Prevalence of perinatal factors in infants with brachial plexus birth injuries and their association with injury severity

To determine the prevalence of perinatal factors associated with brachial plexus birth injury (BPBI) in affected infants and their relationship with BPBI severity. Retrospective study of BPBI infants prospectively enrolled in a multicenter registry. The prevalence of perinatal factors was calculated. Infants were stratified by injury severity and groups were compared to determine the association of severity and perinatal factors. Seven-hundred-ninety-six BPBI infants had a mean 4.2 ± 1.6 perinatal factors. Nearly all (795/796) reported at least one factor, including shoulder dystocia(96%), no clavicle fracture (91%), difficult delivery(84%), parity >1(61%) and birthweight >4000 g(55%). Ten-percent (74/778) had Horner’s syndrome and 28%(222/796) underwent nerve surgery. Birth asphyxia and NICU admission were significantly associated with injury severity. NICU admission and asphyxia were associated with BPBI severity. An improved understanding of the relationship between perinatal factors and BPBI severity may be used to guide early referral to BPBI providers and support prevention efforts.


INTRODUCTION
Brachial plexus birth injury (BPBI) refers to upper extremity weakness or paralysis resulting from trauma to the nerve roots of the brachial plexus sustained during childbirth. This injury occurs in approximately 1.5 per 1000 livebirths and is among the most common birth traumas [1,2]. Although many infants recover spontaneously within the first few months of life, 30% experience more severe injuries [3][4][5][6][7] that result in persistent weakness and sensory deficits, impaired musculoskeletal development [8,9], functional limitations [10][11][12], and psychosocial consequences [10,12,13], which may persist into adulthood [14]. To optimize upper extremity function and mitigate the musculoskeletal consequences of BPBI, infants with incomplete or absent neurologic recovery often undergo nerve surgery and other rehabilitative interventions within the first year of life [15]. To facilitate early intervention and allow providers adequate time for serial physical exams, making treatment decisions, and counseling caregivers, referral to a BPBI provider within the first few months of life is recommended [16][17][18][19][20][21].
Although the exact mechanism of BPBI has not been definitively elucidated, several perinatal factors have been associated with BPBI, including multi-parity, gestational diabetes, fetal macrosomia, fetal malposition, instrumented delivery, shoulder dystocia, and birth asphyxia [1,2,19,[22][23][24][25]. Much of our recent understanding of these factors comes from investigations using large administrative datasets, including insurance claims or state health databases that evaluate factors associated with BPBI at the population level. However, these databases only provide insight into factors associated with BPBI coded during the infant's birth admission and may not capture factors that were not diagnosed or coded incorrectly. In addition, these data often contain information only from the infant's birth record, but not maternal health data which may be relevant to BPBI. Factors identified in infants with BPBI are compared to those without BPBI and may not reflect the prevalence within the affected population. Consequently, it is unknown if the factors identified in these studies are generalizable to the population of infants who present to a BPBI center for care. Furthermore, administrative datasets allow for cross-sectional analysis, but do not follow affected infants longitudinally, and therefore are unable to evaluate the association of these factors with indicators of injury severity, including undergoing nerve surgery. Improved understanding of the prevalence and distribution of perinatal factors associated with BPBI within an affected population and their relationship with injury severity would improve characterization of infants with BPBI, including those with severe injury that are most likely to benefit from early referral and intervention. The primary purpose of this study was to identify the prevalence of perinatal factors associated with BPBI in a large national cohort of affected infants. Secondarily, we sought to describe associations between these perinatal factors and injury severity.

