Hand function was evaluated in 105 patients who had been operated on in early infancy for brachial plexus birth palsy. The mean follow-up after surgery was for 13.4 years (5.0 to 31.5). Fine sensation, stereognosis, grip and pinch strength and the Raimondi scale were recorded.

Fine sensation was normal in 34 of 49 patients (69%) with C5–6 injury, 15 of 31 (48%) with C5–7 and in 8 of 25 (32%) with total injury. Loss of protective sensation or absent sensation was noted in some palmar areas of the hand in 12 of 105 patients (11%). Normal stereognosis was recorded in 88 of the 105 patients (84%), whereas only 9 of the 105 (9%) had normal grip strength. The mean Raimondi scale scores were 4.57 (3 to 5) (C5–6), 4.26 (1 to 5) (C5–7) and 2.16 (0 to 5) in patients with total injury.

The location of impaired sensation was related to the distribution of the root injury. Avulsion type of injury correlated with poor recovery of hand function.

Obstetric brachial plexus palsy is mostly treated conservatively. Spontaneous recovery has been reported in between 66% and 92% of the cases.13 Reports of the incidence of obstetric brachial plexus palsy vary between 0.42 and 5 per 1000 newborns.49 The nerve roots may be damaged by avulsion, rupture or neurapraxia. The lesion is located in the upper plexus (C5–6) most frequently and a total injury (C5-T1) is seen in a minority of the patients.10,11 The indications for surgery to the plexus during the first months of life are not agreed and vary between clinicians,1214 but severe total injuries or upper-middle plexus injuries with no signs of spontaneous recovery are widely-accepted as indications for early operative treatment.

Permanent palsy results in various degrees of impairment of motor and sensory function in the affected arm. Most reports have focused on the recovery of the motor function and the function of the shoulder joint.9,1520 Sensory impairment of the hand has been evaluated by two-point discrimination, identification of small pieces (Moberg21 and Dellon22 tests), thermal thresholds, thresholds of perception of vibration, differentiation of heavy and light touch, proprioception and with Semmes-Weinstein monofilaments.23 Although the latter test is time-consuming, it is commonly-used. The filaments produce repeatable stimuli within very small variances as measured by means and standard deviations, and can also be reliably used in children.24,25 Sensibility and proprioception are essential for the normal development and function of the hand. Severe sensory deficits cause negligence of the affected limb, leading to progressive functional motor and sensory disability.26,27

We are aware of no long-term studies concerning the functional outcome of the affected hand in surgically-treated patients with obstetric brachial plexus palsy. The purpose of our study was to determine the motor and sensory hand function in these patients.

Patients and Methods

All patients with obstetric brachial plexus palsy who had undergone surgery to the plexus during the first 12 months of life, and were operated on before 1998 were included. Between 1971 and 1997, a total of 1706 patients were treated for obstetric brachial plexus palsy in Finland.9

Of these, those who had undergone early plexus surgery were chosen for the study (124 patients, 7.3% of the total). The Mallet scores, the Gilbert elbow scale28 and Raimondi hand scores29 for 112 of the 124 patients have already been described.9 A total of 105 (85%) of the 124 patients (42 males and 63 females) agreed to participate in the study. The mean follow-up was for 13.4 years (5.0 to 31.5). The right upper limb was affected in 53 patients (50%) and the left in 48 (46%). The injury was bilateral in four patients (4%), in which the more adversely affected side was operated on. Physical examinations were performed between September 2002 and October 2003.

The mean age at the time of surgery was 2.8 months (0.4 to 9.2). The operations performed are summarised in Table I. The intra-operative findings and surgical treatment in patients with total palsy are presented in Table II. In six of 25 patients with a total injury, the lower plexus was not explored. The extent of the injury was estimated according to physical examination, intra-operative findings, myelography (47 patients) and electromyography (95 patients). Differentiation between a C5–6 and a C5–7 injury was based on clinical findings, with weak or absent wrist extension in a C5–7 injury, and/or evidence of C7 injury either in myelography and/or intra-operatively. A C5–6 injury was found in 49 patients (47%), C5–7 injury in 31 (30%) and total brachial plexus injury (C5-T1) in 25 (24%). None of the patients had an isolated injury to the lower nerve roots. Root avulsion was found in 34 patients (32%). Pseudo-meningocele(s) were seen in 31 of the 47 myelograms performed and according to intra-operative judgement, avulsions were registered in 29 of 105 patients. Avulsion involved one root in 17 patients (51%), two roots in 12 (34%), three roots in four (11%) and one root in one (1%).

