 with time. A single test is unlikely to be able to capture this range and therefore a battery of tests may need to be employed. The purpose of the present study was to investigate and compare the responsiveness of a battery of sensibility tests on a cohort of patients who had undergone surgical repair of a median nerve laceration.

INTRODUCTION Trauma to a major nerve trunk in the upper limb results in severe sensory impairment, motor paralysis and considerable pain and discomfort. The recovery of strength and sensibility following surgical repair can take up to 5 years and restoration of normal function cannot be expected in adult patients (Birch et al., 1996; Moneim and Omer, 1998). During the often lengthy follow-up of these patients, clinical tests are needed which can indicate that regeneration of nerve fibres and recovery of sensory and motor function are occurring as expected. There is an abundance of tests available to assess sensibility, with little consensus on which test(s) best quantifies outcome or, indeed, what defines a good outcome after peripheral nerve injury (Moneim and Omer, 1998). Much of the research into measures of sensibility after peripheral nerve injury has focused on the reliability and validity of tests. Responsiveness, that is the ability of a test to reflect change, has received less attention but is important. The follow-up of patients with peripheral nerve injury by the hand therapist often involves repeated assessments of sensory and motor function. Their purpose is to assess if regeneration is occurring, establish that interventions are effective, assist with clinical decision-making and provide important feedback to the patient and referring surgeon. This requires tests which are responsive to the changes of nerve regeneration. The process of sensory recovery covers a spectrum of impairment from complete anaesthesia during the first few months, to return of some discriminative sensibility

PATIENTS AND METHODS The method of investigation used was a prospective, longitudinal, observational study based on a dynamic cohort. Repeated assessments of sensibility using a battery of sensibility tests were carried out at set time intervals after surgical repair. Patients at different stages of recovery were recruited into the study in order to maximize sample size, thus making it a dynamic study sample. The variable ‘‘time elapsed since surgery’’ measured in months was used to assign the patients’ test results to standardized time ‘‘windows’’. Patients with a significant median nerve or combined median and ulnar nerve lesion were chosen as the study population in order to allow a comparison of the relative responsiveness of different tests, including functional tests such as the Moberg pick-up test (Moberg, 1958). Patients were identified from surgical lists at three regional centres for Plastic and Reconstructive Surgery and their surgical notes were viewed to obtain further details. Patients who had sustained a 50% or greater median nerve laceration at the wrist or 255

256

forearm level, which was surgically repaired were considered for inclusion in this study. Reasons for excluding patients from the study were: moved out of the region or detained in a prison or a psychiatric unit; development of other neurological impairments which may affect the sensation of the hand, e.g. multiple sclerosis, stroke, inability to understand or follow simple instructions in English, e.g. moderate to severe learning difficulty; suffered other concomitant injuries or developed complications resulting in severely compromized function, e.g. complex regional pain syndrome type I. Approval for the study was given by each hospital’s Research Ethics Committee in compliance with the Declaration of Helsinki. Patients were invited to participate by letter and informed consent was obtained in writing prior to start of the study. The rationale for including certain tests within the battery was based on a comprehensive review of evidence with regard to the test’s validity and reliability. This had to be balanced with practical considerations such as the cost of the test or equipment, the time it takes to administer, its general availability and its use in clinical practice. A test which is highly responsive but also very costly may find little use within the current healthcare climate. The touch threshold test, also referred to as the Semmes–Weinstein Monofilament Test, is considered a valid test of reinnervation (Rosen, 1996). It quantifies the smallest force at which detection occurs. Its interand intra-tester reliability are good (Bell-Krotoski and Tomancik, 1987; Novak et al., 1993; Thibault et al., 1994). A newer development, the Weinstein Enhanced Sensory Test, Bioinstruments Inc, Connecticut (WEST) contains five filaments on one handle thus increasing portability. Also, improvements to the tip geometry have reduced slippage rates (Al-Quattan, 1995) and it is supplied with guaranteed calibration. The WEST (five filament kit) was used as the preferred instrument for testing touch detection and was applied according to the protocol by Bell-Krotoski et al. (1993). Using the heaviest filament first, three applications randomly interspersed with two shams (no actual stimulus though verbal cue offered) are given and, if detection of one in three actual stimuli occurs, the next lightest filament is applied. The touch detection threshold is determined as the lightest filament at which detection of at least one out of three stimuli took place. The differences in force between the five filaments are not equidistant and are therefore converted into an ordinal scale as follows: 0.07 gm=5; 0.2 gm=4; 2.0 gm=3; 4.0 gm=2; 200 gm=1; >200 gm=0. The final score recorded is the summated score for the tips of thumb, index and middle fingers (maximum 15). The two-point discrimination (2PD) test has been criticized as a valid measure of spatial threshold (Johnson et al., 1994) and also on account of its reliability (Bell-Krotoski and Buford, 1997). Despite this 2PD remains a widely used test and was therefore

