To identify antecedent clinical and health services events in infants (35 weeks gestational age (GA)) who were discharged as healthy from their place of birth and subsequently sustained kernicterus. We conducted a root-cause analysis of a convenience sample of 125 infants 35 weeks GA cared for in US healthcare facilities (including off-shore US military bases). These cases were voluntarily reported to the Pilot USA Kernicterus Registry (1992 to 2004) and met the eligibility criteria of acute bilirubin encephalopathy (ABE) and/or post-icteric sequelae. Multiple providers at multiple sites managed this cohort of infants for their newborn jaundice and progressive hyperbilirubinemia. Clinical signs of ABE, verbalized by parents, were often inadequately elicited or recorded and often not recognized as an emergency. Clinical signs of ABE were reported in 7 of 125 infants with a subsequent diagnosis of kernicterus who were not re-evaluated or treated for hyperbilirubinemia, although jaundice was noted at outpatient visits. The remaining infants (n=118) had total serum bilirubin (TSB) levels >20 mg per 100 ml (342 μmol l−1; range: 20.7 to 59.9 mg per 100 ml). No specific TSB threshold coincided with onset of ABE. Of infants <37 weeks GA with kernicterus, 34.9% were LGA (large for gestational age) as compared with 24.7% of term infants (>37 weeks GA). Although >90% mothers initiated breast-feeding, assessment of milk transfer and lactation support was suboptimal in most. Mortality was 4% (5 of 125) in infants readmitted at age 1 week. Along with a rapid rise of TSB (>0.2 mg per 100 ml per hour), contributing factors, alone or in combination, included undiagnosed hemolytic disease, excessive bilirubin production related to extra-vascular hemolysis and delayed bilirubin elimination (including increased enterohepatic circulation, diagnosed and undiagnosed genetic disorders) in the context of known late prematurity (<37 weeks), glucose 6-phosphate-dehydrogenase deficiency, infection and dehydration. Readmission was at age 5 days in 81 of 118 (69%) infants and <10 days in 101 of 118 (86%) infants. TSB levels were 35 mg per 100 ml (598 μmol l−1) in 46 (39%) infants, of whom one died before exchange transfusion, one was untreated and one was lost to follow-up. Timely and efficacious bilirubin reduction interventions defined by ‘crash-cart’ initiation of immediate intensive phototherapy and urgent exchange transfusion were accomplished in 11 of 43 infants, which were compared with 12 of 43 infants in whom a timely exchange sometimes could not be accomplished. No overt sequelae were found in 8 of 11 infants (73%) treated with a ‘crash-cart’ approach compared with none without sequelae when exchange was delayed by pre-admission delays, technical factors or need to transfer to a tertiary facility. None of the remaining 20 of 43 infants treated only with phototherapy escaped sequelae. Regardless of age at readmission and intervention, infants with peak measured TSB >35 mg per 100 ml had post-icteric sequelae (n=73). There was a narrow margin of safety between birthing hospital discharge or home birth and readmission to a tertiary neonatal/pediatric facility. Progression of hyperbilirubinemia to hazardous levels and onset of neurological signs were often not identified as infant's care and medical supervision transitioned during the first week after birth. The major underlying root cause for kernicterus was systems failure of services by multiple providers at multiple sites and inability to identify the at-risk infant and manage severe hyperbilirubinemia in a timely manner.
Classic signs of acute bilirubin encephalopathy (ABE) in the severely hyperbilirubinemic term infant have been described by van Praagh,1 Jones,2 Volpe,3 and Perlstein.4 These include tone abnormalities such as hypotonia alternating with progressive hypertonia of extensor muscles, with retrocollis and opisthotonos, in association with varying degrees of drowsiness, lethargy-decreased feeding and irritability. When described in terms of the infant's mental status, muscle tone and cry, as shown in Table 1,5 progression of ABE can be documented and provides a schema for grading its severity.5, 6, 7, 8 Increasing scores would be indicative of worsening signs of acute neurotoxicity. The earliest signs of ABE are non-specific and subtle and may be missed unless elicited by direct questioning of parents and close clinical observation. Moderate ABE (score of 4 to 6) has been considered as definitive signs of kernicterus and include beginning arching of neck and trunk on stimulation, alternating with increasing lethargy, decreased feeding, unexplained irritability and usually accompanied by a shrill cry. During the early phases, prompt and effective interventions could prevent chronic kernicteric sequelae. Advanced signs are progressive and marked by cessation of feeding, bicycling movements, inconsolable crying with irritability, inability to feed, fever, seizures and coma. These late findings are ominous predictors of the probability of severe kernicteric sequelae, even with intensive treatment. The extent of brain damage is likely to be reduced by rapid reduction of the bilirubin load (by a combination of intensive phototherapy and exchange transfusion). Rate of bilirubin rise, duration of hyperbilirubinemia, adequacy of albumin-binding reserves, level of unbound bilirubin, host susceptibility and presence of co-morbidities, individually or in combination, have been implicated but not confirmed as important to the onset and progression of ABE.6, 9, 10, 11, 12 Death in an infant with ABE is likely to be due to respiratory failure, progressive coma or intractable seizures. Currently, the transition from increasing severity of hyperbilirubinemia to ABE is unpredictable.
