Search tips
Search criteria 


Logo of spsJournal HomeThiemeInstructions for AuthorsSubscribeAboutEditorial Board
Semin Plast Surg. 2004 November; 18(4): 327–338.
PMCID: PMC2884791
Obstetrical Brachial Plexus Paralysis, Part 1
Guest Editors Julia K. Terzis M.D., Ph.D. Saleh M. Shenaq M.D.

Management of Obstetrical Brachial Plexus Palsy: The Stanford Experience


Stanford's experience in the management of obstetrical brachial plexus palsy dates from 1983. A formal clinic service began in 1992. The tenets of management include early evaluations, a dependence on sequential evolution for decision making, and very early neural surgery for babies with abnormal hands. We watch babies with normal hands for a longer time before advising surgery. Intraoperative evoked potentials are used to make surgical decisions. Reconstructive goals for upper plexus injuries include shoulder and elbow control. The paramount goal for babies with global palsies is hand function. Therapy throughout the child's growth years is vital. Sequelae, particularly shoulder contractures, require early surgical intervention. Secondary reconstructive procedures are typically beneficial in improving function. Since 1992, more than 400 children have been examined; 62 have had neural reconstruction and 102 have undergone secondary procedures. Surgery has been remarkably complication free. All children having neural reconstruction except two have benefited.

Keywords: Obstetrical brachial plexus palsy, reconstructive surgery, secondary procedures, intraoperative evoked potentials


The Stanford obstetrical brachial plexus palsy (OBPP) experience began in 1983, the consequence of the author's academic sabbatical experience working in Paris from November 1982 through February 1983 with Dr. Alain Gilbert, assisting him at surgery in a number of primary plexus explorations and secondary reconstructions, primarily to improve external rotation at the shoulder. At that time, Dr. Gilbert had an appointment at Hôpital Trousseau, a major pediatric hospital in the east of Paris. Dr. Gilbert had already accumulated what was at the time the world's largest operative experience in OBPP. During this sabbatical from Stanford, about 40 postoperative children of various ages and times after surgery ranging from months to about 7 years were examined and their postoperative evolution reviewed.

At that time, Gilbert's philosophy of management was strongly influenced by a longitudinal study of several children with OBPP who were allowed to evolve spontaneously. Tassin1 published these findings as a thesis in 1978 and in a later publication. Tassin stratified these children into three groups, based on outcome, and then made correlations regarding outcome and the time of initial recovery of certain muscles, focusing primarily on the time and level of recovery, according to time, of elbow flexion. One group recovered essentially to normal. In this group, rapid recovery of elbow flexion had been observed, and these children could forcefully flex the elbow against the resistance of gravity by 3 months of age. A second group demonstrated some activity in the elbow flexor muscles by about the third month. This could be just a flicker of activity. These children evolved to a good functional outcome but not to a normal or near-normal state. A third group demonstrated no evidence of any activity in the elbow flexor muscles at the third month interval examination. After spontaneous evolution, all these children had poor shoulder function although most eventually recovered the ability to flex the elbow through an important range. All today would be considered good candidates for secondary procedures, and many of the children in this study did have secondary procedures.

Based on this study, Gilbert2,3 elected to recommend a neural surgical procedure if an infant with an upper root injury reached the third month still exhibiting no evidence of any recovery of elbow flexor muscles. This benchmark, elbow flexion, was chosen because this joint is far more easily examined than the shoulder in infants, who demonstrate “trick” shoulder flexion movement brought about by pectoralis and trapezius activation, even with total paralysis of deltoid and supraspinatus muscles. Gilbert contended that all infants with global palsies who do not recover rapidly should be explored at an early period, within the first few months of life. We adopted this philosophy of surgical decision making during our early experience.


Over the ensuing decade, we at Stanford essentially followed the philosophy of Gilbert, that is, recommending plexus exploration for isolated upper root/trunk injuries at about the fourth month of life if there was no evidence of biceps/brachialis reinnervation at the third month examination. It was our practice to see the infant at the third month of life and, if there was no detectable reinnervation of the elbow flexors, surgery was scheduled to take place about age 4 months. If, on reexamination in preparation for surgery, there was a “dramatic” change, surgery would be postponed and the infant would be monitored with frequent reexaminations to be certain that we saw normal continued recovery. For example, we would occasionally but rarely see a baby who exhibited no discernable activity in biceps or brachialis at the third month examination demonstrate the ability to flex the elbow to almost 90 degrees against gravity at the presurgical examination. At least two babies, during this decade, had no biceps function at 3 months but at 4 months of age could put hand to mouth. On the other hand, and this was the case in perhaps 6 out of 10 babies, if there was no clear change, we would proceed with the planned plexus exploration. Not infrequently, perhaps in 2 or 3 of 10 babies, at the time of the immediate preoperative examination, we would note a flicker of activity in the biceps muscle. We did not view this change as dramatic and would proceed with the planned surgery.


Our decision-making algorithm has evolved over time. This was in part due to two factors, a closer study of the thesis of Tassin1 mentioned earlier and the intraoperative findings at the time of the initial neural surgical procedure that made us challenge our philosophy as derived from Gilbert.2,3 The first factor, a closer analysis of Tassin's thesis, convinced us that this third group, those who demonstrated no evidence of biceps or brachialis activity by the third month, were essentially all babies with global palsies (all having poor hand function) and few if any were the far more typical “Erb's” babies, that is, those with essentially normal hands. The second factor was the result of frequently finding strong contraction of the biceps when the C5 and/or C6 root or the anterior division of the superior trunk was electrically stimulated at the time of plexus exploration in babies who themselves could not preoperatively contract their biceps. A third factor was a significant improvement in the outcomes of secondary procedures, especially for improving shoulder abduction and external rotation in babies who presented with C5-C6 palsies and who did not spontaneously recover good shoulder abduction and external rotation. These factors led us to stratify our patients further into three groups and develop separate treatment and decision-making algorithms for these three groups, as discussed subsequently.

