CONTEXT AND OBJECTIVE. Recent reports have raised global concerns about a reemergence of kernicterus. Accurate information on the incidence of kernicterus is unavailable because of the rarity of the condition and the lack of a systematic surveillance strategy. We used nationally representative hospital discharge data to evaluate trends in the diagnosis and management of neonatal jaundice and the incidence of kernicterus in relation to the American Academy of Pediatrics hyperbilirubinemia clinical practice guideline.
PATIENTS AND METHODS. The data came from the Healthcare Cost and Utilization Project family of databases. The Nationwide Inpatient Sample and the Kids' Inpatient Database were combined to generate trend data for the years 1988–2005. All neonatal discharges with primary or secondary International Classification of Diseases, Ninth Revision, Clinical Modification diagnoses codes for jaundice or kernicterus occurring within the first 30 days of life were selected with population incidence rates calculated from estimates of term and preterm newborn hospitalizations derived from the Healthcare Cost and Utilization Project data. To increase the reliability of identified kernicterus hospitalizations, newborns with a diagnosis of kernicterus and a procedure code for phototherapy or exchange transfusion were included as cases.
RESULTS. Hospital diagnosis codes for kernicterus likely included a substantial number of rule-out cases, because ∼70% did not include a procedure code for phototherapy or exchange transfusion. Including only cases with a procedure code for phototherapy or exchange transfusion resulted in 2.7 per 100000 diagnosed with kernicterus over the entire study period. A diagnosis code for jaundice was recorded for 15.6% of newborns. The diagnosis of jaundice and kernicterus differed according to race and gender. Rates also were elevated in preterm relative to term infants for both jaundice and kernicterus. Trends in diagnosis for newborn jaundice were u-shaped, with rates falling in the years before the initial American Academy of Pediatrics guideline (1988–1993) and increasing in the years after publication of the guideline (1997–2005). In contrast, the number of newborn hospitalizations with a diagnosis of kernicterus generally declined throughout the study period. Most of the decline in hospitalizations for term infants with a diagnosis of kernicterus occurred before and immediately after publication of the 1994 guideline, going from 5.1 per 100000 in 1988 to 1.5 per 100000 in the years from 1994 to 1996 and has since remained constant.
CONCLUSIONS. Nationally representative hospital data indicate a declining incidence of hospitalizations with a diagnosis of kernicterus in newborn infants over the period 1988–2005. The decline occurred before and immediately after publication of the 1994 American Academy of Pediatrics guideline on hyperbilirubinemia. Epidemiologic findings were mostly consistent with other studies. Healthcare Cost and Utilization Project data provide an important system for monitoring hospitalizations of uncommon newborn conditions such as kernicterus.
Kernicterus is the chronic form of bilirubin encephalopathy and is associated with significant mortality and morbidity. Deficits in surviving children are typically characterized by choreoathetoid cerebral palsy, paralysis of upward gaze, sensorineural hearing loss, dental enamel dysplasia, and intellectual handicaps.1–3 Hyperbilirubinemia, with total serum bilirubin values >30 mg/dL (513 μmol/L) in seemingly healthy term and near-term infants, increases the risk for developing kernicterus.4 A systems-based approach to the management of jaundice in newborns could largely prevent this permanent and devastating condition in otherwise healthy infants.5
To improve the systematic management of jaundice in newborn infants, the American Academy of Pediatrics (AAP) Subcommittee on Hyperbilirubinemia issued a guideline for the identification and management of hyperbilirubinemia in newborns in 1994.6 This guideline addressed the management of healthy term (≥38 weeks) newborns with jaundice. An update of the 1994 guideline was issued in 20043 that expanded the population of newborns covered by the guideline to include healthy and at-risk newborns (defined as having either isoimmune hemolytic disease, G6PD deficiency, asphyxia, significant lethargy, temperature instability, sepsis, acidosis, or albumin <3.0g/dL) of 35 or more weeks' gestation.