Preventive management of progressive hyperbilirubinemia as the predecessor of potential ABE is the most effective of clinical strategies to prevent kernicterus.13, 14, 15, 16, 17, 18 Implementation needs to be a system-based approach that allows for individualized care to accommodate the clinician's concerns, informed participation of the family and monitoring of the progression of hyperbilirubinemia of at-risk newborns.5 Practice parameters developed by the American Academy of Pediatrics (AAP) provide useful guidelines for the management of term healthy newborns when these are followed diligently.13, 14, 15, 16, 17 These include a ‘crash-cart’ approach to prevent or minimize sequelae of ABE.16, 18
Randomized controlled trials to prevent kernicterus are not ethically feasible because of easy access to effective treatment options for severe hyperbilirubinemia. Thus, the only available evidence that would best delineate strategies to improve access to effective and efficient treatment is through a detailed root-cause analysis of infants who developed kernicterus with current healthcare systems. We conducted a root-cause analysis of a convenience sample of all cases that were reported to the Pilot USA Kernicterus Registry from 1992 to 2004, prior to the updated AAP guidelines.17 Our hypothesis was that we would identify additional clinical observations and common recurring lapses that could be useful to delineate more effective systems strategies to inform a safer management of newborn jaundice. Preliminary data from the Registry5 were provided to the AAP Subcommittee on Hyperbilirubinemia12 and presented at the 2004 National Institute of Child Health and Development Workshop.7, 19, 20, 21, 22, 23, 24
Patients and methods
Kernicterus registry database
176 cases voluntarily reported to the Registry from 1992 were reviewed, scrutinized and entered into a specifically designed computer database (ACCESS) and constituted the Kernicterus Registry Database. Subsequent to raw data entry, each case report was individually assessed by two separate reviews by the research nurse (KK), the co-principal investigator (LJ) and principal investigator (VKB), and, when needed, a pediatric neurologist (SMS). All patient identifications were removed and replaced by unique identification numbers.
The Pennsylvania Hospital Review Board (University of Pennsylvania) and the Stanford University Research Review Board independently approved the project. Enrollment ended on 13 April 2004. To meet a follow-up requirement for at least 18 months, the last calendar year of birth was 2002. Follow-up data are current through April 2004. A large number of parents who were familiar with the Registry participated in detailed interviews and volunteered their participation. Others have thus far chosen to remain anonymous. We conformed to the Health Insurance Portability and Accountability Act Regulations on a case-by-case basis. The database was maintained with unique identifiers for confidentiality of children with kernicterus. As these children grow and mature, this database is likely to be a valuable resource for the children to link their detailed birthing period data to a potential relationship with onset of problems and disorders in childhood and later life.
The specific and unique nature of the custom-built program was adapted to accommodate the multiple clinical datasets gleaned from the medical records, medico-legal depositions, physician office records and parental feedback. Software was configured to accommodate data entry and allow for visualization of temporal relationship and progression of hyperbilirubinemia, intervention, as well as follow-up. The configuration was designed to minimize qualitative data and to use a quantitative scaling or indicators that were collated to the closest age-in-hours. The database directs the operator to record specific information, including missing information, and to seek the closest time of the day to indicate the occurrence of the specific event. The unique nature of the design was to minimize bias, but allow for a recording of comments for any unclear documentation from the chart. Principal investigators adjudicated these comments. Unresolved qualitative data that were prone to a biased interpretation were reviewed and discussed after deliberation and reflection. During the first half of data entry, modifications and labeling of the software and fields for data entry was done simultaneously to ensure user-friendly computer interface. Consensus-based definitions of diagnosis, interventions, sequelae and lapses in care were configured to the software concurrent to entry of raw data from the latter half of the cases. Upon complete data entry and its review for accuracy as well as resolution of queries, software-driven analysis was generated by the program. These unbiased analyses were studied and reviewed. Complexities of data collected were made available for statistical analysis, root-cause analysis and descriptive reports. From a statistical perspective, this sample was small and vulnerable to over-analysis. To better understand the common root causes of these adverse clinical outcomes and identify gaps in the contemporaneous healthcare system, we reviewed the root causes that would be amenable to structural correction or modifications. We were hesitant to extend the data analysis to a specific statistical design and over-analysis.
Clinical analysis included: (a) standardization of definitions for hyperbilirubinemia and kernicterus, (b) demographic analysis, (c) clinical profile and presentation of infants who were readmitted for acute kernicterus and those in whom acute kernicterus was not recognized but developed post-icteric sequelae, (d) assessment of lapses in care provided prior to specific medical intervention.
Standardization of definitions for hyperbilirubinemia and kernicterus
Eligibility to the Kernicterus Registry. With the goal of establishing a national reporting system for kernicterus and to define eligibility criteria for admission to the Pilot USA Kernicterus Registry, it was imperative that a consensus was developed for the clinical diagnosis of both acute stage kernicterus (ABE) and the chronic post-icteric sequelae. Magnetic resonance images (MRIs), when available, confirm acute or chronic neuroanatomic abnormalities in the globus pallidus and subthalamic nucleus. In the absence of a confirmatory diagnosis of magnetic resonance imaging, a consensus-based definition of pathognomic clinical and auditory abnormalities was needed to define ABE and post-icteric sequelae. We convened a workshop with Dr Michael Painter (University of Pittsburgh) to develop definitions for bilirubin-induced neurological dysfunction (BIND). These definitions were presented and published following the National Institute of Child Health and Human Development (NICHD) meeting.7, 8 The medical literature provides recognized descriptions and diagnostic criteria for histopathological signs of kernicterus and extensively details the variety of its clinical signs and manifestations. These have generally been based on literature from the 1950s and 1960s when kernicterus was usually the sequelae of severe hemolytic disorders. Volpe4 and Johnson et al.5, 15 have provided more recent updates. The epidemiological limitations are that clinical signs of early and ABE are often insidious and non-specific, seemingly benign and frequently missed by clinicians. Suggested risk factors have included prematurity, rate of total serum bilirubin (TSB) rise, age at peak TSB, rate of reduction of TSB and duration of hyperbilirubinemia, as well as the mode and timeliness of intervention.11, 12 Therefore, we charted individual neurological outcomes according to standardized clinical definitions and correlated these to specific risk factors and co-morbidities such as gestational age, severity of hyperbilirubinemia, rate of bilirubin load reductions, and sepsis, hemolysis, dehydration, birth trauma, glucose 6-phosphate-dehydrogenase (G6PD) deficiency and idiopathic causes of hyperbilirubinemia. The ‘idiopathic’ group most likely includes infants with unidentified hepatic conjugation and excretion defects as well as infants with undiagnosed or unidentified hemolysis and excessive bilirubin production. The initial template was standardized according to the definitions listed below.