Our current philosophic principles:

  • Evaluate babies as soon after birth as possible.
  • The best decision-making tool is frequent interval examinations.
  • Adjunctive studies influence decisions in a minimal manner.
  • Electrophysiologic studies of babies under 3 to 4 months of age are virtually useless.
  • Unless one's institution has extraordinary capabilities, magnetic resonance imaging (MRI) is an insufficiently sensitive tool upon which to base surgical decisions.


The Stanford OBPP Clinic, formally established in 1992, is a service of the Lucille Packard Children's Hospital at Stanford (LPCHS). New patients, including newborns, infants, and older children, are typically referred by pediatricians, pediatric neurologists, pediatric orthopedists, and physiatrists. More and more frequently, families locate the clinic by searching the Internet. New cases of plexus injury are evaluated by surgeons and occupational therapists. Key historical data such as birth weight, presentation, occurrence of shoulder dystocia, and use of assistive delivery procedures, such as fundal pressure or vacuum, are recorded. We generally refrain from discussing birth events in great detail so as not to be placed into the position of having to answer questions regarding standards of care for a specialty, obstetrics, not ours. We politely decline to review hospital birth records if these are brought by the parents for the purpose of obtaining our opinion regarding causation and deflect as much as possible parents' questions regarding blame. We prefer to speak to parents in the most sympathetic way possible but use only very general terms, for example, “energy that exceeds the tolerable limits of nerve tissues” to discuss causation. I am strongly of the opinion that the majority of our OBPP injuries occur within the standard of care or occur because of less than adequate perinatal care in financially or educationally disadvantaged populations.

Of greater consequence for the purposes of decision making is the history, as recounted by the parents, of what movements existed soon after birth and the history of the evolution of recovery, if any. We try to establish an early benchmark, such as asking the mother what the infant could move at the time of the first feeding and at anniversaries, such as at 1 month of age, 3 months of age, and so on. We key on the history of gross movements such as opening the hand and closing the fingers. For older children we are interested in the parent's impression of key events such as when the child was first able to get the hand to the mouth.

As the history is being obtained, the infant (if this is case) is first observed being held in the mother's arms and, so protected, generally will allow us to move passively key joints, especially the elbow. An infant ,frequently resists the examiner moving a joint out of the position that it is in at a particular time. For example, if the elbow is held extended, as it usually is, no matter the type of OBPP, and if the triceps is functioning, the infant frequently resists the examiner's attempt to flex the elbow. The examination includes observing for the following:

  • Presence or absence of a Horner's sign
  • Resting posture of the affected limb—flaccid, “waiter's tip”
  • Retained ability to grasp
  • Evidence of trophicity

The physical examination of the infant includes:

  • Palpation of both supraclavicular fossae (a large neuroma can frequently be appreciated).
  • Passive range of motion of shoulder, elbow, and forearm rotation, with particular attention to the position of the radial head.
  • Next the infant is placed in a supine position and various stimuli are used to try to generate active movements and, where possible, make some assessment of strength. For years, we have used Gilbert's four-level modification (no activity, flicker, movement with gravity eliminated, movement against gravity) of the British Research Council (BRC) scale for manual muscle testing. More recently, we are incorporating the seven-level Hospital for Sick Children's scale. We assess sensation only grossly by observing for withdrawal from seminoxious stimuli, such as pinching the skin. The pictorial format attributed to Merle D'Aubigne and modified by Gilbert2 is used to document the findings of muscle testing, active and passive joint motion, and sensation. The examination is modified according to the age of the patient.


Erb's (C5, C6)

Babies diagnosed by assessment as having C5-C6 injuries (Erb's palsy) are further categorized according to whether both hand and wrist appear to function well; that is, the infant's wrist does not automatically flex when the fingers flex. Our philosophy for these babies is different from that for babies with more classical postures, that is, wrist flexed, forearm pronated, fingers held primarily flexed (waiter's tip position). If we judge the hand to be normal, we observe these babies for evidence of biceps and deltoid reinnervation for at least 6 months. If, by 6 months of age, there is no evidence of biceps recovery, neural surgery is recommended. Our most recent statistics indicate that less than 20% of these babies undergo neural reconstruction but 50% of babies in this category who have not achieved biceps recovery by 3 to 4 months of age ultimately have or would benefit from secondary shoulder procedures.

Indeterminate (C5, C6, C7, ? C8, T1)

Babies who would by most observers be categorized as having an upper plexus palsy (Erb's) but who present with an abnormal hand are assumed to have a more significant injury and are managed similarly to our earlier philosophy learned from Gilbert. These babies exhibit the more classical posturing. They frequently have weak or absent finger extension and poor triceps strength, although some triceps activity is usually present. By history, the arm and hand were often flaccid immediately after birth but with quick recovery of finger flexion and intrinsic activity. If we see no evidence of biceps recovery by 3 months of age and the hand continues to be abnormal, we schedule neural surgery at about the fourth month of age.

Global (C5-T1)

Babies who present to us with clear evidence of lower root injury, in addition to paralysis of shoulder and elbow muscles, are grouped into the “global palsy” category. In our series, 65% also exhibit a Horner's sign. Hand function varies from total absence to a flicker of activity in the ulnar two fingers. Trophicity is present to some degree. In older infants, parents report that their baby bites or chews on the affected hand. In ideal circumstances (e.g., no prematurity and otherwise healthy), we recommend neural surgery when these babies achieve 10 weeks of age, 10 pounds of weight, and 10 g of hemoglobin.