THE JOURNAL OF HAND SURGERY VOL. 28B No. 3 JUNE 2003

included in the battery of tests. The Dellon-Mackinnon Disk-Criminatort (Dellon-Mackinnon Disk-Criminatort, Post Office Box 16392, Baltimore, Maryland) was used as the instrument and applied according to the protocol described by Jerosch-Herold (1993) and Marsh (1990). Beginning at the widest calliper distance of 15 mm, ten random applications of one or two points are given. If the patient is able to correctly discriminate seven or more, the next smallest calliper width is used. The final threshold is converted into an ordinal scale (Rose! n, 1996): unable to discriminate at 15 mm=0, threshold between 11 and 15 mm=1, threshold between 6 and 10 mm=2, threshold equal or smaller than 6 mm=3. The final score recorded was the total score from the thumb, index and middle fingertips to a maximum of 9. Locognosia (area localization) was assessed using a modified version of Marsh’s test (Marsh, 1990). Each pulp of the radial three digits and radial side of the ring finger were divided into four quadrants giving a total of 14 zones. A suprathreshold stimulus was applied once in each zone in random order using the black filament (200 gm) from the WEST. Patients were asked to point to the zone where the stimulus was felt on a diagram of the hand which had the zones drawn on it. One point was given for each correctly identified orientation within a quadrant, i.e. upper left, lower left, upper right, lower right, and one point for localizing to the correct digit: thus a total of 28 could be scored for each hand. The Moberg pick-up test (Moberg, 1958) and the modified pick-up test (Dellon and Kallman, 1983) require identification of objects and have been widely reported in the literature. However standard test material is not available. For the present study the materials used for the ten objects were similar to those described by Moberg (Moberg, 1958) and were all made of metal. A board with two shallow containers was fabricated and a standardized protocol for administering the tests was devised. For the pick-up task, the subject was required to pick up all ten small objects, one at a time, using the radial three digits from one container and place these in the adjacent dish. A blindfold was used and the ulnar two digits were lightly taped into the palm using Cobant (3 M, St.Paul, MN). The time taken to complete this task was timed using a stopwatch. The final score recorded was the average of three trials. For the object recognition test, each object was placed in the patient’s hand and the time taken to identify the item recorded. The final score was the total time taken for all ten objects as well as the number of correctly identified objects. All assessments were carried out by a qualified occupational therapist with experience in using these tests and the tests were administered in the same order as they are described above. The complete battery of tests took approximately 40 min to administer. Repeated assessments were taken using the same procedure at 4 monthly intervals.

257

RESPONSIVENESS OF SENSIBILITY TESTS

Statistical methods The assessment of responsiveness is problematic and there is no consensus on the best method for evaluating change (Beaton et al., 2001; Liang et al., 1990; Wright and Young, 1997). Different statistics have been described, which give a ratio of the mean change to the variability in baseline or other criterion. Effect size (Kazis et al., 1989) is calculated as the mean of the change divided by the standard deviation of the initial score. The Standard Response Mean (Liang et al., 1990) uses the standard deviation of the change as the denominator, whereas the Guyatt Responsiveness Index (Guyatt et al., 1987) bases its denominator on the standard deviation of a control group or stable population. All three statistics give a standardized score and thus allow comparison between tests even when they are based on different scales or units of measurement. The interpretation of Effect size and Standard Response Mean can be done using Cohen’s (Cohen, 1977) guidelines, whereby a value of or >0.8 is considered a large effect, >0.5 as moderate and one of 0.2 as small. A number of authors (Beaton et al., 2001; Wright and Young, 1997) have observed that the indices can alter depending on which responsiveness statistic is used and also on the actual scale of measurement. They caution against the interpretation of an absolute value, such as Cohen’s, and instead propose that the tests are compared according to their relative rank ordering based on the actual indices calculated. One of the disadvantages of these responsiveness statistics is that they assume linearity, require the use of an interval scale and can only utilize two time points at a time. In this study patients were assessed on repeated occasions and at different times after surgery. It was expected that patients’ scores on the sensibility tests would improve over time, but the assumption that such change would be linear could not be made. A new method of quantifying change was used, which measures the monotonicity of improvement in scores over time and does not require an equal number of time points for each patient. This degree of monotonicity was quantified as follows: within each patient, and only for those patients with at least four consecutive assessments, Spearman’s rank correlation coefficients (rs) were calculated between the test score and the months elapsed since surgery. This provided a correlation coefficient for each test variable and for each patient. The r values for a total of 18 patients and six test dimensions could be calculated. A Fisher transformation (Zr) was carried out in order to obtain a mean for the group and thus compare the average correlation coefficient between the tests. These Zr values can be shown to follow a normal distribution (Fisher and van Belle, 1993). Since the number of observations (assessments) varied between individuals, the variance, and hence the precision of the Zr statistics varied from patient to patient. Therefore, a weighted analysis of the Zr values was carried out and

