Early efforts at studying the incidence of NBPP in the United States were limited by the information systems of the day. In addition to this, most reviews were also focused on a single institution or metropolitan area (Foad et al. 2008). For instance, a 1975 publication from the University of California reviewed over 19,000 newborn infants and identified only 11 babies with brachial plexus palsy, yielding an incidence of 0.57 per 1,000 live births (Specht 1975). A ten-year analysis (1972–1982) of data from the Kaiser Foundation Hospital in San Francisco identified 61 babies with birth palsy out of a cohort of over 30,000 deliveries for an incidence of 2.0 per 1,000 live births (Greenwald 1984). At Johns Hopkins an 11-year (1993–2004) review revealed a surprisingly high 5.8 per 1,000 live birth rate (135 palsies among 23,273 births) (Gurewitsch 2006).

A 23-year (1980–2002) review of records at the University of Mississippi produced 85 NBPP cases from over 89,000 vaginal deliveries (incidence = 1.0 per 1,000 live births) (Chauhan et al. 2005). These same authors offered provocative evidence-based extrapolation of their data in that they estimated that the average obstetrician (typically 140 deliveries per yr for ACOG fellows) should encounter NBPP once every 7 years and a permanent plexus injury once every 71 years (Chauhan et al. 2005). A 1999 publication of computerized database analysis of over 300 hospitals across California found an incidence of 1.5 per 1,000 live births (1,611 palsies in over one million deliveries) (Gilbert et al. 1999). The first large US national database study of NBPP included over 17,000 palsy patients from a pool of more than 11 million births, yielding an incidence of 1.5 per 1,000 live births (Foad et al. 2008). Interestingly, 54 % of the babies with NBPP did not have known risk factors (Foad et al. 2008).

Multiple risk factors have been suggested as predictive of NBPP, but numerous studies have pointed to shoulder dystocia and macrosomia (usually defined as birth weight >4.5 kg) as being exceptionally strong ones (Foad et al. 2008; Mollberg et al. 2005; Chauhan et al. 2005). In an American national database study, shoulder dystocia and macrosomia were associated, respectively, with a 100 times and 14 times higher risk of neonatal plexus palsy (Fig. 11; Foad et al. 2008). It would then seem to simply be necessary to pre-identify macrosomic infants (perhaps by ultrasound) in order to identify those at increased risk for shoulder dystocia and thus at greater risk for NBPP. But in reality ultrasound is an imperfect predictor of macrosomia (Sacks and Chen 2000; Mehta et al. 2005; Goetzinger et al. 2014; Nguyen et al. 2013), macrosomia is an imperfect predictor of shoulder dystocia (Chauhan et al. 2006; Gherman et al. 2006; Alsunnari 2005), and shoulder dystocia is an imperfect predictor of NBPP (Chauhan et al. 2007; Christoffersson et al. 2003).

Other contemporary facts regarding shoulder dystocia muddy the water even further. It has been shown that there was a tenfold increase in the rate of shoulder dystocia in recent decades (Dandolu et al. 2005), and surprisingly there are mixed results at best regarding the effectiveness of obstetrical educational programs aimed at managing shoulder dystocia (Walsh 2011; Inglis et al. 2011). Many obstetricians consider the only reliable predictor of shoulder dystocia to be a prior maternal history of shoulder dystocia (Bingham et al. 2010; Ouzounian et al. 2012). On top of all of this, rates of NBPP have remained stubbornly steady over the course of decades despite the above noted increase in shoulder dystocia rates as well as substantial increases in Cesarean section rates in the United States (Walsh et al. 2011; Graham 1997). This has led numerous authors to point to the shortcomings of commonly accepted risk factors (Backe 2008; Gurewitsch 2006; Sandmire and DeMott 2005) and others to assert that one or more known risk factors are present in only a minority (46 %) of newborn babies with NBPP (Foad et al. 2008).

This previous discussion shows that as clear as it is in many cases that traction-related birth trauma seems to be the explanation for the neonatal brachial plexus injury, in other instances, it is equally mysterious as to exactly how or precisely when the injury occurred. This controversy regarding the mechanism of injury is not at all new. James Warren Sever conducted extensive infantile cadaver experiments and concluded:

…that traction and forcible separation of the head and shoulder puts the upper cords, the fifth and sixth cervical roots of the brachial plexus, under dangerous tension. This tension is so great that the two upper cords stand out like violin strings. Any sudden force applied with the head bent to the side and the shoulder held would without question injure these cords … One thing impressed me, and that was the evident vulnerability of the upper cords of the plexus under any degree of traction, and I was surprised that the paralysis was not of much more frequent occurrence. (Sever 1916)

But in Sever’s same 1916 report of 470 NBPP patients, there is a solid 7 % who were born following “apparently normal labors” (Sever 1916). Many others have also traditionally interpreted William Smellie’s 1746 birth palsy description (cited earlier in this chapter) as supportive of an intrauterine etiology.