Early in our experience we obtained imaging and electrophysiologic evaluations of all babies. Based on several observations, we have greatly modified our recommendations. We noted that within the initial 3 to 4 months of age, in babies with upper root lesions, the conclusions from the electromyographic report were essentially interchangeable from baby to baby. The studies invariably found evidence of significant denervation but also evidence of reinnervation in selected muscles, typically biceps and deltoid. In these circumstances, we found this study of little predictive value and we could not use the data for surgical decision making. The same was found to be true when we analyzed our imaging data. At our institution, the pediatric radiologic staff does not perform computed tomography myelography on infants, arguing that the incidence of complications and side effects makes this procedure unacceptably risky, particularly in this population (the majority of our patients sue their obstetricians or midwives). MRI of the cervical spine and brachial plexus is performed instead. Technically proficient imaging demonstrates important features, such as absence of a root sleeve or cord displacement. However, none of our studies have clearly shown good images at all root levels of interest. Although we have seen strikingly beautiful images from other centers, we have not been able to obtain such images systematically. We are influenced by others who have had the opportunity to study the incidence of false-negative and -positive data to conclude that MRI, at least as performed at our institution, is not sufficiently accurate to allow us to base surgical decisions on this data. Therefore, we have chosen to base decisions on history and our serial clinical examinations.


When neural surgical exploration and reconstruction have been recommended and accepted (~10% of parents decline surgery even when strongly recommended), the baby is evaluated by the pediatric anesthesiology staff, primarily to alert them that we will be performing intraoperative evoked potentials and to allow them to plan their anesthetic regimen accordingly. Until recently, a thermoplastic head-chest splint was constructed in advance of surgery. We have abandoned its use in the last 2 years and have seen no difference in early outcome.

Preoperative Electrophysiology

If the infant is cooperative, scalp electrodes for somatosensory evoked potential (SSEP) studies are attached in the preoperative preparation area; otherwise, they are attached when anesthesia is induced. A stimulating electrode is placed on the volar wrist of both the affected and unaffected side and a recording electrode is placed at Erb's point in the supraclavicular fossa. A spinal electrode is placed just below the occiput. After anesthetic induction, the initial electrophysiologic studies are performed, with stimulation of both the involved and uninvolved sides. Stimulating the normal side allows us to determine whether the equipment is functioning properly and gives some idea regarding the effects of the anesthetic agents on signal properties. Even the normal nerves of the infant are incompletely myelinated and therefore conduction velocities and signal amplitudes are reduced. When these initial electrophysiologic studies have been performed, the wrist and Erb's point electrodes are removed. Occasionally, stimulating at the wrist on the affected side results in some intrinsic muscle contractions that were not noted on the preoperative physical examination.

Positioning and Procedure

The baby's head is placed in a gel head rest, making certain that the occiput does not “bottom out.” We have had two incidences of significant alopecia when we failed to follow this regimen conscientiously. Anesthetic levels can be adjusted so that there is sufficient depth of anesthesia to perform the initial exploration and then adjusted to as light a plane as possible for the second set of SSEP studies. The head is turned away from the operative side. Most of these infants, especially the chubby ones, have little or no neck to speak of and it is difficult to discern what are true skin creases versus folds of fatty skin. The proposed transverse supraclavicular incision is marked and then injected with 0.5% bupivacaine with epinephrine at 1:200,000 units/mL. The neck, shoulder, and entire arm and both legs are prepared. For infants, we no longer use thigh tourniquets to harvest sural nerve grafts. With the infant supine and the leg fully elevated, as is necessary to harvest grafts, bleeding is minimal. Skin staples assist in maintaining drapes in place.

The supraclavicular fossa is explored through a transverse incision made just above the clavicle. We switched from the more deforming longitudinal incision recommended in earlier communications (as have most others) without noticeable loss of visualization. If the epinephrine has had time to be effective, this incision is bloodless. The platysma is essentially nonexistent at this age. The external jugular vein, if encountered, can be mobilized laterally and never needs dividing. The supraclavicular sensory nerves are encountered lying just deep to the platysma and, in general, can be stretched out of the way. The lateral border of the sternocleidomastoid muscle is visualized and retracted slightly medially and the submuscular layer of the superficial fascia is opened in a vertical direction. The omohyoid is divided, as is the lymphatic-rich fat pad. The underlying scalene muscle, the phrenic nerve and plexus, or neuroma-in-continuity remains protected by the injury-thickened deep layer of the deep cervical fascia. At this point a key landmark, the phrenic nerve, is identified. The phrenic nerve is the surgeon's guide to the C5 root. Once identified, it is stimulated with a portable nerve stimulator to confirm that it is functioning. If this nerve is nonfunctional at this early phase of exploration, or if stimulation results in minimal contraction of the diaphragm, the C5 root is almost surely avulsed and quite likely the C4 root is also avulsed. We have seen this only in babies with global palsies and in babies who suffer upper root injuries associated with breech or cesarean deliveries.

The phrenic nerve is followed superiorly and carefully protected if functioning. Permanent surgical injury to the phrenic nerve has led, although thankfully rarely, to significant postoperative respiratory complications. If present, the C5 root is encountered passing obliquely over the middle scalene muscle. More typically, just lateral to the phrenic nerve lies a large neuroma-in-continuity at the confluence of the C5 and C6 roots and the superior trunk. If one begins at the more normal superior part of the C5 root, the root can be more easily dissected into the neuroma-in-continuity.

The C6 root lies deeper in the neck in relation to the C5 root and its proximodistal course is less transverse than that of C5 root (which has a more exaggerated sigmoid course than the C6 root). The C6 root, if present, is invariably larger than the C5 root. At this point, we dissect carefully the phrenic nerve–C5 root relationship, searching for evidence of a large element coming from C4 (typically as part of the phrenic nerve) and then joining the C5 root. This is termed a prefixed plexus. The typical findings are a branch of some size exiting the phrenic nerve and joining the C5 root and then, almost immediately, a smaller branch exiting the C5 root to rejoin the continuing course of the phrenic nerve.