View this table:
Table I.

Surgical procedures according to type of injury

View this table:
Table II.

Injury pattern, surgical treatment and outcome in patients with total palsy.* Mean Raimondi score,29 percentage of normal fine sensation (Semmes-Weinstein)23 and stereognosis are also given for the three subgroups

The sensory function of the hand was evaluated by the Semmes-Weinstein test. Sensibility in 32 palmar areas of the hand (Fig. 1) was tested using five different sizes of Semmes-Weinstein filaments (2.83, 3.61, 4.31, 4.56 and 6.65 mm). These filaments had a constant length of 38 mm with the diameter of the filament corresponding to the strength of its bending force. The finest filament felt by the patient in each defined area of the palm was recorded. The median Semmes-Weinstein filament value for each palmar area of both hands was marked on a map. The sensory function was interpreted as abnormal if the 4.31 filament or higher were used at least once.30 The 3.61 filament was judged to be borderline abnormal and was interpreted as diminished light touch.31 However, among our patients the 3.61 filament was also the size most often needed to produce repeatable stimuli in the unaffected side.

Fig. 1

Diagram showing how the Semmes-Weinstein monofilament sensibility test23 was made in 32 different palmar areas of the hand.

The modified Moberg-Dellon pick up test21,22,32 was used to evaluate stereognosis. Six objects were first shown to the patient (coin, key, ring, screw, pin and key ring). The patient was then blinded and asked to pick up and recognise the items. Each correctly identified object yielded one point, giving a minimum of 0 points and a maximum of 6.

The grip strength was evaluated by using a Jamar Dynamometer (Asimow Engineering Co, Los Angeles, California) and a B&L Engineering pinch gauge (B&L Engineering, Santa Fe Springs, California) for palmar, key and tip pinch. These tests have been shown to be reliable and valid in measuring strength.33,34 All strength measurements were performed as described by Mathiowetz et al.33 A ratio between the unaffected and affected hands was calculated and a ratio below 0.89 was estimated to be abnormal.35 The Raimondi hand scale29 was used to classify motor function.

Motor function of the hand, the Semmes-Weinstein monofilament test, stereognosis and measurements of hand strength were performed by independent examiners.

Oral and written informed consent was obtained from the patients or from their parents in the case of patients who were minors at the time of requesting consent. The research ethics committee of Helsinki University Hospital approved this study.

Statistical methods.

The Mann-Whitney U test was used to examine the differences for continuous variables. Correlations were analysed using the Spearman’s rank correlation test. Odds ratios for the presence of abnormal clinical outcome at 95% confidence intervals (CI) for different patient groups were analysed with logistic regression models (NCSS 6.0 Statistical Software, Kaysville, Utah). The age at the time of the operation, gender, type of injury (avulsion or not), and experience of the surgeon (more than ten previous procedures) were considered as independent covariates in the analyses. The type of the surgery (neurolysis, suture, grafting or neurotisation) was closely associated with the type (avulsion/no avulsion) or extent of the neural injury and therefore both of these covariates could not be included in the same logistical regression model. Two-tailed p-values ≤ 0.05 were considered to be statistically significant.


Semmes-Weinstein monofilament test.

Normal sensory function (2.83 to 3.61) was obtained in 57 (54%) patients. In patients with a C5–6 and C5–7 injury, the median sensibility of the affected and non-affected hand was normal (2.83 to 3.61). In the patients with C5–6 injuries, abnormal Semmes-Weinstein filament test findings (4.31 to 6.65) were registered in 15 (31%) affected hands. In patients with C5–7 injury the Semmes-Weinstein filament test was abnormal in 16 (52%) and in 17 (68%) with a total injury (Table III).