Acute bilirubin encephalopathy. For infants 35 weeks GA (gestational age) with hyperbilirubinemia is described by the constellation of progressive signs of acute kernicterus (listed in Table 1). These data were tabulated retrospectively, using a medical chart review and parental reports. The progression of ABE was categorized as subtle, moderate or advanced.16 For inclusion in the Registry, infants without kernicteric sequelae were required to have documented ABE of at least moderate severity. Those with non-specific ABE were included only if they met the following criteria for post-icteric sequelae.
Post-icteric or kernicteric sequelae. These included extrapyramidal movement disorders (especially dystonia and athetosis), auditory disturbances (especially deafness or hearing loss secondary to auditory neuropathy/dys-synchrony), gaze abnormalities (especially impairment of upward gaze), intellectual impairment (rarely in mentally retarded range) and post-icteric enamel dysplasia of deciduous teeth. Impairment in either neuro-motor or auditory impairment (severe enough to necessitate ongoing specialized therapy or skilled assistance) along with an additional two of the five domains defined post-icteric sequelae and inclusion to the Registry. Post-icteric sequelae were categorized as mild, moderate or severe.16
Definitions used for severity of hyperbilirubinemia. Neonatal hyperbilirubinemia is often referred to as a ‘high’ level of TSB. However, the clinical meaning of this designation must be recognized as a function of the postnatal age-in-hours, the concentration of bilirubin in a percentile-defined track. From this perspective, hyperbilirubinemia was defined as the need for an acute medical intervention. Transcutaneous (TcB) or TSB values charted in medical records were assigned to the known or, if necessary, the closest and plausible postnatal age-in-hours and then plotted on a computerized hour-specific bilirubin nomogram.24 The severity of hyperbilirubinemia was determined by the following indices: (a) >75th percentile TSB for age-in-hours, (b) >95th percentile TSB for age-in-hours, (c) >99th percentile for age-in-hours, (d) TSB level of either >20 or >25 mg per 100 ml, which have been traditional clinical thresholds for intervention.24, 25 TSB levels >20 mg per 100 ml before 60 h of age and >25 mg per 100 ml beyond 72 h of age are consistent with the 99th percentile before 60 h of age and the 98th percentile beyond 72 h of age on the hour-specific bilirubin nomogram. TSB level 25 mg per 100 ml is a recent clinical threshold for aggressive intervention. This value represents the 99.99th percentile before 60 h of age and is the 99th percentile beyond 72 h of age on the hour-specific bilirubin nomogram. (e) Use of phototherapy and/or an exchange transfusion and their initiation in accordance with recommended guidelines has also been used as indices for severity of hyperbilirubinemia. (f) Rate of TSB rise (mg per 100 ml per hour) during the first 96 h after birth, in a healthy newborn, usually increases in approximate linear manner.24, 25 When the pre-discharge TSB level was not measured or available, we calculated the projected rate of TSB rise using a cord TSB of 2 mg per 100 ml (95th percentile for healthy newborns) to a peak TSB measurement (usually at readmission). A non-linear or an erratic increase in TSB levels was considered in an infant with known or proven G6PD deficiency or sepsis that may have sustained an acute unpredictable hemolytic crisis, such as ‘favism.’
Statewide distribution. Each case in the Registry was designated by the city and state of birthing. The total number of cases voluntarily reported from each state was collated for information purposes only. These data most likely represent bias due to location of professional colleagues, sources for medico-legal referrals and concern of specific physicians for litigation. Data analysis will be limited to general impressions. Absence of cases reported from a state carries no significant import.
Occurrence by year of birth. Each case in the Registry was collated by the year of birth. The total number of cases reported for each calendar year was collated for informational purposes. These data were evaluated for any correlation with critical changes in healthcare practices, healthcare policies or prevalent provider educational agendas. Coincidental trends to increase or decrease in relation to the reported occurrences were studied.
Etiological factors. The clinical diagnosis for the major contributor of the hyperbilirubinemia and reasons for the diagnosis, as listed in the medical charts at the time of discharge following the sentinel event, were collated to assess the prevailing clinical perspective. Presence of concurrent morbidities, such as hemolysis, prematurity, hypoalbuminemia, asphyxia/blood–brain barrier disruption, infection, hypoglycemia and drugs that may trigger a ‘favism’ reaction in an infant with G6PD deficiency, were documented.
Definition of late preterm. Late prematurity is defined as a gestation of 34 0/7 to 36 6/7 weeks based on pregnancy dating (last menstrual period and prenatal ultrasound, when available) and was corroborated by clinical examination. For the purpose of this report, infants <35 weeks gestation were excluded and in this report the terminology ‘late preterm’ specifically refers to infants 35 and <37 weeks GA.
Standardization of definitions for major contributors of hyperbilirubinemia. (a) Hemolytic disorders: Hemolysis was defined by presence of anemia (hematocrit 40% within 2 weeks of age), higher than normal reticulocyte counts for postnatal age, a peripheral smear suggestive of Hemolysis, such as spherocytes, schistocytes, RBC fragments, and corroborating signs such as positive direct antiglobulin test (DAT, Coombs's). The clinical use of exhaled carbon monoxide to assess hemolysis was not available for any of the Registry cases. In view of the retrospective nature of data analysis, the documentation was limited by the extent of bedside diagnostic investigation. Using existing clinical criteria, infants with excessive bilirubin production could not be recognized or categorized in the absence of hemolytic disease or extra-vascular hemolysis. Thus, these infants were considered ‘idiopathic’. To better understand this group of infants, we estimated their rate of TSB rise during the first 96 h after birth.
(b) G6PD deficiency: The clinical and laboratory diagnosis of deficiency required a quantitative kinetic G6PD assay interpreted for age and gender and carried out in a reputable laboratory, done on the infant's pre-exchange transfusion blood. The diagnosis was confirmed, or sometimes first made by assay of blood collected after the age of 3 months. DNA analysis was sometimes available but not required.
(c) Birth trauma: Excessive trauma was defined as presence of extensive bruising, subgaleal hemorrhage, presence and size of a cephalohematoma or other concealed or contained hemorrhage, a fractured clavicle or other trauma related to shoulder dystocia.