the degree of monotonicity calculated as the mean of the weighted Zr values. Further analysis of variance was used to establish if the difference between tests in the degree of monotonicity was statistically significant.

RESULTS The subjects were patients who had sustained a median nerve transection which had been repaired using either epineural or combined epineural and perineural sutures. All patients received occupational therapy and/or physiotherapy post-operatively including some sensory re-education. However this varied across the centres and was not standardized. A total of 115 patients who met inclusion criteria were invited to participate. Unfortunately only 32 were successfully recruited and nine of these were later excluded due to repeated non-attendance or moving away. The final sample consisted of 23 patients, who had at least two consecutive assessments during a postoperative period spanning 6 months to 8 years. The mean number of repeated assessments were five per patient and a total of 125 assessments were undertaken over a 4-year period. The key demographic details of the study sample are presented in Table 1. The number of patients assessed at different time intervals since surgery varied. The earliest assessment was taken at 6 months after repair and the latest at 99 months. The follow-up of ‘‘older’’ injuries was problematic and the number of patients available for assessment at 4 or more years after surgery was small. A standardized time interval from 6 to 18 months after nerve repair was used to compare the relative responsiveness across tests. This was the longest time interval in which scores could be calculated for the largest available sample size (n=13). All 13 patients were assessed at 6, 10, 14 and 18 months after repair. The sample sizes for later time intervals were too small (five to ten patients), to allow for meaningful comparison of responsiveness indices. However, the data from assessments taken at later time intervals were included in the analysis of monotonicity. Table 2 gives the descriptive statistics and responsiveness indices for effect size and the standard response mean for all five tests (six dimensions). The Guyatt Responsiveness Index has not been included in the results. It is based on the change in the affected hand over the standard deviation of change in a control group. In the present study the unaffected hand was used as the control, but the scores for the unaffected hand did not change at all in the tactile gnosis test (number of objects correctly identified) and the 2PD test, resulting in a standard deviation of 0. Where either nominator or denominator equals zero an index cannot be calculated resulting in missing values. This means that the Guyatt Responsiveness Index could

258

THE JOURNAL OF HAND SURGERY VOL. 28B No. 3 JUNE 2003

Table 1—Demographic details Total number of patients

23

Gender Men Women Age Mean (range) age at time of injury Delay between injury and repair Mean delay Number of patients with delay=0 days Number of patients with delay=1 day Number of patients with delay = 4 days Injury type Complete median nerve Partial median nerve1 Combined median and ulnar nerves Cause of injury Hand through glass (deliberate or accidental) Self-inflicted2 Work related (glass or sharp objects) Hand affected Ratio of dominant to non-dominant First assessment after surgery (months) Median (range) Final assessment after surgery (months) Median (range) Duration of follow-up (months) Mean (range)

19 4 33 (12–53) years 0.5 days 16 2 1 17 6 8 14 2 7 16:7 9 (6–67)

methods described above. A total of 18 patients who had at least four consecutive assessments could be included, thus allowing a greater number of later test scores to be included in the analysis. The relative rank ordering of the mean Zr values shows that the degree of monotonicity between score and time is highest for the WEST, followed in descending order by the pick-up test, locognosia test, the object recognition test (dimensions: no of objects and time) and finally the 2PD test (Table 3). Inferential statistics were used to compare the degree of monotonicity between tests. A two-way analysis of variance was carried out using the Zr values as the dependent variable, the tests as a fixed factor, the subjects as random factor, and using a weighted leastsquares analysis to compensate for different precision of Zr values. The difference between the tests was statistically significant (P=0.02, df=5, F=2.78). Post hoc pairwise comparison between the tests using Tukey’s honestly significant difference showed that only the results for the WEST and the 2PD tests were statistically significant (Po0.05).