Despite the development of guidelines for managing hyperbilirubinemia, recent reports continue to raise global concerns about a reemergence of kernicterus.7–9 However, accurate information on the incidence of kernicterus is unavailable because of the rarity of the condition and the lack of a systematic surveillance strategy.4 It is unknown whether the 1994 guideline was successful in improving the management of jaundice and the prevention of kernicterus. Only 1 study examined changes in the management of jaundice in relation to the 1994 AAP guideline and found that both phototherapy and exchange transfusion rates declined in 2 community hospitals in Jerusalem.10
The 1994 guideline and the subsequent 2004 revision raised a number of concerns. The 1994 guideline embraced a “kinder and gentler” approach to the management of jaundice in neonates.11 The “kinder and gentler” approach allowed the use of phototherapy alone at bilirubin levels that previously would have indicated the need for an exchange transfusion.6 Even before publication of the 1994 guideline, some medical professionals were concerned that such an approach would lead to an increase in kernicterus.12 Subsequent case reports suggesting a reemergence of kernicterus raised questions about whether the “less aggressive” guideline was responsible or whether sampling issues resulted in a perceived increased incidence of kernicterus.13 Both guidelines recommend universal follow-up of infants within 72 hours if they were discharged between 1 to 3 days after birth. Such a recommendation has the potential to significantly increase cost with uncertain benefits given the lack of data on the incidence of kernicterus.14
Evidence on the effectiveness of the AAP guideline is lacking largely because of the inability to obtain reliable incidence data over time.4 Evidence describing the effectiveness of the guideline thus has the potential to provide important information for use in clinical practice. An important test of the effectiveness of the guideline is whether the clinical management of newborns has changed consistent with the intent of the guideline and whether changes in clinical management have led to an increase or decrease in the incidence of kernicterus. Unfortunately, there is no formal process for documenting or reporting kernicterus in the United States.15 A registry in operation from 1992 to 2002 reported 125 cases of kernicterus from 34 states and 2 US Army camps.16 A recent study from Canada based on voluntary reporting raised concerns of a high incidence of infants with severe hyperbilirubinemia.8 However, the reported incidence was lower than reports from previous studies17,18 and could not be construed as a resurgence of severe hyperbilirubinemia.19 Despite the lack of population-based reporting systems, the prevention of kernicterus remains a high priority among public health agencies.20–22
We used national hospital discharge data to determine whether (1) the management of newborn jaundice has changed over the 18-year study period and (2) the incidence of hospitalizations with a diagnosis of kernicterus has subsequently changed. Examination of these trends provides important information as to whether the 1994 guideline was effective in changing clinical practice in the best interests of patients and providers. The data also provide important information for use in considering clinical management changes and subsequent evaluations of the more recent 2004 guideline.
Data for this study come from 2 large nationally representative hospital discharge databases created by the Agency for Healthcare Research and Quality (AHRQ) Healthcare Costs and Utilization Project (HCUP): the Nationwide Inpatient Sample (NIS) and the Kids' Inpatient Database (KID). HCUP databases were developed through a federal–state–industry partnership and contain patient-level information compiled in a uniform format with privacy protections in place.23,24 HCUP includes the largest collection of longitudinal hospital discharge data in the United States, with all-payer, encounter-level information beginning in 1988. These databases enable research on a broad range of health care services and health policy issues at the national, state, and local market levels.
Both the NIS and the KID were sampled from the State Inpatient Databases of those states that participated in the HCUP for a given year. AHRQ has developed appropriately scaled discharge weights to generate national estimates of hospitalizations from both the KID and the NIS. With these weights, national estimates of hospitalizations and hospitalization rates are comparable across years despite the varying number of states participating in each year of the HCUP project.25–27
The NIS is designed to approximate a 20% stratified random sample of all US community hospitals (defined as short-term, nonfederal, general and specialty hospitals, excluding hospital units of other institutions) from states that contribute their State Inpatient Databases to the HCUP. Data elements in the NIS are constructed in a uniform format with quality checks in place. Each year NIS data are compared with data compiled by the American Hospital Association and the National Hospital Discharge Survey. The NIS includes 100% of discharges for all age groups and all payers from each sampled hospital. It contains data from ∼1000 hospitals and includes 7 to 8 million hospital discharges annually. NIS data are available from 1988 to 2005, allowing analysis of trends over time. However, fewer states participated before 1993 and could reduce the ability to capture rare conditions. In 1988, 8 states were included in the NIS, representing ∼30% of the US population. In 1993, 17 states were included in the NIS representing ∼50% of the US population. In 2005, 38 states were included representing ∼90% of the US population.