If a large neuroma-in-continuity is present, it can be rapidly dissected along its medial border in a proximal-to-distal direction. Dissection along its lateral margin should proceed with some care so that too deep a dissection does not risk injury to the C6 contribution, if existing, to the long thoracic nerve of Bell. As the clavicle is approached, the neuroma-in-continuity gives off the suprascapular nerve (SSN), another key landmark. This is dissected distally and surrounded with a rubber vascular loop. The traction of the birth injury will have displaced this nerve more distally and laterally than normal. Finding this branch allows rapid dissection of the anterior and posterior divisions of the superior trunk. These are also surrounded with a vascular loop. If a neuroma-in-continuity is present, this part of the dissection can be performed rapidly and safely. If the C5 and/or C6 roots are avulsed or completely ruptured and retracted proximally, dissection can be tedious and missteps, such as confusing a band of scalene muscle for a nerve root, frequent.

The C7 root can be identified using several landmarks. It lies even deeper in the neck and its course is even less transverse than that of the C6 root. If the transverse scapular artery, a branch of which accompanies the SSN, can be identified, it can be followed medially. It crosses immediately under or occasionally over the C7 root as it becomes the middle trunk. The neuroma-in-continuity, if present, can be elevated and retracted laterally to expose the interval between anterior and middle scalene muscles. The C7 root lies deep in this interval. The foramen of exit of the C7 root lies much more superiorly than anticipated. The C7 root, in my experience, is better hidden in these injuries than the C6 or C8 root. If present, this root is circled with a rubber loop and dissected proximally and distally. The C7 root and middle trunk are often confluent with the large neuroma-in-continuity and separating middle from superior trunk is difficult as there is no longer an anatomic plane. These two elements become one because axonal sprouting has occurred from one into the other. For example, we have systematically noted that babies who, at the time of surgical intervention, have begun to show flickering of their biceps demonstrate the same flicker of activity after surgery, even when the C5 and C6 root neuroma, or superior trunk neuroma, is resected and C5 and C6 reconnected to distal targets via nerve grafts. In these infants, preneuroma resection and stimulation of the C7 root, when present, or even the C8 root result in a contraction of the biceps that is much more forceful than can be appreciated on the preoperative clinical examination. These findings convince us that the reason why the majority of babies, even unoperated babies with severe upper trunk injuries, achieve biceps reinnervation is collateral (side-to-end) sprouting.

The subclavian artery is a useful guide to the C8 root and the inferior trunk. The C8 root lies posterior to the curve of the subclavian artery. Even if injured, the C8 root is rarely bound to the artery by scar and, if present, is typically more easily identified than the C7 root and middle trunk. The T1 root is not always easily dissected. Frequently it emerges from so deep within the interval between anterior and middle scalene muscles that what appears to be the C8 root is actually the inferior trunk. If the baby has a global palsy, and particularly if a Horner's sign is present, it is senseless, in our opinion, to dig deep into this interval because the T1 root in these cases is typically avulsed. We have never grafted onto a ruptured T1 root stump, having no faith that any motor regeneration will result. We have seen babies in whom a Horner's sign is present and who, at exploration, have a perfectly normal-appearing C8 and T1 root. However, stimulation of the inferior trunk results in neither motor movement nor SSEPs. We have judged these roots to be avulsed at the intraforaminal level. All of this dissection can take place through the supraclavicular incision.


Once the pathoanatomy has been identified, systematic electrical stimulation of each root or trunk is performed. We try to alert the anesthesiologist as this time is approaching so that anesthetic agents or levels can be adjusted. The contralateral side is again stimulated to determine that spinal and cortical signals are obtainable. We compare these with the signals obtained prior to any deeper planes of anesthesia that might have been necessary to allow dissection of the plexus. If there is a large difference, the anesthesiologist is asked to reduce the level of anesthesia to the minimum necessary to avoid extraneous muscle-generated noise. As a practical matter, electrically noisy operating room equipment unnecessary for the moment is unplugged. In our experience, the most significant source of electrical “noise” is the warm-air Baer Hugger warmer.

Depending on findings, both nerve-to-nerve and nerve-to-brain (or spinal cord) studies are performed. We begin by stimulating the most superior root identifiable, almost always the C5 root. If we see a large motor response in appropriate muscles and a good cortical signal, we then stimulate the C5 root and record in turn from the SSN, the anterior division of the superior trunk (ADST), and the posterior division of the superior trunk (PDST). Each root is stimulated using signal-averaging techniques, and this is repeated to confirm that the response is reproducible and real (not artifact). We judge the response on a scale of 0 to 3, with 0 being no discernable (above noise level) response, 1 being an easily seen and reproducible response, 2 being a response equal to at least 50% of that of the contralateral side, and 3 a response equal or nearly equal to that of the opposite side.

If we see motor movement and a clear signal that does not require any signal averaging, we strongly consider simple neurolysis of the involved area of root, trunk, division, or cord. This is a more common circumstance in older babies and rarely occurs in babies who have global palsies and who meet our operative indications. Once all recordings are accomplished, decisions regarding how to manage each level of injury can be made and reconstruction can proceed.


For babies with lesions restricted to the upper roots, reconstructive goals are similar to those for adults with traumatic brachial plexus injuries. We focus on restoring function to shoulder movers, particularly abduction and external rotation, elbow flexion, and wrist extension. These babies have several typical pathoanatomic patterns. The most common, as mentioned earlier, is a large neuroma at the confluence of the C5 and C6 roots as they form the superior trunk. Exiting the neuroma are the SSN and the anterior and posterior divisions of the superior trunk. Usually, the C7 root or the middle trunk is adherent to this neuroma. If both C5 and C6 roots are identifiable, we use the SSEP studies as a guide to determining which is likely to be the healthier root. If the studies indicate poor regeneration across the neuroma-in-continuity, the neuroma is excised proximally in a stepwise fashion using the operating microscope to study the cut face of the root for characteristics of a healthy root. If both roots have a favorable appearance and good response on SSEP studies, both are used to reinnervate the distal elements. The SSN and anterior and posterior divisions are dissected superiorly into the neuroma as far as they remain recognizable. These elements are sectioned at this level and their cut facies are examined under the operating microscope. Additional distal sectioning is performed as necessary.