(d) Idiopathic disorders: The definition of idiopathic hyperbilirubinemia required apparent absence of increased hemolysis, G6PD deficiency and excessive birth trauma after a detailed chart review. In some cases assigned to this category, studies had excluded the possibility of Gilbert's syndrome. A family history of jaundice in the absence of known isoimmune disease was supportive of the diagnosis of idiopathic hyperbilirubinemia and suggested the presence of as yet unidentified familial hepatic conjugation or excretion compromise.
(e) Sepsis: It was defined as culture-proven sepsis or presumed (culture negative) sepsis based on laboratory studies and clinical risk factors and treated with a full course of antibiotics. Sepsis was considered a highly significant co-morbidity and was designated as a secondary contributing cause of hyperbilirubinemia. These instances are highlighted in the clinical vignettes (Appendix 1). For analysis purposes and tabulation of results, however, it was not considered as a primary diagnostic category.
(f) Co-morbidities: These were identified and defined as shortened gestation (35 to <37 weeks GA), being born to a diabetic mother, dehydration (based on a birth-weight loss of 15% and/or a serum sodium of >150 mEq per l), hypoalbuminemia (<3.0 g per 100 ml) if albumin had been measured (n=16 of 125), polycythemia (when documented), poor feeding and hypothyroidism (abnormal metabolic screen).
(g) Non-preventable causes: Each case was individually assessed by one or more co-authors, including at least one physician, followed by software evaluation to identify preventable steps that could have been taken with reasonable ease and without undue cost to avert the occurrence of kernicterus.
(h) Severity of hyperbilirubinemia exposure: Based on (as discussed above) (i) peak bilirubin level, (ii) duration of hyperbilirubinemia, (iii) TSB level >20 and >25 mg per 100 ml or (iv) >4 weeks duration.
Assessment of lapses in care
Using the patient safety matrix recommended by the Institute of Medicine27 and Joint Commission Accrediting Hospital Organizations,28 we used both qualitative and quantitative data to determine the lapses in care in the context of patient centeredness, patient safety, effectiveness and timeliness of care. For example, we assessed the following. (i) Patient centeredness: Analysis for patient-centered care was assessed for: (a) experiences of newborn services at the birthing hospital, interval home care, outpatient follow-up, emergency room management, hospital readmission, hospital to hospital transfer or emergency room to another facility transfer, site for targeted and presumed therapeutic intervention and follow-up services; (b) patient–provider relationships at each of these sites were ascertained based on communications charted in medical records, depositions (if a medico-legal case) and parent interview; and (c) value of the services to the family was evaluated on the continued care by the attending physician, referral for developmental follow-up, accuracy of medical records, medico-legal consequences and feedback from parental interviews. (ii) Family experiences: Parental experiences were ascertained by their comments to measure how they perceived the management of jaundice, the attitude, the language used, the message received, the concern, the recognition of emergency, and the healthcare provider realization that the outcome was not desirable. (iii) Relationships with healthcare providers: Parental perspectives as to the relationships they established with medical staff and the nursing staff during the birthing experiences, post-discharge care, emergency care and follow-up care and continuity of these relationships were obtained.
Results and outcome
176 case reports to the Registry by August 2004, 125 infants met the inclusion and eligibility criteria for the term and near-term Registry. Infants were excluded (n=23) because of their gestational age being <35 weeks or because of concurrent complex neonatal conditions that were managed in neonatal intensive care facilities; 28 infants were additionally excluded because they did not meet the eligibility criteria for diagnosis. Eligible infants had a follow-up for up to 18 months other than those who died (n=6) or were lost to follow-up before age 3 months (n=1). Cases reported by colleagues and by families were assessed for their calendar year of birth (Figure 1) and location of birthing site. No extrapolation of these reports to a state or community incidence of kernicterus can or should be made as the reporting was done on a voluntary basis by pediatricians (usually neonatologists) and as the clinical diagnosis was made over several years subsequent to birth. Brief clinical vignettes are provided for each case in a de-identified tabular list categorized by the most likely cause for excessive hyperbilirubinemia and ranked by ascending peak TSB levels (Appendix 1). These vignettes represent our most updated information and thus may occasionally differ from an earlier reported preliminary analysis. An overview of the characteristics: gestational age, outcome and mode of treatments for all 125 infants, is tabulated in Table 2. The birth weight (compared for term and late-preterm infants) and gestational age of these infants are shown in Figures 2 and 3. Data relating to occurrence of kernicterus in the late preterm have been published earlier.29 The subsequent tables and figures provide a perspective of the clinical data from this convenience sample.
Mortality attributed to kernicterus
Of the 125 infants in the Registry, overall six infants died during their first year. First week mortality was 5 of 125 (4%) with hazardous hyperbilirubinemia. Mortality included two infants who died earlier to exchange transfusion and one who died before any treatment could be initiated. Additional risk factors were late prematurity, G6PD deficiency and sepsis. Four of the five deaths were in infants of <37 weeks GA. G6PD deficiency appeared to be a confounding risk factor for both mortality and morbidity. Of the 26 infants with a confirmed enzyme deficiency, 4 (15%) infants died. Concurrent sepsis was noted in 8 of 26 (31%), two of whom died.
Morbidities associated with outcome of kernicterus
Individual case vignettes (Appendix 1) detail the concurrent impact of sepsis: either proven or presumed in infants with negative blood cultures. Genetic disorders were confirmed in three infants with Crigler–Najjar syndrome (n=2) and galactosemia (n=1). These three infants were excluded from further analysis because their clinical profiles were considered as usually representing non-preventable causes of icteric sequelae.
Presumption of sepsis was based on serial white blood cell and differential counts and treatment with antibiotics. Of those tested and confirmed to have G6PD deficiency (n=26), six had blood culture proven sepsis and in two infants sepsis was presumed, including one with probable viral sepsis. In the remainder 99 infants (in whom G6PD status was often not investigated), six were proven to have sepsis (positive blood culture) and four had presumed sepsis (including two with probable viral illness).