The KID differs from the NIS in 2 major ways: (1) the KID only includes data on pediatric patients aged ≤20 years; and (2) the KID includes an 80% sample of discharges from all community hospitals in participating states, except for uncomplicated newborn hospitalizations of which there is a 10% sample. It includes between 2500 and 3500 hospitals and 2 to 3 million discharges per year. The KID is produced every 3 years and included 22 states in 1997, 27 states in 2000, and 36 states in 2003.24 Because of its larger sample of pediatric patients, the KID is substituted for the NIS in the years where it is available. Identical statistical methodology is used to derive appropriate national estimates and standard errors for both the KID and the NIS. This method for constructing trend data from the HCUP has been used to study several newborn conditions.28–30
All neonatal discharges that had a primary or secondary International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code for jaundice (774, 774.0–774.3 774.30–774.31, 774.39, 774.5, 774.6, and 782.4) or kernicterus (773.4 and 774.7) were selected. Infants were classified as preterm birth if they were discharged with an ICD-9-CM diagnosis code in the range of 765.0 to 765.28. Because of the lack of unique identifiers in the NIS and KID, initial hospitalizations of infants which resulted in a transfer to another short-term, acute care facility (pretransfer hospitalizations) were excluded from study to prevent double-counting.31 The subsequent hospitalization record from the receiving hospital captured cases of jaundice and kernicterus with pretransfer hospitalizations. To limit double counting of infants sent home and subsequently readmitted during the neonatal period, age at admission was limited to a maximum of 30 days.
To increase the reliability of our case definition of kernicterus, cases were required to have a diagnosis code for kernicterus and a procedure code for phototherapy or exchange transfusion. This case definition can limit the number of cases with a “rule-out” diagnosis as the process for generating diagnosis codes in administrative data are subject to considerable variation and potential error through a variety of mechanisms.32 There are up to 15 diagnosis and procedure codes in the HCUP databases with some states abstracting 10 or fewer codes and some abstracting >15 codes.33 Diagnosis codes are first assigned by physicians based on available information. The knowledge and experience of the physician with the illness can be a significant factor in coding variation. The physician's narrative discharge summary is used by hospital coders to assign primary and secondary diagnosis codes and sequencing for billing purposes. Because kernicterus is rare and potentially difficult to diagnose, it is recommended in such situations to combine diagnosis codes with other measures.32 For kernicterus, we used evidence of treatment with phototherapy or exchange transfusion.
We also sought to reduce potential errors associated with polycythemia, because there is no separate ICD-9-CM procedure code for partial and double-volume exchange transfusions. From 1988 to 2005 there were 17991 weighted (4479 unweighted) exchange transfusions recorded for infants with a jaundice diagnosis. Of these, ∼40% of infants (7067 weighted, 1899 unweighted) had a comorbid diagnosis of polycythemia. Therefore, we did not include newborns with a diagnosis code for polycythemia and a procedure code for exchange transfusion in our rate calculations for exchange transfusions.
Overall rates of newborns identified with jaundice and kernicterus were calculated. Rates were then stratified by premature status, race/ethnicity, insurance status, income level, region of the country, and year of discharge. Newborn rates of phototherapy and exchange transfusion (excluding polycythemia) were calculated from ICD-9-CM procedure codes (99.83 and 99.01), and newborns with a diagnosis of polycythemia and exchange transfusion were excluded. Denominators for rate calculations involving newborn infants were obtained from the HCUP data.
Because kernicterus is such a rare disorder and the analysis of NIS data requires a minimum unweighted cell size of 70 cases, trend analyses were conducted using 3-year groups. All statistical analyses were performed by using Stata statistical software (Stata Corp, College Station, TX). Stata survey commands properly accounted for the weighted stratified sample designs of the NIS and the KID.