In babies with global palsies, the goals of neural reconstruction differ from those recommended for adults with traumatic brachial plexus injuries. For these babies, restoring hand function is of paramount importance. The next priority is restoring elbow flexion, followed by shoulder stabilization and shoulder adduction. Some ingenuity is required to address the variety of injury presentations.

It is rarely possible to perform direct nerve coaptation, and, as in the adult, nerve grafts are almost universally necessary and nerve transfers frequently necessary. We preferentially harvest one or both sural nerves and no longer use thigh tourniquets in our babies. The assistant fully flexes the leg at the hip and, so elevated, there is very little bleeding. A stocking-seam incision is used, extending from lateral malleolus to the flexion crease at the popliteal fossa. A careful dissection is needed to identify the anatomy of the sural nerve and its contributions from the popliteal (or sciatic) nerve and a frequently large contribution from the peroneal nerve. If additional graft is needed, we harvest the medial brachial and antebrachial cutaneous nerves through an incision extending from axilla to midhumeral level. Leg incisions are closed in layers with absorbable sutures and dressings supported by elastic wraps.

The placement of nerve grafts is determined by the anatomic findings. For reinnervation of the elements of the superior trunk (SSN, ADST, PDST) we respect the anterior-posterior aspects of the C5 and C6 root (if both are available), placing grafts bound for the SSN and PDST on the posterior aspect of the root and grafts destined for the ADST on the anterior aspects. The typical sequence is first to determine how many grafts can be placed on each available root and how many grafts can be accepted on each target. A map is sketched, to assist with operative dictation, and distances from root to target are measured. Nerve grafts are cleared of unnecessary connective tissue and are sectioned into appropriate lengths. The posterior grafts are set into place on the appropriate root and their distal ends segregated in an accessible site for later distal placement. The anteriorly designated grafts are layered on top of the previously placed grafts and their distal ends segregated from the posteriorly directed grafts. Fibrin glue is used to fix the grafts to the proximal root stumps. We now rarely suture grafts in place unless exposure of the target root is so poor that gluing is not feasible. Once the fibrin glue has solidified, any excess is sharply trimmed and the root-graft junctures inspected. The distal ends of the grafts are then distributed according to the plan of reconstruction. Typically, the posterior division is reconstructed first, then the anterior division, and finally the SSN.

For babies with global palsies, nerve grafts are placed according to findings and reconstructive goals. Several common scenarios have been encountered, including avulsions of C8 and presumably T1, based on clinical presentation and intraoperative SSEP determinations and ruptures of C5, C6, and C7. In such a case, if the C7 root has a healthy appearance and reasonable SSEP signals and stimulation of the C8 root results in no SSEPs, we divide the C8 root as high as possible into its foramen and then can suture this directly onto the C7 root stump. If the C7 root is also avulsed, we try to connect the anterior aspect of the healthiest root to reinnervate the C8 root or lower trunk. The more severe the injury, the more likely we are to use nerve transfers, particularly the distal branches of the spinal accessory nerve led to the SSN and, rarely, the phrenic nerve to the ADST. We have not used the contralateral C7 root as a source of axons in our babies. On only three occasions have we performed the Oberlin4 procedure (motor fascicles of the ulnar nerve to innervate directly nerve branches to the biceps) to reinnervate the biceps. All were breech-born babies and, at exploration, were found to have relatively uninjured-appearing C5 and C6 roots but totally unstimulatable and lacking any SSEP signals. We judged these root to have suffered intraforaminal avulsions.


We have rarely had to divide the clavicle in babies. We have never had to divide the clavicle in babies with isolated upper root injuries and have had to do so in less than 20% of babies with global palsies. If necessary, an additional infraclavicular incision is made, beginning just below the outer third of the clavicle and curving into the deltopectoral groove. The lateral origin of the pectoralis major muscle is elevated off the clavicle and a small area of clavicle is cleared and two reciprocal periosteal flaps elevated. A bone cutter usually suffices to section the clavicle. The sharp point of a self-retaining (Gelpi-type) retractor can be placed into the end of each side of the divided clavicle and opened progressively. The subclavius muscle can be divided to expose the sub- and infraclavicular portions of the plexus. Once reconstruction is complete, a heavy absorbable suture is used to approximate the clavicular elements.

Following wound closure with absorbable sutures and skin tape, a soft cotton “T-shirt” is applied and the affected arm is adducted and wrapped against the chest with an elastic bandage. We no longer use a prefabricated head immobilizer. Babies usually stay overnight and are discharged the next day. No babies have required transfusion. The average operative time, including SSEP studies, is about 4 hours.


The shoulder immobilizer is removed and babies are allowed to be bathed after the initial 48 hours with the parents controlling head and shoulder and then replacing the shoulder immobilizer. Otherwise, the shoulder remains immobilized for 2 weeks. At this time, exercises are resumed and babies followed up at sequential clinic visits beginning at the third month. We have not experienced a postoperative infection, nor have we experienced injury to healthy and preoperatively functioning nerves. Two babies suffered an inadvertent injury to a preoperatively functioning phrenic nerve, as manifest by decreased O2 saturation in the recovery room and by a complicated recovery following later-derived respiratory infections. Both recovered within several months.


The first evidence of recovery is usually improved shoulder movement as the supraspinatus begins to recover at about the third to fourth month. Soon after, the biceps begins to flicker. During this time, directed therapy to maintain supple joints is a key element of treatment and prevention. Parents are taught the necessary exercises and are encouraged to perform these at every opportunity, meals, bath time, play time, and so forth. As recovery progresses and as soon as feasible, toddlers and young children are enrolled in swimming programs as this is the single best exercise for the joint most commonly affected in the majority of children with OBPP, the shoulder. Daily exercise should be continued, ideally for years. It is disappointing to see a child, at age 6, who has good passive and active motion, particularly at the shoulder, return at age 13 having lost function, presumably because of self and parental inattention to this needed exercise regimen.