Clinical diagnoses for severe hyperbilirubinemia
There were multiple co-morbidities that were associated with excessive hyperbilirubinemia (Appendix 1), and the severity of clinical sequelae was regardless of its contributory clinical cause. Dehydration at readmission was defined as 15% weight below birthweight and/or Na 150 mEq l−1, such as 22% weight loss in one infant, and hypernatremia, 166 mEq l−1 in another infant. The data in Table 3 represent the relative frequencies of diagnostic criteria related to readmission age. Most babies were admitted to the hospital for excessive jaundice, but were often not recognized or diagnosed as having ABE (Table 4).
Infants readmitted on days 2.5 to 3.5 (<84 h of age) are infants with acute early-onset hyperbilirubinemia (Table 3). The highest TSB reported in these 13 infants was 52 mg per 100 ml in a baby girl, with an older sibling who had jaundice treated with phototherapy, and who had no blood-group incompatibility, evidence of hemolysis, excessive birth trauma or infection and was not jaundiced in later infancy or childhood. The total number of infants in whom no specific diagnosis was established, 9 of 13 (69%), was categorized as having an idiopathic cause (Appendix 1 for clinical vignettes). These nine infants, six of whom were girls, represent an unusual group of infants who had excessively rapid rates of TSB rise associated with the following: (a) probable increased bilirubin production due to ABO incompatibility with a negative direct antiglobulin test, a hematocrit >50% and reticulocytosis at the upper limit of normal for postnatal age, plethora and visible bruising; (b) severe dehydration and sepsis; (c) possible unidentified genetic polymorphisms or coding errors and delayed bilirubin elimination based on a strong family history of Gilbert's syndrome or unexplained jaundice.
Infants not readmitted
These infants represent those with unrecognized ABE and presumed excessive hyperbilirubinemia (7 of 125, 4.8%) who subsequently showed moderate or severe post-icteric sequelae. Four of these seven babies who were not readmitted had severe sequelae: two had idiopathic jaundice, one had jaundice associated with birth trauma and one was associated with hemolysis. The remaining three infants never readmitted had moderately severe sequelae.
Pre-discharge risk assessment
Most infants were discharged at <48 h of age, but about 28% (34 of 121) were discharged beyond 48 h of age. Frequency of jaundice was identified and recorded in the medical records prior to discharge in 70 of 125 infants: ‘no jaundice’ in 11 infants and ‘jaundice present’ in 52 infants (in 11 of these jaundice was evident prior to age 30 h). There was no record of jaundice in 55 of 125 infant medical records. Pre-discharge TcB or TSB levels were measured infrequently and were done in only 22 of 125 infants. The readmission ages of these 22 infants ranged between day 3 and day 11. Severity of hyperbilirubinemia was unrecognized in all 22 infants because their TSB levels were not interpreted on the basis of postnatal age-in-hours or even age-in-days, and earlier or targeted follow-up was not considered, even though there was significant hyperbilirubinemia.
Post-birthing hospital follow-up visit
First contact with practitioner
Data after discharge from the birthing hospital were often inadequate, and responses to concerns raised by parents whose infants were in early stages of ABE were inappropriate. The ‘first-contact practitioners’ were: (a) an office triage nurse, (b) a nurse (random) at the birthing hospital, (c) a home-visiting nurse, (d) a practicing physician (often not the discharging physician), (d) an emergency room physician and (e) a lactation consultant. The first contact was most likely an office visit (38%), emergency room (29%) or call to a physician's office (24%). Approximately 63% had made one contact, but 37% made between 2 to 5 contacts. A review of medical records indicated that most nurses, pediatricians and neonatologists were unfamiliar with the clinical signs of ABE and the progressive clinical manifestations.
Only two infants were exclusively formula fed. Lactation failure was identified in over 90% of infants discharged on exclusive breast-feeding. There was uniformly sub-optimal lactation support (both at the birthing hospital and at follow-up). There was a high incidence of excessive weight loss and dehydration, and in some supplements were recommended. These infants also had confounding influences of other co-morbidities, such as prematurity, bruising, lethargy, possible exposure to maternal medications or hemolytic triggers, and first-time experience for breast-feeding.
Inappropriate parental reassurance
Commonly used reassurances given to the babies who developed kernicterus were: ‘jaundice does not cause brain damage in healthy babies’, ‘kinder and gentler approach to jaundice’, ‘do not worry, all babies get jaundice’, ‘you can easily get rid of jaundice with sunshine’, ‘bilirubin is an anti-oxidant’, ‘it is like a rash’, ‘it is like a cold’, ‘jaundice gets worse before it gets better’ and ‘put them in the window’.
Bilirubin and jaundice management at first follow-up visit
Most infants were not given an appointment for a follow-up visit within 3 to 5 days of discharge. Office-based follow-up appointments, within 48 h of discharge from the birthing hospital, were provided for only 29 of 125 (23%) infants. Measurements of TcB or TSB at the follow-up office visit were not done in almost half of the infants. At the office visit, eliciting history for clinical signs of ABE in the context of a jaundiced or symptomatic infant was not a practice. Response to excessive TSB levels (when measured) was slow or referrals were made to the emergency room. Among the 29 infants who returned for an office visit, follow-up TSB was measured within 48 h of discharge in 15 infants, and of these five were readmitted within 2 days of the follow-up visit, to be treated with ‘crash-cart’ interventions. In the remaining 14 infants, response to prolonged jaundice (beyond 7 days of age) was reassurance.