MATERIALS AND METHODS
With Institutional Review Board approval, we conducted a retrospective cohort study of infants with BPBI who were prospectively enrolled in the multicenter Treatment and Outcomes of Brachial plexus Injury (TOBI) registry. This study was performed as a secondary study aim. The TOBI registry is comprised of infants with BPBI who were evaluated at one of 6 regional BPBI centers across the United States between 2000 and 2016. Infants in the registry were enrolled prior to 12 months of age and followed until 5 years old. Upon study enrollment, demographic information and birth characteristics were collected from the infant's caregiver. Additionally, data was collected regarding twelve perinatal factors associated with BPBI in previous studies: maternal parity, gestational diabetes, pre-eclampsia, delivery position, shoulder dystocia, birthweight greater that 4000 g, gestational age greater than 40 weeks, no clavicle fracture, history of difficult delivery with a previous infant, difficult delivery with the present infant, asphyxia, and neonatal intensive care unit (NICU) admission following birth. All infants enrolled in TOBI were included. Infants missing data for 2 or more of the 12 factors were excluded. Injury severity was defined as presence of a Horner's sign on the enrollment exam (asymmetric eyelid ptosis on the ipsilateral side as the BPBI) or undergoing nerve surgery for BPBI (at least 1 nerve graft or transfer) within the first 12 months of life. Surgical decisions were made by the operating surgeon; at study onset, all operating surgeons agreed to adhere to the principle of offering surgery to infants with an Active Movement Score (AMS) elbow flexion score of <5 of 7 and/or a Toronto score of <3.5 of 10 [26].
The prevalence of BPBI risk factors were summarized for the cohort and stratified by injury severity (presence vs absence of a Horner's sign; undergoing nerve surgery in infancy or not). Comparisons across groups were conducted using chi-squared tests or Fisher's exact tests as appropriate. Statistical analysis was performed using R Statistical Software (v4.1.2; R Core Team 2021). All statistical tests were two-sided. Statistical significance was evaluated at p < 0.05.

RESULTS
796 infants with BPBI (55% female, 55% right-sided) were included in this investigation. Their demographic and birth characteristics are reported in Table 1. The frequency of the 12 evaluated risk factors are included in Table 2. The most prevalent risk factors that occurred in more than half of the cohort included shoulder dystocia (96%), absence of clavicle fracture (91%), difficult delivery (84%), parity >1 (61%) and birthweight >4000 g (55%). Asphyxia and NICU admission following delivery were reported in 25% and 35%, respectively.
Nearly all infants (795/796, >99%) had at least one of the evaluated factors, with a mean of 4.2 ± 1.6 perinatal factors. Factors frequently occurred in combination.
Ten percent of the cohort (74/778) were diagnosed with a Horner's sign. Compared to infants without a Horner's sign, infants with a Horner's sign had a significantly higher prevalence of asphyxia (36% vs. 24%, p = 0.03) and NICU admission (54% vs 33%, p = 0.001), and lower prevalence of gestation age >40 weeks (4% vs. 13%, p = 0.04). (Table 3) There was no significant difference in prevalence of the other evaluated factors between infants with and without a Horner's sign. There was a statistically, but not clinically, significant difference (p = 0.02) in the number of risk factors in the Horner's sign group (4.6 ± 1.6) compared to the non-Horner's group (4.1 ± 1.6).
Twenty-eight percent of the cohort (222/796) underwent nerve surgery for BPBI within the first 12 months of life. There was a significantly higher prevalence of asphyxia (31% vs. 23%, p = 0.03) and NICU admission (44% vs. 32%, p = 0.003) in infants who underwent nerve surgery compared to those who did not. (Table 4) There was no significant difference between groups in the prevalence of the other analyzed factors. There was a statistically, but not clinically, significant difference (p = 0.002) in the mean number of risk factors in nerve surgery group (4.5 ± 1.6) compared to the non-nerve surgery group (4.1 ± 1.6). 43% of BPBI infants admitted to the NICU and 34% with asphyxia underwent nerve surgery.

DISCUSSION
This study found that nearly all infants with BPBI had at least one, and often multiple, of the commonly reported factors. The most common factors in this study are similar to those frequently identified in other studies [1,2,19,[22][23][24]. In addition to many of the most commonly recognized characteristics, NICU admission and birth asphyxia, which are less frequently investigated, were prevalent in our cohort, at 35% and 25%, respectively. Moreover, both of our indicators of injury severity were significantly associated with NICU admission and birth asphyxia. Of particular relevance to providers caring for affected infants, 43% of BPBI infants admitted to the NICU and 34% of BPBI infants with asphyxia underwent nerve surgery in infancy.