Table 1 provides data on characteristics of newborn infants hospitalized with a diagnosis of jaundice or kernicterus. Column 1 in Table 1 indicates that over the entire study period, 20.6% of neonates diagnosed with jaundice had a procedure code for phototherapy. Only a small percentage (0.2%) had a procedure code for exchange transfusion. For those children that did have an exchange transfusion, ∼50% also had a procedure code for phototherapy. The in-hospital mortality rate for these children was ∼0.5%.
Column 2 in Table 1 provides similar data for infants with a diagnosis code for kernicterus. This group represented our original case definition and resulted in 1395 (unweighted) hospitalizations; a number that is 3 times higher than the number of cases found using a case definition that requires evidence of treatment with phototherapy or exchange transfusion (column 4). The large number of cases in column 2 likely is because of the inclusion of rule-out admissions. Indeed, column 3 indicates that infants with a diagnosis of kernicterus only had a higher rate of admissions involving birth (80.1%), a lower length of stay (8.7 days), and a higher age at admission to the hospital (6.6 days). Thus, we used the case definition from column 4 to describe trends in hospitalizations of children with kernicterus. The phototherapy rate (93.9%) for these infants was much higher than the rate of exchange transfusion (11.6%). They also had the lowest rate of birth hospitalizations and the longest length of stay in the hospital relative to the other groups summarized in Table 1.
Table 2 reports the percentage of infants diagnosed with jaundice and the rate of infants diagnosed with kernicterus using the case definition described above in Table 1 (column 4). Data in Table 2 are organized by selected patient and hospital characteristics over the entire 18-year study period (from 1988 to 2005). Overall, 15.6% of hospitalized newborns received a diagnosis of jaundice and 2.7 per 100000 newborns received a diagnosis of kernicterus. The percentage of those diagnosed with jaundice was higher in Asian infants (19.4%) and lower in black infants (13.8%). Asian infants also had a higher rate of diagnosis for kernicterus (4.8 per 100000), which was significantly different from other racial/ethnic groups (P < .01). Male infants were more likely to be diagnosed with jaundice. The largest difference in diagnosis rates was evident in infant born prematurely. Preterm infants were more likely to be diagnosed with jaundice (52.4%) and kernicterus (9.2 per 100000) than term infants.
Data in Table 2 indicate a u-shaped relationship for rates of diagnosis of jaundice over time. Diagnosis rates were decreasing before the release of the initial AAP guideline on management of hyperbilirubinemia and increasing in the periods after their release. During 1994–1996, 11.9% of infants were diagnosed with jaundice compared with 17.3% for 1988–1990 and 20.0% in 2003–2005. In contrast, hospitalizations with a diagnosis of kernicterus decreased over the entire study period from 5.8 seconds to 1.6 per 100000 newborns.
Table 3 provides a weighted logistic regression model depicting the adjusted odds of being diagnosed with jaundice and kernicterus in the neonatal period. The adjusted odds ratio (aOR) for any characteristic in the table is calculated using all of the other characteristics as covariates in the logistic regression model. Examination of the adjusted odds ratios in Table 3 confirms the main findings from Table 2. A number of characteristics were associated with differences in the diagnosis of kernicterus. Asians were more likely to be diagnosed (aOR: 2.64 [95% confidence interval (CI): 1.66–4.20]) as were preterm infants relative to term infants (aOR: 4.22 [95% CI: 3.36–5.29]). Infants with public insurance also were more likely to be diagnosed with kernicterus (aOR: 1.34 [95% CI: 1.01–1.77]) and less likely to be diagnosed with jaundice (aOR: 0.87 [95% CI: 0.85–0.88]). There was no evidence of a relationship between rates of diagnosis of jaundice and kernicterus by region of the country. There was clear evidence of a reduction in hospitalizations with a diagnosis of kernicterus over time. Adjusted odds ratios for all the periods relative to the period 1988–1990 were significantly less than 1 and ranged from 0.54 to 0.23.