Peak TSB levels and diagnostic categories by age at readmission for infants with ABE
The likelihood of progressive hyperbilirubinemia could have been predicted based on: (a) severity of pre-discharge hyperbilirubinemia or excessive levels at first follow-up visit; (b) unrecognized hemolysis due to major or minor blood type incompatibility; (c) unrecognized G6PD deficiency, which may have been suspected either because of hyperbilirubinemia or because of racial/ethnic background; and (d) delayed bilirubin clearance complicated by suboptimal breast milk intake with diminished urine and stool output and unrecognized genetic bilirubin elimination disorders. Data are presented for peak TSB levels following admission and were reviewed in the context of the clinical diagnosis shown in Table 3.17, 25, 26 Among the infants who were not hospitalized (n=7) but subsequently showed moderate to severe post-icteric sequelae (Appendix 1), none was treated for prolonged (>2 weeks) jaundice in spite of repeated parent phone calls voicing concerns. The TSB level that was measured in only one of these infants at age 4 days was 20 mg per 100 ml, and no subsequent measurements were obtained for persistent jaundice. Diagnostic categories in the seven infants included hemolysis in one, birth trauma in one and idiopathic in five.
Rate of total serum bilirubin rise
We determined the rate of TSB rise for ages 24 to 72 h from a cord level of 2 mg per 100 ml and the peak TSB measured (almost always at readmission) (Table 5). This rate of TSB rise was corroborated in 22 of 118 infants in whom an hour-specific pre-discharge TSB/TcB had been measured. In 18 of 22 infants, the graphic line for the projected bilirubin rate of rise from a cord TSB level of 2 mg per 100 ml to the peak TSB level approximated or, was below, the measured hour-specific pre-discharge bilirubin level plotted on the hour-specific bilirubin nomogram. The projection over-predicted the pre-discharge bilirubin in 4 of 22 infants: two infants with G6PD deficiency (one with a known exposure to moth balls and the other in whom pre-discharge TSB was >99th percentile), one infant with nosocomial sepsis acquired 4 days after hospital readmission on day 7 and another infant with severe birth trauma. The rate of TSB rise in infants discharged at age <24 h, 24 to 47 h and <72 h was determined in a select cohort of 58 of 125 infants with kernicterus who were discharged at age <72 h and readmitted at age 5 days (10 of these infants had a pre-discharge bilirubin value). We also calculated what would have been the hour-specific TSB level at discharge for each infant. Table 5 shows TSB rates of rise stratified for 0.20 to <0.30 mg per 100 ml per hour and 0.30 mg per 100 ml per hour. For infants discharged at <48 h age, 29 of 39 (72%) would have had TSB levels that meet the 2004 AAP thresholds for intensive phototherapy. Out of 58 infants discharged before age 72 h, 43 (75%) had projected rates of TSB rise 0.30 mg per 100 ml per hour and met the criteria for exchange transfusion at discharge. All 58 infants would have showed pre-discharge TSB levels >75th percentile for age in hours at the time of discharge.
Outcome of infants with peak TSB levels >35 mg per 100 ml
All infants with peak TSB levels >35 mg per 100 ml who were readmitted (n=73) had post-icteric sequelae. Of these, four died soon after readmission; 61 (83.6%) had severe sequelae, six (8.2%) had moderate sequelae and two (2.7%) had mild sequelae (including one term male infant admitted with subtle signs of ABE and treated only with phototherapy for a peak TSB level of 41 mg per 100 ml). Most infants who presented with moderate (n=8) or advanced (n=55) ABE had severe sequelae (55 of 63; 87%, including three infants who presented with moderate ABE). Severe sequelae were also noted in three of the four infants who were readmitted with subtle signs of ABE. Phototherapy was the only treatment for four infants with TSB levels >35 mg per 100 ml. Of these, two infants (50%) died at readmission and postnatal age <8 days as compared with 2 of 69 (3%) infants who died after subsequent phototherapy and exchange transfusion.
Outcome of infants with TSB 35 mg per 100 ml
Two infants died: one before any treatment could be initiated and the other treated with brief phototherapy. One infant was lost to follow-up before age 3 months. The remaining infants (n=43) were treated with phototherapy alone (n=20) or exchange transfusion and phototherapy (n=23). Interventions in 8 of these 23 infants was characterized by: (i) a pattern of immediate institution of phototherapy with multiple lights to expose as much of the body surface as feasible; (ii) performance of procedures while the infant was under phototherapy; and (iii) urgent preparation for exchange transfusion, implemented as soon as possible or indicated by clinical signs. As shown in Table 6 and Figures 4 and 5, we noted that the interval between admission and institution of an exchange transfusion was <6 h in seven of eight infants who survived without sequelae. On the basis of these data, we defined this pattern of intervention strategy as a ‘crash-cart’ approach. In the 15 of 23 infants treated with both phototherapy and exchange transfusion who sustained sequelae, treatment of three met our definition for ‘crash-cart’ treatment. In these three infants and considering all 23 infants treated with phototherapy and exchange transfusion, there was overlap in length of time that elapsed from parents’ call for help and admission to a treatment facility, between infants with and without sequelae. However, this overlap tended to be shorter in the eight babies who survived without sequelae (Table 6). Thus, the outcome of eight infants (73%) who did not have excessive pre-admission delays out of a total of 11 infants managed by a ‘crash-cart’ approach had no sequelae. In 12 of the 23 infants treated with exchange transfusion, ‘crash-cart’ intervention was not feasible because of unavoidable circumstances such as need for transport to a tertiary care facility or technical issues such as resuscitation, requirement for the interosseous route for stabilization and parenteral correction of severe dehydration or surgical venous access to accomplish exchange transfusion. As shown in Table 6 reversal was more likely in infants readmitted by age 5 days.
Treatment of ABE with phototherapy alone (20 of 43) was uniformly dismal. Five of the 20 babies treated with only phototherapy were readmitted by age 4 days, one on day 2.5 (TSB of 22.4 mg per 100 ml) and four on day 4 (TSB ranged from 28.3 to 34.5 mg per 100 ml). These infants had acute early-onset hyperbilirubinemia and were exposed to rapid rates of TSB rise. The readmission age of eight of the 20 babies treated with only phototherapy was after day 9 (TSB ranged from 20.7 to 29.3 mg per 100 ml including one baby with congenital spherocytosis admitted on day 27). These eight infants had prolonged duration of jaundice and presumed significant hyperbilirubinemia that was treated with phototherapy of unspecified intensity.