Characterizing factors associated with a medical condition is important for several reasons, including informing the diagnosis, determining prognosis, guiding interventions, and ultimately, understanding the mechanism of disease or injury to develop prevention strategies. Presently, BPBI risk factors are principally used to support the diagnosis and guide perinatal and intrapartum prevention strategies; the association of shoulder dystocia, fetal macrosomia, and gestational diabetes with BPBI forms the basis of many of the obstetricians' perinatal and intranatal prevention strategies. Less commonly have factors associated with BPBI been used for prognosis, in part because many datasets that provide information regarding these factors do not contain longitudinal data that would allow assessment of their relationship with longer terms outcomes. Many previous investigations of BPBI risk factors consist of single institution case series which may not be generalizable to a broader population [27,28]. More recently, factors associated with BPBI have been studied at the population level using large administrative databases to compare the presence of potential risk factors in affected and unaffected infants. These studies leverage the power of large datasets to investigate the association of perinatal and intrapartum factors with BPBI that would not be possible in smaller datasets, and have vastly expanded our understanding of BPBI risk factors at the population level [2,23]. Several authors have looked at the prevalence of BPBI-associated factors using the Kids' Inpatient Database (KID) [1,2,24], a multistate claims database that provides an estimated sample of pediatric inpatient hospitalizations, including childbirth, which permits extrapolation to population level estimates. Using the KID, DeFrancesco et al evaluated the incidence of BPBI from 1997 to 2012 and identified shoulder dystocia, macrosomia, instrumented delivery, and birth hypoxia as the most common risk factors for BPBI. In addition, they found that increasing number of risk factors was associated with an increasing risk of BPBI [2]. These findings were replicated by Foad et al. [1] and Abzug et al. [24] also using KID data.
Other investigators have used state administrative databases, which contain actual count of infant births rather than estimated samples, to evaluate the incidence of BPBI and perinatal risk factors and reported similar findings as the KID dataset. Lalka et al. [23] used Colorado Hospital Association data to review the birth claims of 966,447 infants born in a Colorado hospital from 2000 to 2014. The authors found that shoulder dystocia was associated with the highest increased risk for BPBI, followed by forceps birth, breech delivery, and gestational diabetes. Similarly, Gilbert and colleagues [22] evaluated the prevalence of factors associated with BPBI using linked maternal-infant data from California and found that shoulder dystocia (53.4%), fetal malposition (25.1%), and birth hypoxia (7.5%) were among the most common factors associated with BPBI. The finding that shoulder dystocia is one of the most common factors associated with BPBI has been replicated in populations outside the United States [29][30][31][32][33].
The most striking difference between our investigation and studies using administrative datasets is the proportion of infants with and without risk factors. Whereas KID studies report up to 55% of BPBI infants have no known risk factor [1,2], nearly all infants in our study had at least one of the perinatal factors Cumulative risk factor data was calculated for subjects who were missing data on no more than 2 risk factors. Cumulative risk factor data was calculated for subjects who were missing data on no more than 2 risk factors. associated with BPBI. One explanation is the population of infants diagnosed with BPBI during their birth admission are different than the population who present to a BPBI provider; it may be that that infants without risk factors sustain more mild BPBI that resolve within the first few weeks of life and do not present to a BPBI provider. Alternatively, this difference may arise from the fact that the present study evaluated a different and broader set of factors than many of the studies using administrative datasets. Because claims datasets are often designed for billing or distribution of state resources, rather than research, they may not contain clinically relevant birth data that this study was able to capture from caregivers. In addition, administrative datasets principally contain information from only the infant's medical record and may not include maternal data relevant to the injury that we obtained from caregiver report; in particular, risk factors identified prenatally (gestational diabetes, maternal obesity, preeclampsia, history of a previous difficult delivery) or intranatally (labor induction, instrumented delivery) may not be reliably or accurately coded in the infant's medical record. For these reasons the number and type of factors analyzed in this study differs from that utilized in the studies using administrative datasets. An important finding of this study is the identification of perinatal factors associated with BPBI severity, including NICU admission and birth. Previous studies evaluating the association of perinatal factors and injury severity have identified increasing fetal birthweight and macrosomia as a risk factor for "permanent" BPBI compared to "transient" BPBI [31,[34][35][36]. Furthermore, Pondaag and colleagues [37] evaluated the effect of birthweight on intraoperative nerve findings in 206 infants with BPBI treated surgically and found that increasing birthweight was significantly associated with increasing extent of injury. Lastly, Shah et al. [38] evaluated a subset of the TOBI data to identify factors that predicted nerve surgery in infancy in 365 infants with BPBI. After adjusting for known BPBI risk factors, the authors identified early physical exam findings (weak (Active Movement Scale score <4.5) finger/thumb flexion, wrist flexion, wrist extension, and elbow flexion) independently predicted undergoing nerve surgery in infancy. Unfortunately, different definitions of injury severity preclude direct comparison between investigations.