View this table:
Table III.

Sensory status of the affected hand (Semmes-Weinstein monofilament test).23 Filament sensation was recorded in 32 palmar dermatomes. Classification into different sensory grades is based on the worst single filament recording

Loss of protective sensation, deep pressure sensation only or complete loss of sensation (12 of 105, 11%) was found mainly in patients with total injuries (Table III). Two of five patients (Table II) with a total injury which had a repair or neurotisation of the lower roots registered normal filament sensation of the affected hand compared with four of 20 patients without surgical intervention in the lower plexus (p = 0.35).

In the patients with upper and/or upper and middle root injuries, diminished sensation occurred most commonly on the radial dermatomes of the palm (Fig. 2). In contrast to this, abnormal Semmes-Weinstein recordings in patients with total injury were generally found in the proximal and ulnar dermatomes (Fig. 2).

Fig. 2

The Semmes-Weinstein monofilament sensibility test in patients with C5–6 (n = 49), C5–7 (n = 31) and total (n = 25) injury: non-affected hand (left), median affected hand (middle), worst affected hand (right). Green = normal touch (Semmes-Weinstein monofilament 2.83), blue = diminished light touch (Semmes-Weinstein monofilament 3.61), violet = diminished protective touch (4.31), red = loss of protective touch (4.56), orange = deep pressure only (6.65), white = no response.

Stereognosis (Moberg-Dellon).

The results of the pick-up test were normal in 86 patients (82%). This was observed in 46 of 49 patients (94%) with C5–6 palsy, 27 of 31 (87%) with C5–7 palsy and 15 of 25 (60%) with total palsy. All patients who did not recognise any of the six items belonged to the total injury group. Four of the five patients with reconstruction of the lower nerve roots had normal pick-up tests compared with 11 of 20 patients without this reconstruction (p = 0.31) (Table II). The results of the Moberg-Dellon pick-up test were not age related. The results according to the extent of the injury are given in Table IV.

View this table:
Table IV.

Stereognosia (modified Moberg-Dellon test)21,22 of the affected hand. The patients had to identify six objects without vision (coin, key, ring, screw, pin, key ring)

Hand strength.

Only nine of all 105 patients had a normal grip strength ratio between the affected and the non-affected hand. The group with a C5–6 injury did not differ significantly from those with a C5–7 injury (p = 0.48). Patients with a total injury had significantly lower measurements of hand strength, including seven patients (28%) who could not get hold of the dynamometer or the pinch gauge. Their results were recorded as zero. The mean ratio between the affected and non-affected hand in patients with C5–6 injury in grip strength was 0.61 (0.13 to 1.79), in palmar pinch 0.81 (0.39 to 1.23), in key pinch 0.88 (0.36 to 1.43) and in tip pinch 0.80 (0.31 to 1.3). The results were 0.37, 0.61, 0.57, 0.54 for the C5–7 group and 0.09, 0.19, 0.28, 0.17 for the total injury group, respectively. More detailed information is given in Table V.

View this table:
Table V.

Hand strength on the non-affected and affected side with minimum and maximum values (Jamar dynanometer for grip strength and Pinch gauge for palmar, key and tip pinch)

Motor function of the hand.

The mean scores in the Rai-mondi scale for each group were as follows: 4.57 (3 to 5) for patients with a C5–6 injury, 4.26 (1 to 5) with a C5–7 injury and 2.16 (0 to 5) with a total injury. All patients, except one with poor hand function (Raimondi scale between 0 and 2), had a total brachial plexus injury (Table II). The Mallet and Gilbert scale scores were used in further statistical analysis. The mean values for the Mallet shoulder scale scores were 3.4 (2 to 5) (C5–6), 2.9 (2 to 4) (C5–7), 2.4 (2 to 3) (total injury) and for the Gilbert elbow scale the scores were 4.2 (1 to 5), 3.6 (1 to 5) and 2.5 (1 to 5), respectively. These results have been described earlier in a publication with a slightly larger patient population (n = 112).9 The unaffected hand was dominant in all but three (3%) of the patients.