On the basis of this data, we believe that ‘crash-cart’ management after hospital admission has the potential to prevent sequelae in some babies even if advanced signs of ABE are present. However, emergency post-admission management cannot reverse all acute-stage damage. Our review and analysis of the Registry cases lead us to recommend that infants whose TSB levels are close to exchange transfusion thresholds17 should be screened with an automated auditory brainstem response and treated using a ‘crash-cart’ approach.
Infants with peak TSB <25 mg per 100 ml who developed kernicterus
Six of the 118 infants with peak measured TSB level <25 mg per 100 ml developed chronic sequelae. One infant was treated with both exchange transfusion and phototherapy after being readmitted on day 6 following a traumatic home birth and subsequent development of E. coli sepsis (see Appendix 1). In the other five infants who were treated with phototherapy alone, we explored the probable antecedent clinical factors: (a) Early postnatal age. One infant was readmitted at age 2.5 days with TSB of 22.4 mg per 100 ml. Concurrent co-morbidity included severe dehydration in the absence of hemolysis and sepsis. The cause of hyperbilirubinemia was not identified and listed as idiopathic. For this infant, increased risk for bilirubin toxicity was most likely because of decreased binding affinity of albumin for bilirubin during the first 3 postnatal days and a rapid rate of TSB rise. (b) Prolonged exposure to severe hyperbilirubinemia (>2 weeks). Three infants with TSB range from 20.7 to 24.9 mg per 100 ml had uncomplicated delivery and newborn courses. Their significant jaundice and hyperbilirubinemia persisted for over 4 weeks in one and over 2 weeks in the other two. (c) Sepsis. One infant, at term gestation, developed sepsis and positive blood culture for clostridium perfrigens and enterococcus fecalis. He was admitted with severe dehydration (Na 160 mEq/l) on day 11.5. This infant was treated with a ten-day course of antibiotics, 2 days of intravenous fluids and three days of phototherapy. These five infants highlight the confounding and deleterious contributions of sepsis, hypernatremic dehydration and slow post-admission reduction of the body bilirubin load with sub-optimal phototherapy.
Association of neonatal apnea as a clinical manifestation of ABE
Case records that provided detailed records of apnea, bradycardia, desaturations, periodic breathing as well as notations for presence or absence of apnea by either the neonatal nurses or physicians were available and reviewed in 108 infants. Occurrence of symptomatic apneic events, noted on readmission for ABE and for the ensuing 72 h, was ascertained and are listed in Table 7. These data suggest that apnea associated with severe hyperbilirubinemia could be a manifestation of ABE and may be confounded by gestational immaturity as well as gender.
Absence of definitive signs of acute kernicterus
Using the BIND score, we identified nine infants with clinical signs for ABE and scores <4. These infants were labeled to have subtle or non-specific ABE. None had an auditory brainstem response (ABR) testing during their acute phase of hyperbilirubinemia. Clinical profile and the post-icteric sequelae are shown in Table 8. Severe sequelae were noted in 4 of 9 infants. In contrast, among 91 of 116 infants with advanced ABE (BIND scores >6), nine infants subsequently had no (n=3), mild (n=1) and moderate (n=5) post-icteric sequelae. The three infants with no icteric sequelae had ‘crash-cart’ management at hospital admission. These observations are consistent with earlier reports1, 2 of 15 to 16% of babies with Rh disease who developed kernicteric sequelae and were not recognized as having definite neurological signs of acute kernicterus in the first week after birth. Retrospective use of the BIND score based on hospital records and reported parental concerns was helpful to document and monitor the progression of ABE (as shown in Table 1). However, its usefulness, in conjunction with serial ABR measurements, as a guide for treatment, as well as its prognostic value considered in association with the timeliness and efficacy of treatment, needs to be validated prospectively.
Identified ‘breakdowns’ in the health services for newborn jaundice management
All cases were assessed to have multiple and specific root causes or lapses in the care provided by multiple providers at multiple sites. Delays in interventions were often related to a pervasive lack of awareness of the impending irreversible neurotoxicity. The most common root causes for failures of health systems (Table 9), based on the Institute of Medicine patient safety matrix, were similar to those provided as preliminary data in an earlier report.16
The antecedent clinical and health services events experienced by infants reported to the Pilot USA Kernicterus Registry inform gaps in the clinical and public health management of newborn jaundice that may not be evident by traditional epidemiological investigation or surveillance. These include the following:
Kernicterus cases are continuing to be reported in the United States.
Over 95% of the cases were attributed to a multi-factorial failure of the post-partum and newborn healthcare delivery system.
National incidences of kernicterus have been defined in reports from Denmark, the United Kingdom and Canada. Incidences of severe neonatal hyperbilirubinemia have also been reported from Sao Paulo and Jerusalem.31, 32, 33, 34, 35, 36 The Pilot USA Kernicterus Registry cases do not provide any basis to define a national or state incidence of kernicterus in United States. These cases represent the minimum number of infants diagnosed with kernicterus and are likely to be a ‘tip-of-the-iceberg’. A review of these data and the current literature support the role of a systems approach to decrease the incidence of extreme hyperbilirubinemia.37, 38
Virtually all cases of kernicterus in the Registry could have been prevented with early identification of potentially severe hyperbilirubinemia (TSB levels above the 40th percentile), recognition (before discharge or at early follow-up) of TSB increase at >0.2 mg per 100 ml per hour and institution of timely medical care, intensive phototherapy and exchange transfusion as needed. On the basis of this data and the availability of current treatment interventions, we conclude that kernicterus is almost always an unacceptable outcome in healthy >35 weeks gestation infants in the United States. This opinion is supported by recent reports in the peer reviewed literature.37, 38, 39, 40
Kernicterus is a low-frequency disease with current interventions, but an unacceptable outcome in prevailing health practices.39 It may be characterized as an iatrogenic event when the management of severe neonatal hyperbilirubinemia is not effective or timely.