32 (10–99) 19 (4–36)

1

DISCUSSION

be calculated in only four out of six sensibility dimensions. The WEST exhibited the highest degree of responsiveness, followed in descending rank order by the object recognition test, the pick-up test and the locognosia test. Large (>0.8) indices were obtained for these four tests. The two-point discrimination test had small Standard Response Mean (0.36) and Effect size (0.11) values, indicating low responsiveness. Further analysis of the monotonicity between test score and time elapsed was carried out, using the

The ability of a test to reflect change over time is an important aspect of its validity, especially when it is used for the long-term follow up of patients with peripheral nerve injuries. Use of outcome measures which are highly responsive allows fewer patients to be studied when investigating the effectiveness of surgical or therapeutic interventions. The present study investigated the relative responsiveness of five tests for the assessment of sensibility using both standard indices of responsiveness and a two-way analysis of variance of the monotonicity between scores and time. The touch threshold test exhibited the highest degree of sensitivity to change which agrees with previous findings. Rosen et al. (2000) conducted a longitudinal

Injuries were to 50% or more of the nerve trunk, as documented in surgical notes. 2 Suicide attempts.

Table 2—Responsiveness for the six sensibility dimensions over the 6 months to 18 months time interval (n=13), quantified using SRM and ES

WEST (score 0–15) Gnosis objects (score 0–10) Gnosis time (seconds) Pick-up test (seconds) Locognosia (score 0–28) 2PD (score 0–9)

6 months Mean (7SD)

10 months Mean (7SD)

14 months Mean (7SD)

18 months Mean (7SD)

Standard response mean

Effect size (Cohen’s interpretation)

9.2 (2.2) 2.6 (4.3) 137.8 (72.5) 135 (66.4) 19.7 (4.3) 1.4 (2.85)

10.5 4.4 115.4 110.2 20.7 1.53

10.8 (1.9) 5.2 (4.0) 101 (63.4) 115.6 (53.6) 23.2 (2.3) 0.77 (1.9)

11.2 6.4 82.4 98.1 23.2 0.77

2.4 1.4 1.3 1.3 0.9 0.4

1.2 1.1 1.0 0.8 0.9 0.1

(2.2) (4.3) (71.3) (64) (4.4) (3.1)

(1.7) (3.2) (53.9) (52.6) (2.9) (1.9)

(large) (large) (large) (large) (large) (small)

Table 3—Weighted mean Zr (SD) scores for the six sensibility dimensions. A high mean Zr value indicates strong monotonicity between time and test score. Data based on correlation coefficients calculated for a total of 18 patients who has at least four consecutive assessments (range 4–10) Test

WEST

2PD

Locognosia

Pick-up

Gnosis time

Gnosis objects

Mean Zr

1.2 (70.7)

0.5 (70.6)

1.1 (70.6)

1.1 (70.9)

0.8 (70.8)

0.8 (70.8)

RESPONSIVENESS OF SENSIBILITY TESTS

follow-up of 19 median and ulnar nerve injuries over a period from 3 to 48 months after repair and reported a similarly large effect size (ES=0.73) for the touch threshold test. Two-point discrimination exhibited the lowest sensitivity to change in the present study. Although the 2PD threshold values were converted into an ordinal scale as described earlier, this is unlikely to explain the poor responsiveness. Examination of the raw data showed that very few patients regained a 2PD threshold of 15 mm. These findings concord with previous studies, which have shown that 2PD is a test which yields strong flooring effects (Jerosch-Herold, 2000; Rosen et al., 2000; Rosen and Jerosch-Herold, 2000) and therefore should not be used as a critical measure to quantify sensory outcome. The locognosia test used in the present study demonstrated good responsiveness and should be considered as an adjunct or alternative to 2PD testing. Both tests rely on spatial discrimination and, as for the 2PD test, the error of localization decreases in a proximal to distal distribution. Sieg and Williams (1986) reported a mean error of localization on the forearm of 17.33 and 5.41 mm on the palm of the right hand. Weinstein (1968) obtained similar values for the palm (5–6 mm) and 1–1.5 mm in the fingertip. Nakada (1993) reported a mean error of localization on the fingertip of 1.8 mm. Weinstein (1968) also demonstrated that there is a strong correlation between 2PD values and localization in normal subjects (Spearman rank r=0.92), suggesting that these measures overlap and therefore may measure similar constructs. The lack of an agreed standardized method for assessing localization may account for the relatively low frequency use of this test when reporting outcomes. The pick-up test and object recognition test used in the present study showed good sensitivity to change. However these results have limited clinical relevance as test materials are not available commercially and these tests only have applicability in median nerve injuries, where the area of interest is the radial three digits. The shape–texture identification test (STITM test) developed by Rosen and Lundborg (1998) is a standardized test of tactile gnosis and should be considered as an alternative to the tactile gnosis test employed in this study. This study has a number of limitations, including its sample size. Only 35% of all patients who met the inclusion criteria were successfully recruited into the study. However, the long-term follow-up of patients with median nerve lesions is problematic and other studies reporting outcomes have been based on relatively small samples. In a similar study, Rosen et al. (2000) reported outcomes at 3–48 months after repair in a sample of 19 patients, of which only eight had median nerve injuries. Multi-centre studies may yield greater sample sizes, however this benefit needs to be balanced against potential measurement errors due to the use of multiple testers. The sample of patients used in the

