Neonatal subgaleal hemorrhage: twenty years of trends in incidence, associations, and outcomes

In 2011, we reported 38 neonates with subgaleal hemorrhage (SH), relating an increasing incidence. It is unclear whether the incidence in our hospitals continued to rise and which risk factors and outcomes are associated with this condition. We retrospectively analyzed every recognized case of SH in our hospitals from the end of our previous report (2010) to the present (2022). We redescribed the incidence, scored severity, tabulated blood products transfused, and recorded outcomes. Across 141 months, 191 neonates were diagnosed with SH; 30 after vacuum or forceps. The incidence (one/1815 births) was higher than in our 2011 report (one/7124 births). Also, severe SH (requiring transfusion) was more common (one/10,033 births vs. one/20,950 births previously). Four died (all with severe SH) and 12 had neurodevelopmental impairment. Recognized cases of SH are increasing in our system without a clear explanation. Adverse outcomes are rare but continue to occur.

Twelve years ago, we reported 38 neonates with SH over an 8½ year period [14]. Our overall incidence was one case per 7124 live births, with one case/598 vacuum-assisted births and one/1059 forceps-assisted births. Case numbers increased annually over that 8½ year span, but it was unclear whether this was an actual increasing incidence or an increase in awareness and diagnosis. We reported that four of the 38 had adverse outcomes (death in two and neurodevelopmental impairment in two). We also found that blood transfusions were administered only to those who had vacuum or forceps-associated SH, and never to those with SH in the absence of those instrumentations [14].
Since that report, we have been aware that SH has continued to occur in our healthcare system. However, we have not tracked recent rates, nor have we assessed whether any trends occurred in incidence or outcomes. With a desire to better understand these issues in our system, we undertook the present retrospective analysis. It includes all cases of neonatal SH recorded at all Intermountain Healthcare hospitals beginning with the end-date of the previous report (July 31,2010) through March 31, 2022. Our aims were to compare the following over the past two decades; (1) the annual incidence of recognized cases of neonatal SH, (2) known fetal/neonatal and obstetrical risk factors for SH, (3) the number of transfusions, and the specific blood components administered, to each neonate recognized to have a SH, and (4) adverse outcomes (death or neurodevelopmental impairment) among these neonates when assessed at 24 ± 6 months-old.

METHODS
Prior to data collection, the Institutional Review Board (IRB) of Intermountain Healthcare classified this project as quality improvement. The IRB reviewed this manuscript prior to our submitting it for publication and judged it to be compliant with the institutional privacy-protection policies.
Our review was a multihospital retrospective records analysis of neonates with dates of birth from August 1, 2010, through March 31, 2022 in the Legacy Intermountain Healthcare databases. Intermountain Healthcare is a not-for-profit organization that owns and manages hospitals in the Intermountain West of the USA. All diagnoses were initially ascertained from electronic data marts; Case Mix (the billing, coding, and financial data mart used by Intermountain Healthcare), EVOX (the extended Vermont-Oxford database), Storkbytes (the labor and delivery database), Fetal Link (a replacement for Storkbytes), provider problem lists, and ICD-9 and ICD-10 coding.
The electronic medical records of each neonate identified by using the above methods were reviewed by the authors, and the severity of the SH was determined for each case, using the criteria of Chadwick et al. [1] as modified by us (Table 1). Our modification divided moderate cases into three subgroups depending on whether phototherapy, or fluid bolus, or both were administered. To be classified as having a severe SH, we required administration of one or more blood-product transfusions, independent of degree of hyperbilirubinemia. To provide additional emphasis on severe SH we performed an analysis of the original 13 neonates with severe SH from our previous report [10] and compared them to those we identified during the present reporting period.
Charts of all neonates with a diagnosis of SH were individually reviewed by a member of the research team, not relying on coded information or on data tables to assign the delivery practices, clinical features, diagnoses, or neurodevelopmental follow-up information. For most neonates with SH of mild or moderate severity, outcome at 24 ± 6 months was assessed by a general pediatrician in our healthcare system. Those with neurodevelopmental problems were evaluated in either the pediatric neurology clinic or the neonatal follow-up clinic. A statement by the pediatrician or neurologist that the child had "normal" development resulted in the classification of "no neurodevelopmental impairment (NDI) detected". If the pediatrician or pediatrician stated that developmental delays were present, but the child had neither blindness, profound hearing loss, or severe delays we classified this as "moderate NDI". Blindness, or profound hearing loss, or severe delays constituted "severe NDI".
The dataset was collected and managed using an Intermountain Healthcare REDCap (Research Electronic Data Capture) electronic data capture tool. REDCap is a secure cloud-based application designed to support data capture and provide an intuitive interface for validated data entry, audit trails for tracking data manipulation and export procedures, automated export procedures for seamless data downloads to common statistical packages, and procedures for importing data from external sources. Summary statistics (means, counts, and proportions) were the primary quantitative tools used for analysis. Differences in continuous variables by group were assessed using one-way ANOVA. Differences in categorical variables were assessed using either chi-square tests or Fisher's exact test. Data management and statistical analysis were done in the R language and environment for statistical computing (R Foundation).