Univariate analysis revealed that patients with an avulsion type injury had significantly lower Mallet (p = 0.017), Gilbert (p = 0.0079), Raimondi (p = 0.032), grip strength (p = 0.0070), palmar pinch (p < 0.001), key pinch (p < 0.001), and tip pinch (p < 0.001) values than those without avulsion. In addition, there was a significant correlation between the Mallet score and the Gilbert score (Spearman’s rank correlation coefficient (rs) = 0.69, 95% CI 0.57 to 0.78), the Raimondi score (rs = 0.55, 95% CI 0.38 to 0.68), grip strength (rs = 0.68, 95% CI 0.57 to 0.77), palmar pinch (rs = 0.64, 95% CI 0.51 to 0.74), key pinch (rs = 0.67, 95% CI 0.55 to 0.76), as well as with tip pinch (rs = 0.66, 95% CI 0.54 to 0.75) (p < 0.001 for all correlations). The Gilbert score correlated significantly with the Raimondi score (rs = 0.44, 95% CI 0.26 to 0.60), grip strength (rs = 0.69, 95% CI 0.57 to 0.78), palmar pinch (rs = 0.74, 95% CI 0.59 to 0.79), key pinch (rs = 0.67, 95% CI 0.55 to 0.76), and tip pinch (rs = 0.69, 95% CI 0.57 to 0.78) (p < 0.001 for all correlations). The Raimondi score correlated significantly with grip strength (rs = 0.54, 95% CI 0.37 to 0.67), palmar pinch (rs = 0.50, 95% CI 0.33 to 0.64), key pinch (rs = 0.53, 95% CI 0.36 to 0.66), and tip pinch (rs = 0.55, 95% CI 0.39 to 0.68) (p < 0.001 for all correlations).

The type of injury, whether an avulsion or not, was closely associated with poor outcome after adjusting for age at surgery, gender, and the experience of the surgeon (Table VI).

View this table:
Table VI.

Odds ratios (OR) and the 95% confidence intervals (CI) for outcome (Semmes-Weinstein monofilament test,23 stereognosis, key pinch) when age, gender, injury (avulsion or not), experience of surgeon are considered as independent covariates in logistic regression models


Little has been written about hand function after surgery for obstetric brachial plexus injury.1,11,27,3639 We present a retrospective population-based study with a long-term follow-up of 85%. The results of this study have to be considered in the light of the development in Finland during the study of the indications, timing and surgical technique for operations in obstetric brachial plexus palsy. Direct suture after excision of the neuroma was most commonly used if possible to theoretically allow more rapid regeneration compared with nerve grafting. The operations were performed in nine hospitals by 11 different surgeons, of whom only two had gained experience of more than ten operations. Therefore, our results are not directly comparable with more recent series from specialised centres with uniform treatment regimes and more experienced surgeons.

A uniformly-accepted outcome measurement for hand function does not exist, especially when motor function is concerned. However, several centres have adopted the Raimondi scale for current use. Sensory function has also been evaluated by different methods including differentiation of heavy and light touch, testing of the thermal and vibration threshold, two-point discrimination and the Semmes-Weinstein filament test. Two-point discrimination was originally defined as the distance between compass points necessary to feel two contacts.40 The pressure applied is not standardised and the outcome in different adult studies varies.23,41 Furthermore, children may experience difficulty in understanding the principle of two-point discrimination.42 The Semmes-Weinstein monofilaments test has been shown to detect sensory changes after peripheral nerve repair better than the two-point discrimination test.43,44 It is a simple and pain free test which is especially suitable for children.25,45,46 Two independent observers carried out all clinical examinations. The effect of each independent variable on the satisfactory outcome was analysed using the logistical regression model.