Figure 1 shows the percentage of hospitalized newborns diagnosed with jaundice, stratified by term and preterm gestation, over the 18-year study period. Hospitalizations with a diagnosis of jaundice seem to have increased in both term and preterm infants in the periods after publication of the AAP guideline. Figure 2 shows the rate of hospitalized neonates diagnosed with kernicterus stratified by term and preterm gestation over the 18-year study period. There is a large decline in the rate of term infants diagnosed with kernicterus in the periods before the publication of the AAP guideline. The rate of hospitalized term newborns diagnosed with kernicterus has remained relatively constant in the periods after publication of the guidelines at ∼1.5 per 100000 hospitalized newborns. The rate of hospitalized preterm infants diagnosed with kernicterus declined throughout the study period with the exception of the years between 1997 and 1999.
Table 4 provides evidence on trends in the use of phototherapy and exchange transfusions for term and preterm infants with jaundice. The data in Table 4 describe rates (as percentages or as rates per 10000) in relation to all term or preterm newborns identified from the HCUP database. Over the course of the study period, use of phototherapy associated with a diagnosis of jaundice in term infants declined before the publication of the guidelines and increased after their publication. In the most recent period, 2.62% of all newborns received phototherapy for the treatment of jaundice. For preterm infants, phototherapy for the treatment of jaundice increased over the entire study period from 15.2% to 27.6%. This change was significant (P < .01) and represents an 81.6% increase in utilization.
In contrast to the use of phototherapy, the use of exchange transfusions declined even as hospitalizations for newborns with a diagnosis of jaundice increased. The largest decrease in the use of exchange transfusions occurred in the periods before and immediately after the publication of the 1994 guideline for term infants and declined steadily in preterm infants. For term infants, rates fell 60% from 1.81 (95% CI: 1.51–2.12) to 0.73 (95% CI: 0.58–0.88) per 10000 over the period 1988–1996. From 1997 to 2005, rates of exchange transfusions fell only 17% in term infants. In preterm infants, use of exchange transfusions fell 42.5% from 16.0 (95% CI: 13.2–18.8) to 9.2 (95% CI: 7.4–11.0) per 10000 infants over the period 1988 to 1996. From 1997 to 2005, use of exchange transfusions fell 27.6% in preterm infants to a low of ∼6 in 10000 newborns.
Kernicterus is a rare newborn condition with tragic consequences. The best evidence available suggests that at least 10% of infants receiving the diagnosis will die acutely and survivors will have long-term morbidity.4 Given the severe consequences of kernicterus, recent case reports of kernicterus resulted in the Joint Commission on Accreditation of Healthcare Organizations issuing a sentinel alert and a number of authors warned of a possible global reemergence of the disease.8,9,34,35 Indeed, some authors36 attributed the reports of an increased incidence of kernicterus to the initial AAP guideline that adopted the “kinder and gentler” approach to the management of hyperbilirubinemia outlined by Newman and Maisels.11 Whether the incidence of kernicterus has changed over time was previously unknown as population-based data were not available.13 The lack of a large scale surveillance system for identifying cases of kernicterus in the population makes it impossible to evaluate the impact of the AAP guideline and more importantly, to design effective public policy recommendations for the optimal management of jaundice to prevent kernicterus.
Our study used 18 years of HCUP data covering over 13.6 million unweighted and nearly 70 million weighted newborn hospitalizations to evaluate trends in the diagnosis and management of newborn jaundice and kernicterus relative to the 1994 AAP guideline on the management of hyperbilirubinemia.6 An important test of the guideline is whether clinicians were more likely to diagnose and treat jaundice in newborn infants and whether hospitalizations associated with the diagnosis of kernicterus fell over time. The data indicate that hospitalizations for term and preterm newborn infants with a diagnosis code for jaundice and treated with phototherapy increased after the publication of the guideline. However, the relationship differed for term and preterm infants. Over the entire study period, the relationship was u-shaped for term infants with hospitalizations for newborn jaundice and utilization rates for phototherapy falling before publication of the guideline. In contrast, for preterm infants, hospitalizations with a diagnosis of jaundice increased after publication of the guideline, whereas utilization of phototherapy for the treatment of jaundice increased over the entire study period.