The most common and earliest functionally important sequela of this injury is progressive loss of passive external rotation. Occasionally, the loss can be sudden, and this sudden loss seems to be associated with sudden posterior subluxation of the glenohumeral joint. Once these babies begin to develop an internal rotation contracture, recovery of passive external rotation by therapy is very difficult, especially if the infant has strong internal rotators. Such infants or young children may develop loss of passive external rotation while still seeming to maintain essentially normal passive shoulder abduction. In fact, glenohumeral abduction is being lost simultaneously; however, the arm can still be elevated secondary to increased (compensatory) scapulothoracic movement. The most meaningful examination to document that external rotation is being maintained by exercise and therapy is passively externally rotating the arm with the arm adducted. The scapula cannot rotate as easily, compared with its freedom to rotate when the arm is abducted.

The other sequelae that are commonly seen include progressive loss of elbow extension, occasionally associated with subluxation or dislocation of the radial head; loss of passive supination, similarly associated with radial head pathology; and loss of radial deviation at the wrist (ulnarization of the hand). The latter occurs in children whose radial wrist extensors fail to become reinnervated or do so very late and who have weakness or inability to activate the abductor pollicis longus (APL). The growing child substitutes for this lack by extending the wrist using digital extensors, the extensor digitorum communis (EDC), balanced by the flexor carpi ulnaris (FCU). The carpus becomes radialized and the hand and fingers point excessively in the ulnar direction. Once initiated, few of these sequelae respond adequately to conservative care, except perhaps flexion contractures at the elbow, which in our recent experience have responded to progressive dynamic (Dynasplint) therapy.


The need for secondary (postneural surgical) operative procedures depends upon many factors, some still incompletely recognized or understood. Foremost is the level of recovery from injury, in the case of spontaneous evolution, or the level of recovery following neural surgical intervention. It is useful to discuss the potentially beneficial secondary procedures in terms of anatomic locations.

The Shoulder

A primary goal of frequent (every 3 months) follow-up during the initial 2 to 3 years is to recognize the development of a tightening shoulder. Strong active external rotation is an infrequent outcome of either spontaneous reinnervation or surgical neural reconstruction. If the external rotators (the infraspinatus and teres minor) do recover or become reinnervated, they are severely disadvantaged in their contest for balance against the many very strong internal rotators (pectoralis major, teres major, latissimus dorsi, and subscapularis). Unless passive range-of-motion exercises are compulsively performed, and often even in spite of this, these internal rotators become shortened, the anterior capsule tightened, and the coracohumeral ligaments foreshortened. In addition, the coracobrachialis and pectoralis minor muscles may shorten. The consequence is a significant imbalance of muscle forces acting as force couples across the growing glenohumeral joint. The resultant vectors push the humeral head posterior until it subluxes. In this position, the humeral head begins to deform and the glenoid recontours. Over time, unless corrected, the humeral head becomes less rounded, more oval, and may develop areas with differing radii of curvature. The joint moves back and forth along these two radii, and this is characterized by a click or clunking of the shoulder. If untreated, the shoulder will remain posteriorly dislocated. The angle of the glenoid with respect to the scapula will tilt posteriorly.5

As mentioned earlier, once deformity becomes established, it is almost impossible to recover normal passive external rotation by conservative measures such as splinting, functional taping or bracing, or exercise. The child fights attempts at passive external rotation secondary to pain. For us, rapid loss of passive external rotation is almost always a sign of posterior subluxation or dislocation, and this needs to be addressed, typically with surgery, quickly.

The child is examined by passively ranging the shoulder. Active range is difficult to determine unless the child is older than 3 to 4 years of age. Younger children can be coaxed by toys or treats to reach overhead as much as possible, but encouraging active external rotational movements is less successful. Passive external rotation of the affected side is compared with that of the normal limb. Our criterion regarding when to consider surgical release is passive external rotation less than 15 degrees with the humerus adducted by the side. A humerus that clunks as it moves from internal to external rotation is another strong indicator to consider release of contracture and muscle rebalancing. One characteristic sign is the presence of an extra skin fold at the proximal humerus. The older child who cannot adequately externally rotate his or her arm runs with the arm elevated at the shoulder and held either across the chest or abducted at the shoulder and internally rotated in a very unmistakable posture.

A good ultrasonographer can document the development of posterior glenohumeral subluxation or dislocation. More commonly, we have relied on MRI of both shoulders to assess the presence of cartilaginous contour changes, subluxation, and dislocation. Plain x-rays are not particularly helpful because of the lack of ossification at this age.

Early in our experience, we preferred to perform only release of contracted tissues in children younger than 3 and combined release with muscle transfers only for older children. We anticipated that simple release in these very young children (typically age 2–3) followed by therapy would maintain the correction achieved at surgery and that, over time, these children's external rotators would become stronger and provide balanced forces about the shoulder. A significant number of these children, in spite of aggressive therapy, experienced recurrence of their internal rotation contracture. Therefore, we now combine a more aggressive and systematic release with transfer of internal rotators to a position of external rotation, no matter the age of presentation (except in the older children where imaging studies indicate severe deformity of the glenohumeral joint). In these older children, we recommend derotation osteotomy.


Smaller children have an orthoplast “Statue of Liberty” style of brace constructed preoperatively by the therapist. For older children, a fiberglass shoulder spica is constructed at the end of the surgical procedure.

Depending on the age and size of the child, two or three small incisions are planned. The anterior structures are accessed via an incision in the deltopectoral groove. The coracoid is exposed and the origins of the coracobrachialis and pectoralis minor muscles delineated. These tendons of origin will be lengthened. The musculocutaneous nerve is just deep to these and can be neurolyzed if desired. The coracohumeral ligaments are divided, but the coracoacromial ligaments are preserved. If the child is older than age 5, we perform a lengthening of the tendon of insertion of the subscapularis through this approach. The insertion of the pectoralis major can be lengthened if necessary.