Sensory function of the hand was highly related to the type of the injury in this study. In C5–6 or C5–7 type injuries, 39% of the patients had impaired sensation, especially on the radial side of the palm. The majority (76%) of the patients with total injury had abnormal Semmes-Weinstein filament findings in at least some ulnar dermatomes and in one third of the hands without protective sensation. The difference in the distribution of abnormal filament recordings between C5–6 and C5–7 injuries compared with total injuries is logical, since the upper root injuries were either repaired, grafted or neurotised in most of the patients regardless of the extent of the injury, whereas repair or neurotisation of the lower roots was seldom performed (20%). Comparison with previous reports is difficult partly because sensation has not been previously recorded with Semmes-Weinstein filaments and partly because of differences in the study populations. However, the reported frequency of absent protective sensation has varied between 0% and 30% in patients with total palsy.1,36,42,47

Stereognosis, as assessed by the Moberg-Dellon pick up test, was normal in 91% of patients with C5–6 or C5–7 injuries and in 60% of the patients with total palsy. Only four patients (16%) in the total group could not recognise any of the six items. Good stereognosis in our patients can in part be explained by few patients having poor sensation affecting many digital rays. Furthermore, stereognosis also depends on the development of central neural connections in addition to the presence of the peripheral sensory connections.46 These findings are in accordance with the results of Strömbeck et al,36 where nearly all patients with C5–6 and C5–7 lesions had normal stereognosia compared with 16 of 26 patients (62%) with total injuries.

Measurements of strength in the affected hand were significantly diminished in nearly all patients, with a more profound decrease in grip strength compared with pinch. Only 9 of the 49 patients (18%) with a C5–6 injury had a normal grip strength ratio (> 0.89) between the affected and the non-affected hand, according to the criteria of Petersen et al.35 There was a negative correlation between the type of the injury and all four strength measurements with very low values recorded in patients with total injury. Normal grip strength has been previously reported in 75% of surgically-treated patients with C5–6/C5–7 lesions and in 0% of patients with total injuries.36 The discrepancy between our results and this earlier study concerning patients with upper and upper/middle type injuries can perhaps be explained by the more lenient criteria for normal grip strength, with a ratio between the affected and healthy hand > 0.79, used by Strömbeck et al.36

Motor function of the affected hand was significantly better in patients with upper or upper/middle type injury compared with those with total palsy: the mean Raimondi score were 4.6, 4.0 and 2.2, respectively. Furthermore, the majority (52%) of the patients with total palsy had a Raimondi score of 2 or less. The Raimondi score assesses mainly the motor function of the hand, which explains the poor correlation with testing of sensibility (Table II). Our results are inferior to those published by Smith et al,47 which can again be explained by the low frequency (20%) of lower root repair in our series. The unaffected hand was dominant except in three patients who had a C5–6 injury with a good outcome after surgery. This finding underpins the previous results of Yang, Anand and Birch.49

We found a negative correlation with avulsion type injuries and functional scores (Mallet, Gilbert and Raimondi) and hand strength (grip and pinch measurements). There was a slightly stronger positive correlation with the Mallet and Gilbert score than the Raimondi score with hand strength. A grip strength ratio between the affected and the healthy hand of 0.61 in C5–6 injuries and 0.37 in C5–7 injuries was also relatively low, indicating that good shoulder and elbow function is an important factor for the development of normal hand strength.

It is still unclear whether the natural history in obstetric brachial plexus palsy concerning sensory and motor hand function can be improved surgically. According to Strömbeck et al36 surgical repair of the plexus does not favour hand function, including sensibility, whereas Anand and Birch30 found better restoration of sensibility in patients with total palsy who were treated surgically. Our results support the findings of Anand and Birch30 because five of 25 patients with total injury who had lower root reconstruction had better hand function than the remaining 20 patients.

Good hand function can be expected in obstetric brachial plexus palsy after plexus surgery in upper root lesions. Some patients with total lesions are left with absent protective sensation and stereognosis with no grip strength and insignificant motor functions of the affected hand.


  • This study was supported by the Foundation for Pediatric Research and Finnish state study grant.

  • No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.

  • Received June 18, 2007.
  • Accepted October 24, 2007.


View Abstract