Perhaps the most important clinical application of this study's findings is that it provides additional guidance to healthcare providers regarding perinatal factors that should prompt early referral to BPBI provider. Although no conclusive evidence exists for the association between BPBI outcomes and timing of referral, early evaluation by a BPBI practitioner, ideally a multidisciplinary clinic, is recommended by BPBI providers [16][17][18]33], the American College of Obstetrics and Gynecology [19], American Academy of Pediatrics [20], and the Canadian Obstetrical Brachial Plexus Injury practice guidelines [21]. Expert consensus indicates that early referral allows caregiver education [39], serial assessments for spontaneous recovery and making treatment decisions [40] and timely operative intervention [26,41,42]. Moreover, caregivers of affected infants report preferring early referral to alleviate confusion and frustration due to lack of information about BPBI [39,43,44]. Despite these benefits, delayed referral or failure to refer is common [16,32]. Coroneos et al. [32] reviewed a demographic sample of infants born in Canadian hospitals between 2004 to 2012 and found that the majority of infants with BPBI (55-60%) were not referred to BPBI specialist, and among those who were referred only 28% were referred within Canada's recommend time of one month of age and only 66% within three months of age. Given the lack of a nationalized health care system and designated BPBI centers in the United States, it is likely that referral practices in the United States are no better. Although reasons for delayed referral are likely multifactorial, one commonly reported reason is inadequate understanding of current referral and treatment recommendations [16]. By reporting perinatal factors commonly observed in a cohort of infants with BPBI and factors associated with injury severity, we provide additional guidance for providers regarding which infants in particular benefit from prompt referral. While we agree with previous recommendations that all infants with BPBI should be referred to a BPBI provider early in life, it is imperative that infants with known risk factors, including NICU admission and birth asphyxia, be referred promptly.
There are limitations to this study. Although this cohort is likely typical of infants with BPBI evaluated at referral centers, it may not be representative of all infants with BPBI and may be subject to referral bias. Nevertheless, this information is likely clinically useful for providers at these types of centers. Another limitation is that factors evaluated in this study were identified by caregiver report, which is subject to recall and other biases. Additionally, we used Horner's sign as a proxy for global lesions rather than a more traditional measure of severity, such as Narakas type (number of nerve roots affected), because Narakas type is determined by clinical exam at 3 weeks of age and this was not always possible in our study cohort. Previous analyses indicate that Horner's sign is a Cumulative risk factor data was calculated for subjects who were missing data on no more than 2 risk factors.
reliable measure of injury severity compared to clinical exam (AMS scores) [26]. Lastly, we were unable to assess all factors that may be associated with BPBI, including other obstetric, intrapartum and prenatal factors, demographic characteristics, and social determinants of health, and therefore our study does not comprehensively characterize BPBI risk. This study characterizes the prevalence of risk factors associated with BPBI in an affected population and identifies risk factors associated with injury severity, including undergoing nerve surgery in infancy. This information provides guidance for providers regarding which infants benefit from early referral to a BPBI provider. Based on this study's findings, we conclude that is imperative that infants with known risk factors, including NICU admission and birth asphyxia, be referred promptly to a BPBI specialist.

DATA AVAILABILITY
All data generated or analyzed during this study are included in this article and/or its supplementary material files. Further enquiries can be directed to the corresponding author.