Through an axillary incision, the latissimus dorsi (LD) muscle and its tendon of insertion are dissected from the teres major and its tendon of insertion. The tendon of the LD is divided as close as possible to its point of insertion on the humerus. We do not transfer the teres major. In young children, we reflect the LD to protect its nerve and then define the seam between teres major and subscapularis. The subscapularis is then detached off the inner surface of the scapula. The arm is maximally externally rotated to determine that a complete release has been performed. If the humeral head has become deformed, the joint clunks as the arm is externally rotated and clunks again with internal rotation. As recommended by MB Ezaki (personal communication), we plicate the posterior capsule in these children to try to maintain reduction of the joint and to prevent postoperative posterior resubluxation.

In a smaller child, the humeral head can be reached through the axillary incision by entering the seam between the long head of the triceps and the posterior deltoid. In older children, or in children with a large deltoid, a third incision paralleling this interval is made on the posterior aspect of the shoulder. The dissection exposes the humeral head superior to the axillary nerve and the target is the tendon of the infraspinatus or the rotator cuff tissues superior to this. Two suture anchors are placed and the tendon of the LD is brought over the long head of triceps and anchored with some tension as high as possible onto the humerus, with the arm maximally externally rotated and abducted to 90 degrees. Wounds are closed and the Statue of Liberty splint applied or constructed. This is worn continuously, except for removal to bathe, for 2 to 3 weeks. At this time, the child begins directed passive exercises out of the brace but wears the brace for decreasing periods of time during the day. The brace is worn at night for 8 weeks. This has been a predictable and beneficial procedure in our experience.


For older children with humeral head deformity, a derotational osteotomy of the humerus is performed. The osteotomy site is between the insertions of the pectoralis major and the deltoid. A 2.7-mm plate and screws are used for fixation. A light posterior fiberglass splint is applied to keep the elbow in flexion and a sling and swath dressing is used to immobilize the shoulder. We have not placed these children in a shoulder spica cast. It is possible to overrotate the arm, and this may lead to a more significant functional problem than the internally rotated arm.



Almost all babies with OBPP regain adequate elbow flexion power and range. The few that do not are candidates for one of the many muscle transfers described for restoring elbow flexion. In over 100 operated babies, we have had to resort to these transfers in only three instances. In two, a Clark-type pectoralis transfer was performed. In the third, triceps to biceps was performed.


In four instances, we have used Botox injections to a cocontracting triceps muscle to improve elbow flexion. This resulted in variable improvement in flexion in three and very remarkable improvement in one child that has persisted long after the anticipated pharmacologic effectiveness of the agent. Other investigators6 have surmised that cocontractures may not necessarily be a consequence of coinnervation of muscle groups by axons originating from the same groups of motor neurons but rather may be a learned response that can be unlearned. We believe that the same may be true for the typical cocontracture of deltoid and biceps when the child tries to place his or her hand to the mouth (the so-called trumpet sign). Lacking good external rotation, the infant or child “learns” that if he or she is to reach the mouth with the hand, the arm must be brought away from the body before the elbow is flexed. Otherwise, lacking external rotation, the chest blocks the hand's ability to reach the mouth.


In the past, we discouraged treating the typical mild contracture (20–30 degrees) that characterizes the great majority of children who developed a flexion contracture. The pathophysiology of this contracture is not well understood as it can occur even in children with triceps function. Its development is insidious and may be related to fibrosis of the muscle as a consequence of the trauma of delivery (the same has been proposed as a reason why the deltoid muscle frequently remains weak in spite of spontaneous recovery of other previously paralyzed muscles). More recently, we have had success with progressive dynamic splinting using Dynasplints fitted and monitored by therapist. We have not performed release of soft tissues, nor have we performed osteotomies to correct what are typically relatively mild contractures.


Many children with OBPP lack full supination. One cause of this loss is subluxation or dislocation of the radial head. This pathology may not be easily recognized unless one carefully examines the elbow in these children. If recognized early, the radial head can be passively relocated but it will not remain reduced. A triceps fascial sling can be constructed to serve as an annular ligament. If there has been a long delay in recognizing this deformity, the radial head cannot be reduced. An MRI is needed to determine whether the head and fossa are still relatively symmetrical. If so, at surgery, the fossa is cleared of fibrous tissue, the radial head relocated and pinned to the capitellum, and a sling constructed. If the radial head is no longer concave, it may be best to leave the radial head dislocated.

We have discouraged any attempts at muscle or tendon transfers to restore better forearm supination in these children for fear of losing the more functionally important movement of pronation.


A few children (almost all having global palsies) develop a persistently supinated forearm. For these children, we have performed a Zancolli7 biceps rerouting when the forearm can be passively pronated. In two older children with fixed supination deformities, we have performed rotational osteotomy of the radius. One child required osteotomy of both radius and ulna.



For children (again typically those presenting with global palsies) who fail to regain adequate wrist extensor function, we have recommended one of two procedures. If there are sufficient numbers of transferable muscles (e.g., the FCU) and the child has decent active finger extension, we recommend transfer of the FCU into the tendon of the extensor carpi radialis brevis (ECRB). If the hand has become “ulnarized,” a slip of the APL, left attached to the base of the thumb, is divided and led to the FCU transfer and sutured to it to try to provide some radial balance to the hand at the wrist. This has proved more successful than adding the tendon of the extensor carpi radialis longus to the FCU-to-ECRB transfer. If the child lacks finger and wrist extension but has good finger flexion, we prefer to attach the FCU to the combined tendons of the EDC, with or without including the tendon of the extensor pollicis longus. These procedures are always a compromise because these children frequently lack functionally significant palmaris longus or flexor carpi radialis muscles and thus cannot readily stabilize their wrists. Nonetheless, the FCU in these children is more of a deforming force and its transfer, although surely predetermined to perform suboptimally once transferred, may be all that is available.