259

present study did reflect the typical demographic characteristics of patients with nerve injuries, who are predominantly young men with clean nerve lacerations, most commonly due to glass injuries. An important consideration in the interpretation of these findings is the question of whether the change captured by these instruments is of clinical importance. For the WEST, the median change over the 6–18 months post-operative assessments was three points. This means that for at least two digits, the detection increased to the next lightest filament during the study period, for example from 4 to 2 gm. Patients improved on average by 4 points on the locognosia test, which means that at least two further zones out of a total of 14 were correctly identified. For the timed pick-up test and object recognition test the time taken decreased by means of 53 and 74 s, respectively, and patients were able to identify four more objects out of a total of 10 at 18 months post-surgery. I consider that these improvements translate into better functional use of the affected hand and are therefore of clinical significance. A further limitation of the study relates to the use of responsiveness indices. The differences in Effect size or the standard response mean between tests were used as the basis for arguing that one test is better at capturing change than another. However this is not based on inferential statistical methods and therefore holds limited validity. The possibility that the difference between the responsiveness of tests could be due to sampling or random error cannot be ruled out. Inferential methods to estimate the sampling variability of responsiveness indices have only been documented once. Liang et al. (1990) described a Jacknife procedure for calculating the 95% confidence interval for the standard response mean. Their estimates were based upon a sample of 38 patients and the authors comment on the small sample size. In the present study the number of observations within each time interval was 15 or less and were considered too small to adequately use the Jacknife approach. However, an alternative method for assessing change over time was used and included inferential statistical methods. The findings from the analysis of monotonicity between time and scores concurred with the relative rank ordering of the tests on the responsiveness indices, and provide further evidence that the WEST, locognosia, pick-up and tactile gnosis tests are responsive measures. The responsiveness indices presented here are based on relatively early time interval, 6–18 months postrepair. However nerve recovery may take 5 years or longer (Rosen and Lundborg, 2001) which is why the analysis of monotonicity was performed in the present study. This included patients who were assessed at time intervals of up to 8 years post-surgery and the relative rank ordering between tests on this suggests the high responsiveness for the WEST and the low responsiveness of the 2PD test is not necessarily dependent on the stage of recovery.

260

Acknowledgements The author thanks the Elisabeth Casson Trust, The Research Innovation Fund of the University of East Anglia and the School of Occupational Therapy and Physiotherapy, University of East Anglia for financial support; Dr Lee Shepstone, School of Medicine, Health Policy and Practice, University of East Anglia for expert statistical advice