For nearly all cases, responses to parental and family concerns and involvement were inadequate.40 In most cases, parents were not aware, sometimes for months subsequent to the event, that bilirubin neurotoxicity was the cause of their child's developmental delay and abnormalities.41
Data from this Registry validate the consensus opinion that there is no evidence of a specific bilirubin level linked to the onset of ABE. As reported by Newman et al.,42 when treated with phototherapy or exchange transfusion, total serum bilirubin levels in the range between 25 and 29.9 mg per 100 ml were not associated with adverse neuro-developmental outcomes in infants born at or near term. However, this retrospective study does not report concomitant signs of ABE or timeliness of intervention. There are no new clinical data on which to base a new danger TSB level.40, 42, 43, 44 Data from the Pilot USA Kernicterus Registry note that infants with any signs of ABE and peak TSB levels >35 mg per 100 ml sustain post-icteric sequelae and underscore the need to better assess clinical signs of acute bilirubin neurotoxicity. In the absence of alternate strategies to predict the risk of neurological injury, the narrow margin of safety between severe hyperbilirubinemia and acute onset of progressive encephalopathic changes, the clinical ability to predict and manage severe hyperbilirubinemia, both medical and public health agenda should focus on prevention.
Successful implementation of a systems approach to manage newborn jaundice could serve as an index for the integrity of the postpartum health delivery system. On the basis of family expectations in the United States, birthing safety standards need to be transparent and impeccable, and aspire to the highest feasible standards with the least adverse encounters.
National guidelines for US clinicians have been in place since 1994. In effective implementation of the 1994 AAP guidelines has been suggested as the basis for the continuing reports of kernicterus. Effective implementation of the newly revised 2004 AAP guidelines is a matter of urgent concern.
National awareness and partnership with new and expectant parents remains an urgent need.
Optimization of clinical practice
Preliminary reports from this Registry have informed the steps initiated by the AAP,45 the Center for Disease Control and Prevention,46 the Association of Women's Health, Obstetrics and Neonatal Nurses,47 and the Joint Commission Accrediting Hospital Organizations.29 These include standardized guidelines for management of neonatal jaundice and hyperbilirubinemia for pediatricians and emergency rooms.48 Most of the patient education materials (brochures, videos, resource links, posters and transcripts) are currently available online for public access and use at www.aap.org/jaundice; www.cdc.gov/jaundice, www.jcaho.org/kernicterus and www.pickonline.org. The California Perinatal Quality Care Collaborative has developed and implemented a ‘Severe Neonatal Hyperbilirubinemia Toolkit’ (http://www.cpqcc.org/quality_improvement.htm).49, 50 State-wide or regional initiatives to report severe neonatal hyperbilirubinemia (TSB levels >25 mg per 100 ml) along with outcomes for newborn screening for other inherited disorders may be feasible through a three-pronged partnership of providers and society to implement a state-driven national program: (1) apply an aviation safety standard to address gaps in newborn health delivery using healthcare-information-technology solutions; (2) implement state-wide (regional) and national surveillance for severe neonatal hyperbilirubinemia; (3) provide national and global education and empowerment aids for new and expectant parents.
Need for a national initiative
Following the initial report by Johnson et al.10 delineating the need for a systems approach to prevent kernicterus and the 2004 AAP guidelines,13 several international reports have provided further insight into the gaps that exist in health systems, including those with institutionalized home-based post-birthing follow-up.31, 32, 33, 34 Recent data from US hospitals and regional health systems37, 38, 48, 50, 51 have provided an insight into the impact of a vigilant healthcare system and its ability to change the incidence of severe neonatal hyperbilirubinemia and frequency of exchange transfusions. Effective implementation of both 1994 and 2004 AAP guidelines remains a challenge to the AAP and the Pediatrics community. The AAP has initiated an implementation plan.52 Strategies for implementation should include monitoring for its effectiveness. Currently, there is no formal reporting of kernicterus or severe hyperbilirubinemia in the United States, and the incidence of these indices of adverse outcomes is unknown. There is an urgent need to establish a formal mechanism to identify the success or failure of any national outreach program for both the professional and public communities.
Regardless of the cause for jaundice, the potential risk for unrecognized, unmonitored and untreated severe hyperbilirubinemia is a matter of newborn safety. As chronic kernicterus is preventable but not treatable, our focus needs to be rooted in a preventive approach. We need to educate clinicians and society about the warning signs of bilirubin neurotoxicity because early and intermediate stages of ABE may be reversible with prompt and effective bilirubin reduction strategies, and thereby ensure access to effective and timely interventions. Future evidence of the adverse effects of either under-treatment or over-treatment of hyperbilirubinemia should continue to impact clinical practice. In progress effort by the AAP to monitor and facilitate the implementation of the guidelines offers a promise for effective clinician implementation of the guidelines.52 As practicing clinicians at the front-line of health delivery who have to deal with the realities of a non-existent seamless transfer of clinical information during the transition after birth, continued vigilance is necessary. As we educate and empower expectant parents and clinical providers, three salient perspectives based on this research need to be considered. First, the focus of the key messages should remain evidence-based and transparent. Second, the timeline for outreach should be coordinated and consistent with the prevailing infrastructure of healthcare system. Third, the educational format should be that of learning and empowerment and, when necessary, advocacy. As we balance evidence-based medicine with patient safety, let us be truly prudent and protective of all newborns entrusted to the care of health professionals. With parents as equal partners in this endeavor, we can bridge these gaps in the safety net of US newborn health services.
Always, there is a need for continued vigilance!
This work was supported in part by AAMC/CDC PERT Grant MM-0448. This research was also supported in part by funds from the Sandy Eglin Fund and her generous support of the Pilot USA Kernicterus Registry at Pennsylvania Hospital, Philadelphia. We remember the late Audrey K Brown, MD, and value her immeasurable contributions to the initiation and maintenance of the Pilot USA Kernicterus Registry. We thank the parents logged on the newborn jaundice list-serve as well as our colleagues who contributed their experiences to the Registry. We also appreciate the administrative support of Donna Spitz (Philadelphia) and Stella Dina-Gengania (Stanford).
About this article
The authors have declared no financial interests.