RESULTS
From August 2010 through March, 2022, 346,681 live births were recorded at the Intermountain Healthcare hospitals. During that period, 191 neonates were diagnosed with SH ( Table 2). The overall incidence for the recent period was one case per 1815 births. This included one case per 558 vacuum-assisted births, one case per 849 forceps-assisted births, one case per 68 births where both vacuum and forceps were used, and one case per 2,048 births where neither vacuum nor forceps were used. Figure 1 shows the annual incidence of SH during the past 20 years stratified by severity (increased from the previous assessment of 1 per 7124 births). The increase involved not only mild and moderate cases, but also severe cases. In our 2011 report, the rate of severe SH was one case per 20,950 live births. In our present report the rate doubled, at one case per 10,033 live births. Table 3 displays features of the neonates diagnosed with mild, moderate, and severe SH. Almost all were term or late-preterm neonates. Those who developed moderate SH and received fluid boluses, and those who developed severe SH and received transfusions, were more likely to have had low Apgar scores at one and five minutes. Almost all cases were diagnosed on the day of birth. The majority (65%) were born to primigravida women, and 45% were delivered by cesarean section after arrested 2nd stage of labor. Twenty-one percent were born after vacuum, forceps, both, or manual head rotation, but the remaining 79% had none of those interventions charted. Fourteen underwent whole body therapeutic hypothermia for moderate (n = 12) or severe (n = 2) hypoxic ischemic encephalopathy. Adverse outcomes were more likely with severe SH: 31% died or had moderate or severe NDI at two years. The two deaths were the result of unremitting acidosis and hypotension accompanying DIC, despite multiple blood product transfusions and intensive care support. Both deaths in the recent 12-year period had been transferred within the Intermountain system for a higher level of NICU care.
During the past 20 years, 47 neonates with SH received one or more blood transfusions in our healthcare system (Supplementary Table). Thirty-seven of the 47 also received fluid boluses of normal saline ranging in volume from 10 to 120 mL/kg, in addition to blood transfusions. None of the fluid boluses were administered using a fluid/blood-warmer system [15]. The transfusions consisted of 0 to 120 mL/kg of PRBC, 0 to 115 mL/kg fresh frozen plasma (FFP), 0 to 60 mL/kg platelets, and 0 to 60 mL/kg cryoprecipitate. None of these were administered using a blood-warmer system, thus, as with the fluid boluses, all were transfused at or below room temperature [16]. Three were treated with rFVIIa; these include one of the four who died. None were treated with tranexamic acid.