This provocative issue has been taken on in recent years by multiple authors at a variety of centers (Jennett et al. 1992; Sandmire and DeMott 2000; Gonik et al. 2000). In 1992 Jennett et al. concluded that “the data are strongly suggestive that intrauterine maladaptation may play a role in brachial plexus impairment” (Jennett et al. 1992). Bernard Gonik’s classic paper published in 2000 used mathematical modeling to contrast endogenous forces (uterine and maternal expulsive efforts) to exogenous forces (clinician applied forces), concluding that the endogenous forces were four to nine times greater than clinician-related forces (Gonik et al. 2000). This line of research drew attention to potential differences between difficult and seemingly uncomplicated labors.

Gurewitsch and her colleagues looked at this question from the perspective of plexus palsies that occurred with and without shoulder dystocia and concluded that non-shoulder dystocia-associated NBPP was likely to be a mechanically distinct entity (Gurewitsch 2006). They tracked 13 separate risk factors and found that 11 % of shoulder dystocia-associated brachial plexus palsy patients had no identifiable risk factors as compared to 30 % of non-shoulder dystocia plexus palsy patients. These risk factors ranged from well-known ones such as macrosomia and instrumented delivery to somewhat less well-known risk factors such as long second stage of labor and excessive maternal weight gain (Gurewitsch 2006). It thus becomes increasingly clear that alternate mechanism of injury explanations need to be entertained.

A growing number of authors have drawn attention to intended and unintended outcomes associated with oxytocin augmentation of labor (Wei et al. 2010; Mori et al. 2011; Bugg et al. 2011; Mehlman et al. 2012; Costley and East 2012; Miller 2009; Doyle et al. 2011; Ouzounian et al. 2005). High- and low-dose approaches to oxytocin administration exist, and tachysystole (the term used for uterine hyperstimulation) occurs more frequently with high-dose regimens (Wei et al. 2010). In a case-control study of 52 mothers and their NBPP babies versus 132 mothers and their unaffected babies, oxytocin administration and tachysystole were statistically significant risk factors for NBPP (respective odds ratios of 2.5 and 3.7) (Mehlman et al. 2012). It would seem plausible that a child traversing the birth canal with a leading head and trailing shoulder (or perhaps other positions as well) who experiences significant uterine forces might very well sustain NBPP.

In order to have hope regarding scientific progress in the clinical care of NBPP patients, there must be valid and reliable tools on both the “front end” of the disease (e.g., disease severity measures, age-specific rating scales) and the “back end” of the disease (e.g., health-related quality of life, functional outcomes) (Chang et al. 2013). Validity addresses the issue of whether the instrument (questionnaire, rating scale, etc.) measures what it is intended to measure, and reliability speaks to its ability to yield appropriate reproducible measurements. Collectively these are referred to as psychometric properties. Some of these NBPP tools have yet to undergo formal reliability testing while a number of researchers at different neonatal brachial plexus centers have performed important reliability and validity studies for others. New NBPP-specific instruments also continue to evolve (Ho et al. 2012). This section will focus on the most widely used and most widely studied NBPP tools that are currently available.

The classic NBPP disease severity measure is the Narakas classification. In 1987 Algimantas Narakas proposed that based on physical examination at 2–3 weeks of age, patients could be practically divided into four groups (Narakas 1987). The original classification recognized groups 1 and 2 as classic Erb-Duchenne palsy (C5 and C6) and extended Erb-Duchenne (C5, C6, and C7), while groups 3 and 4 were total plexus palsies without and with the presence of Horner syndrome. The very presence of Horner syndrome (aka Klumpke’s sign as she was the first to describe the association between brachial plexus injury and these oculopupillary findings) has been shown to have independent unfavorable prognostic value (see Fig. 12 baby photo) (Al-Qattan et al. 2000). These Narakas groupings have been shown to have prognostic power, with dramatically lower full recovery rates for Narakas 3 and 4 patients (Narakas 1987; Sibinski and Synder 2007; Foad et al. 2009).

Al-Qattan and his Saudi Arabian colleagues have recently reviewed over 500 of their NBPP patients and produced a useful modification of the Narakas classification (see Table 1) (Al-Qattan et al. 2009). They found that Narakas group 2 patients who recovered their wrist extension early (between 2 and 3 weeks of age and 2 m of age) had better rates of spontaneous recovery than those who did not (Al-Qattan et al. 2009). For instance, 76 % of Narakas 2-A patients recovered shoulder abduction, while only 18 % of Narakas 2-B patients did the same. It can be argued that the validity of the Narakas classification system (and now the modified Narakas classification) is self-apparent, but formal reliability studies have yet to be published.

Disease severity is also commonly assessed by the Toronto Test Score , a tool aimed at assessing children less than 1 year of age. This scale was introduced in 1994 by brachial plexus specialists at the Hospital for Sick Children in Toronto, Ontario (Michelow 1994). The Toronto Test Score combines physical examination-derived data from five distinct muscle groups, one of which is elbow flexion (see Table 2). Prior to the Toronto Test Score, many brachial plexus surgeons used just the presence or absence of elbow flexion at a particular age (e.g., 3 months of age) as their main criterion for recommending nerve reconstruction surgery. The Toronto Test Score effectively changed this to multidimensional decision making. The original article showed that the Toronto Test Score at 3 months of age correctly predicted the child’s clinical status at 1 year of age 95 % of the time, while similar use of only elbow flexion was predictive in only 13 % of cases (Michelow 1994).