References Al-Quattan M (1995). Semmes–Weinstein Monofilaments versus Weinstein enhanced monofilaments: their use in the hand clinic. Canadian Journal of Plastic Surgery, 3: 51–53. Beaton D, Bombardier C, Katz J, Wright J (2001). A taxonomy for responsiveness. Journal of Clinical Epidemiology, 54: 1204–1217. Bell-Krotoski J, Buford W (1997). The force/time relationship of clinically used sensory testing instruments. Journal of Hand Therapy, 10: 297–309. Bell-Krotoski J, Tomancik E (1987). The repeatability of the Semmes–Weinstein monofilaments. Journal of Hand Surgery, 12A: 155–166. Bell-Krotoski J, Weinstein S, Weinstein C (1993). Testing sensibility, including touch-pressure, two-point discrimination, point localisation and vibration. Journal of Hand Therapy, 6: 114–123. Birch R, Bonney G, Payan J, Wynn-Parry C, Iggo A (1996). Symposium: peripheral nerve injuries. Journal of Bone and Joint Surgery, 68B: 3–21. Cohen J. Statistical power for the behaviour sciences. New York, Academic Press, 1977. Dellon A, Kallman C (1983). Evaluation of functional sensation in the hand. Journal of Hand Surgery, 8: 865–870. Fisher L, van Belle G. Biostatistics — a methodology for the health sciences. New York, Wiley, 1993. Guyatt G, Walter S, Norman G (1987). Measuring change over time: assessing the usefulness of evaluative instruments. Journal of Chronic Disease, 40: 171–178. Jerosch-Herold C (1993). Measuring outcome in median nerve injuries. Journal of Hand Surgery, 18B: 624–628. Jerosch-Herold C (2000). Should sensory function after median nerve injury and repair be quantified using two-point discrimination as the critical measure? Scandinavian Journal of Plastic Reconstructive and Hand Surgery, 34: 339–343. Johnson K, van Bowen R, Hsiao S. The perception of two-points is not the spatial resolution threshold. In Boivie J, Hansson P, Lindblom U (Eds), Touch, temperature and pain in health and disease: mechanisms and assessment. Seattle, IASP Press, 1994: pp. 389–404. Kazis L, Anderson J, Meenan R (1989). Effect sizes for interpreting changes in health status. Medical Care, 27: S178–S189. Liang M, Fossel A, Larson M (1990). Comparison of five health-status instruments for orthopaedic evaluation. Medical Care, 28: 632–642.

THE JOURNAL OF HAND SURGERY VOL. 28B No. 3 JUNE 2003 Marsh D (1990). The validation of measures of outcome following suture of divided peripheral nerves supplying the hand. Journal of Hand Surgery, 15B: 25–34. Moberg E (1958). Objective methods for determining the functional value of sensibility in the hand. Journal of Bone and Joint Surgery, 40B: 454–476. Moneim M, Omer G. Clinical outcome following acute nerve repair. In Omer G, Spinner M, von Beek A (Eds), Management of peripheral nerve problems. Philadelphia, W.B. Saunders, 1998. Nakada M (1993). Localization of a constant touch and moving touch stimulus in the hand: a preliminary study. Journal of Hand Therapy, 6: 23–28. Novak C, Mackinnon S, Williams J, Kelly L (1993). Establishment of reliability in the evaluation of hand sensibility. Plastic Reconstructive Surgery, 92: 301–310. Rosen B (1996). Recovery of sensory and motor function after nerve repair: a rationale for evaluation. Journal of Hand Therapy, 9: 315–327. Rosen B, Dahlin L, Lundborg G (2000). Assessment of functional outcome after nerve repair in a longitudinal cohort. Scandinavian Journal of Plastic Reconstructive and Hand Surgery, 34: 71–78. Rosen B, Jerosch-Herold C (2000). Comparing the responsiveness over time of two tactile gnosis tests: two point discrimination and STI-test. British Journal of Hand Therapy, 5: 114–119. Rosen B, Lundborg G (1998). A new tactile gnosis instrument in sensibility testing. Journal of Hand Therapy, 11: 251–257. Rosen B, Lundborg G (2001). The long-term recovery curve in adults after median or ulnar nerve repair: a reference interval. Journal of Hand Surgery, 26B: 3: 196–200. Sieg KW, Williams WN (1986). Preliminary report of of a methodology for determining tactile location in adults. Occupational Therapy Journal of Research 6: 4: 196–206. Thibault A, Forgt R, Lambert J (1994). Evaluation of cutaneous and proprioceptive sensation in children: a reliability study. Developmental Medicine and Child Neurology, 36: 796–812. Weinstein S.Intensive, extensive aspects of tactile sensitivity as a function of body part, sex and laterality. In Kenshalo, D. (Ed), The skin senses. New York, Plenum Press, 1968: 195–222. Wright J, Young N (1997). A comparison of different indices of responsiveness. Journal of Clinical Epidemiology, 50: 239–246.

Received: 25 July 2002 Accepted after revision: 21 January 2003 Dr C. Jerosch-Herold, Senior Lecturer in Occupational Therapy, School of Occupational Therapy and Physiotherapy, University of East Anglia, Queens Building, GB-Norwich NR4 7TJ, UK. Tel.: +44-1603-593316; fax: +44-1603-593166; E-mail: c.jerosch-herold@uea. ac.uk. r 2003 The British Society for Surgery of the Hand. Published by Elsevier Science Ltd. All rights reserved. doi:10.1016/S0266-7681(03)00017-2/jhsb.2003.0917, available online at http://www.science direct.com