DISCUSSION
Severe neonatal SH is an uncommon but high-acuity condition. Because severe SH can rapidly lead to hemorrhagic shock and death, or neurologic injury leading to disability, all professionals who provide medical care to newborn infants should be familiar with this entity. The primary purpose of the present data extraction was to produce updated clinical outcome information for all such professionals in our healthcare system. Combined with our previous report [14], we now have a twenty-year assessment of neonatal SH cases diagnosed at the Intermountain Healthcare hospitals. Viewing these analyses, it appears that our incidence is increasing. In our previous report, we identified one case of SH per 7124 live births, and an increase over the 8 ½ year period analyzed. In the present analysis, we identified one case per 1815 live births, an almost four-fold increase in incidence between the two reports. We recognize that the apparent increase in our incidence might be due, at least in part, to increasing ascertainment or to unrecognized changes in referral practices. Our most  recent incidence figure is higher than that reported by Plauche, who found an incidence of one per 2500 deliveries [17], and by Swanson et al. who reported an incidence of one in 2000 births [8]. As seen in Fig. 1, in recent years we may be identifying more mild and moderate cases than we found in previous years. However, the number of severe cases (those receiving a transfusion) has also increased during the two reports. In our earlier analysis we identified one severe SH case for every 20,950 live births, and in the present analysis we found one severe SH case for every 10,033 live births. Thus, more accurate diagnosis and better reporting might account for some of our apparent increase in incidence. However, the fact that our incidence of severe cases doubled causes concern that our true incidence of SH is increasing, and without a clear explanation.
Devising and implementing more effective treatment strategies for severe SH cases is needed in our healthcare system, because over the past 20 years we have had 16 poor outcomes from this condition including four deaths. We see four potential treatment advances that we had yet not organized into a "severe SH treatment pathway". As listed in Table 4 these include: (1) recombinant activated Factor VII to facilitate hemostasis [18,19]; (2) tranexamic acid (TXA) to facilitate thrombus stabilization [20][21][22]; and (3) transfusing hemorrhaging neonates through a blood warmer system with low-titer, cold-stored, type O whole blood (LTOWB) [23][24][25][26], as opposed to the serial non-warmed component transfusions we have been using; and (4) stimulating red cell production using darbepoetin.
Treating hemorrhagic shock in a neonate often requires treating coexisting disseminated intravascular coagulation (DIC). Therapies for this condition have historically involved sequential component blood product transfusions: packed red blood cells to correct anemia, fresh frozen plasma (FFP) to replace consumed clotting factors, cryoprecipitate to increase concentrations of fibrinogen, factor VIII, factor XIII, and von Willebrand factor, and platelets to correct thrombocytopenia [27]. Case reports of neonates with hemorrhage and DIC suggest a benefit of adjunctive treatment with TXA and/or with recombinant activated Factor VII [18][19][20][21][22].
Recent advances in mass transfusion for pediatric hemorrhagic shock include the benefit of LTOWB transfusion. Leeper et al. reported that whole blood transfusion is safe in the pediatric population, with no difference in adverse events compared to component transfusions [28][29][30]. However, LTOWB has the benefit of reduced donor exposure, decreased coagulopathy, and  potentially shorter ICU stay [31]. Our healthcare system recently approved incorporating LTOWB for emergency release blood transfusion of neonates, as an alternative to component transfusions in the setting of acute severe blood loss [24]. This protocol, which we have used successfully in extremely low gestational age newborns [26], supports the hypothesis that emergency transfusion of LTOWB could reduce total volume transfused and potentially improve outcomes.
Additionally, inadvertent or unplanned hypothermia can be detrimental to NICU patients [32,33]. Fluid boluses and largevolume transfusions administered below physiological temperature can contribute to hypothermia. Red blood cells are stored at 1°C to 6°C, plasma and cryoprecipitate is frozen, and platelets are kept at room temperature. Normal saline for fluid boluses is typically maintained at room temperature but is sometimes refrigerated. Several studies of adult patients indicate that bringing bolus fluids and blood products up to physiological temperature, with blood/ fluid warmers, is safe and effective, is much more comfortable to conscious patients, and helps maintain normothermia [34][35][36][37]. In preliminary studies, we found that warming fluids and blood products, including LTOWB, to physiological temperature appears to be safe for neonatal recipients [15,16]. Although additional study is needed, perhaps bringing bolus fluids and transfusions to physiological temperature during the infusions could be another benefit to neonates with severe SH.
The use of red cell growth factors such as erythropoietin and darbepoetin has increased in preterm and term infants, and would benefit infants with moderate or severe SH by decreasing transfusion needs [38][39][40]. In addition, studies evaluating potential neuroprotective effects of darbepoetin are ongoing. A neuroprotective effect may turn out to be an additional benefit of weekly administration to these infants during their hospitalization.
We recognize that our report has shortcomings. First, because this was a retrospective analysis it lacks the rigor of prospectively collected data. A prospective accumulation of relevant SH information, with prescribed data-acquisition fields, would be a superior design. Also, we recognize that the population of the Intermountain West is not representative of the entire United States. It is unclear to what extent genetic, cultural, and regional differences are relevant to SH incidence or outcomes. Consequently, it is uncertain how confidently our findings can be applied to other regions of the country or world. Another shortcoming is our failure to clarify the mechanism of severe SH among neonates without recognized obstetrical risk factors. In fact, 23 of our 37 severe cases occurred after non-instrumented deliveries. Six of the 23 underwent manual rotation of the fetal head from an occiput-posterior position, to facilitate delivery [41]. Attempted or successful manual rotation of the fetal head is not routinely documented in the Intermountain Healthcare system. Given that an accurate denominator cannot be obtained for that intervention, we cannot conclude whether it is indeed a risk factor for SH. Improved documentation of this practice, and communication to the pediatric team, is warranted and should help better understand the associated risks. Also, absence of a systematic universal neurodevelopmental assessment for patients diagnosed with subgaleal hemorrhage limits strong conclusions about neurodevelopmental impairment. An additional weakness is the lack of consistent imaging in attempt to document and quantify the hemorrhage, specifically head ultrasound or CT, since these procedures were rarely performed on our patients, even in severe cases.
Despite these weaknesses in our report, we maintain that the new information describing our trends and outcomes can add to our educational efforts for professionals caring for newborn infants in our healthcare system, and perhaps in other hospitals too. Such efforts will have the primary aims of improving recognition of SH, and in severe cases, improving awareness of new potential adjunctive treatments. A practice model for emergency transfusion of a neonate with SH, developed for the Intermountain Healthcare hospitals, is included as a supplemental figure. This material and other educational efforts and programs have the goal of producing better outcomes for neonates with SH.

DATA AVAILABILITY
A deidentified dataset is available by written request to the corresponding author.