For the few children lacking any transferable muscles, we have elected to perform a chondrodesis procedure. The cartilage is sequentially shaved from the surfaces of the radius, lunate, and capitate to just expose secondary ossification centers. A large Steinmann pin is placed down the third metacarpal, across the carpus and down the radial shaft. The arm is placed in a long-arm cast for 8 weeks. We have performed this procedure in only two children.


At the level of the hand the primary sequela seen in our population of patients has been inadequate digital extension. Its treatment was discussed earlier. All children with hand sequelae have experienced global palsies. Most have very few reliable options available to improve hand function. We have not performed free functional muscle transfers and are reluctant to perform tenodesis procedures in growing children.


The OBPP Clinic was formally established in 1992. Over the past 10 years we have evaluated, as new patients, 475 children with OBPP. Age at presentation to the clinic ranged from 4 weeks to 16 years. We see on average three newly injured infants each month. About one in three requires neural reconstruction. Of these operated infants, 70% have global palsies and the remainder have C5, C6, or C5-7 palsy.

Of 30 children presenting with C5-6 palsy, 60% had achieved a modified Mallet8 score of IV by 3 years after surgery and the remainder a modified Mallet score of III. Secondary procedures to improve shoulder external rotation and abduction were recommended for the children who achieved less than a score of IV, and this recommendation was followed in 8 of 12 children. These shoulder scores were all initially improved to the IV range following secondary surgery. However, we have been disappointed to learn that as some of these children reach the age of 10 to 12, they frequently lose shoulder range of motion. We have surmised that these children have ceased performing dedicated exercises. In these cases, neither child nor parent has opted to have further surgery. These results suggest that daily exercise during the growth years is necessary to maintain functional gains achieved by surgery.

It is difficult to quantify the outcomes for children operated to mollify the consequences of global palsies. It is somewhat easier to categorize these children by what they did not recover as opposed to what they can do because, from a functional perspective, these children adapt so readily and substitute so cleverly. We studied 30 operated children and can state the following:

10% (3 of 30) failed to recover any useful hand function.

3% (1 of 30) failed to recover useful elbow flexion.

45% (13 of 30) eventually recovered shoulder function to a modified Mallet score of III.

Ten children were candidates for transfers to improve wrist stabilization. The FCU was transferred in all cases and 60% were thought to have been significantly improved to the point that they no longer required a wrist orthosis.

Secondary Procedures

We studied a small series of children, half presenting with C5, C6 palsy, (+,–) C7 and half presenting with a global palsy, who accepted release of internal rotation contracture and tendon transfer as described earlier. Their outcomes at 1 year following surgery differed, with C5, C6 children having far better outcomes in terms of maintaining the correction achieved at surgery. This has become a very predictable procedure, one we feel confident in recommending.

What we now believe:

  • There are nonobstetrical etiologies of this condition.
  • It is becoming less clear that neural surgery for babies with isolated C5-C6 palsies (babies who have absolutely normal functioning wrists and hands) leads to a better functional outcome than allowing spontaneous recovery with plans for early secondary shoulder surgery and, for the rare baby with poorly recovering biceps, an Oberlin4 type of biceps (boost) procedure.
  • All babies with global palsies should have neural surgery.
  • Surgery can be well performed without complications and without loss of preexisting function by most competent plastic and orthopedic surgeons, and good outcomes can be obtained in circumstances other than a high-volume specialized centers.
  • A good functional result at 4 years of age may become less so as the child ages. This argues for more longer term studies of operated and nonoperated children with OBPP.
  • The loss of passive external shoulder rotation should lead to operation as early as possible.
  • Follow-up and dedicated therapy are needed throughout the child's growing years and probably beyond.
  • Contrary to the claims of litigators and their all too well-paid physician and nonphysician “experts,” these children are not, as a rule, made susceptible as adults to a lifetime of secondary problems, such as nerve entrapments or shoulder arthropathies.


  • Tassin J. Paralysies obstétricales du plexus brachial: Evolution spontanée, résultats des interventions réparatrices process. Paris: Université Paris VII; 1983.
  • Gilbert A, Khouri N, Carlioz H. Birth palsy of the brachial plexus—surgical exploration and attempted repair in twenty one cases (author's transl) [in French] Rev Chir Orthop Reparatrice Appar Mot. 1980;66:33–42. [PubMed]
  • Gilbert A, Tassin J. Réparation chirurgicale du plexus brachial dan a paralysis obstétricale. Chirurgie. 1984;110:70–75. [PubMed]
  • Loy S, Bhatia A, Asfazadourian H, Oberlin C. Ulnar nerve fascicle transfer onto to the biceps muscle nerve in C5–C6 or C5–C6-7 avulsions of the brachial plexus. Eighteen cases. Ann Chir Main Memb Super. 1997;16:275–284. [PubMed]
  • Pearl M L, Edgerton B W. Glenoid deformity secondary to brachial plexus birth palsy. J Bone Joint Surg Am. 1998;80:659–667. [PubMed]
  • Rollnik J D, Hierner R, Schubert M, et al. Botulinum toxin treatment of cocontractions after birth-related brachial plexus lesions. Neurology. 2000;55:112–114. [PubMed]
  • Zancolli E A, Zancolli E R., Jr Palliative surgical procedures in sequelae of obstetrical palsy. Hand Clin. 1988;4:643–669. [PubMed]
  • Mallet J. Obstetrical paralysis of the brachial plexus. II. Therapeutics. Treatment of sequelae. Priority for the treatment of the shoulder. Method for the expression of results [in French] Rev Chir Orthop Reparatrice Appar Mot. 1972;58(suppl 1):166–168. [PubMed]

Articles from Seminars in Plastic Surgery are provided here courtesy of Thieme Medical Publishers