For hospitalizations with a diagnosis of kernicterus, there were notable differences between term and preterm infants. Most of the decline in hospitalizations with a diagnosis of kernicterus occurred before the publication of the guideline for term infants with the average incidence remaining flat over the period 1994–2005. In this period, ∼1.5 per 100000 newborn infants received the diagnosis compared with over 5 per 100000 in 1988–1990. For preterm infants, hospitalizations for a diagnosis of kernicterus decreased over most of the study period with the exception of an increase in incidence in the period 1997–1999 and no change in 2003–2005 relative to 2000–2002.
The findings from this study provide useful information for clinicians and health care organizations interested in the optimal management of jaundice in newborns. There is no evidence in the HCUP database of a resurgence of kernicterus or that the management strategies outlined in the 1994 guideline led to an increase in hospitalizations for kernicterus. This important feedback can promote optimal management strategies within and across health care organizations.18,37 Development of optimal management strategies requires evidence of the incidence of kernicterus.21 This study suggests that the incidence estimate for the diagnosis of kernicterus in term infants has fallen to ∼1.5 per 100000 newborns with a higher incidence in preterm infants (4 per 100000 births). Thus, the study supports increased attention to preterm infants with jaundice. Given its low incidence, achieving substantial reductions in kernicterus among term infants may be difficult or costly.14
Despite known limitations of hospital administrative data, the findings from the HCUP database are largely consistent with available epidemiologic data. The 1994 AAP guideline emphasized increased monitoring of hyperbilirubinemia in term infants, allowed for intensive use of phototherapy before initiating an exchange transfusion, and raised bilirubin thresholds for starting an exchange transfusion.38 If the recommendations led to an increase in the number of newborns diagnosed with jaundice, one would expect to see an increase in the number of newborns treated with phototherapy and a decrease in the use of exchange transfusions. We found evidence of these effects in the HCUP data in addition to a decline in hospitalizations with a diagnosis of kernicterus. Given the observed trends in diagnosis of jaundice and kernicterus, it seems that causal explanations for the decline in reported kernicterus hospitalizations are multifaceted.
The decreased use of exchange transfusions found in the HCUP data has been demonstrated with clinical data.38 Exchange transfusion is a more invasive treatment with a higher risk of an adverse event.39,40 Steiner et al38 note the trend toward declining use of exchange transfusions before the publication of the 1994 guideline and they point to literature predicting declines in its use dating back to the late 1960s.41
Finally, the data are consistent with the literature describing the epidemiology of kernicterus and hyperbilirubinemia. Risks of hyperbilirubinemia are increased in infants of Asian race, male gender, and lower gestational age.18 Risk is reduced in black infants. Data from this study support an increased risk of kernicterus in Asian and preterm newborns and a reduced risk of jaundice in black infants. We did find a statistically significant reduced risk of diagnosis for jaundice in publicly insured children who also had an increased risk of diagnosis of kernicterus, but support for a possible relationship seems limited, as there were no relationships between rates of jaundice and kernicterus by region or over time.
One potential inconsistency in the findings concerns the similar trends in hospital diagnosis and treatment rates for newborn jaundice in both term and preterm infants. The 1994 guideline addressed the management of hyperbilirubinemia in term infants, yet our analysis indicates an increase in diagnosis and treatment for both term and preterm infants. Increased treatment in preterm infants may simply be a spillover effect from the focus on term infants in the 1994 guideline.10,42 Alternatively, the finding in preterm newborns could be a consequence of the multi-factorial nature of changes in management practice. It is likely that clinical practice changed before the published guideline.
HCUP data, like other administrative databases, have a number of limitations that should be addressed. Hospital discharge data rely on accurate coding of diagnoses and therapies, with these codes principally used for reimbursement purposes. Coding of diagnoses and therapies, as well as racial and ethnicity data, can vary by hospital, state, and time period. Potential errors exist in the way ICD-9 codes get into the database. As an example, an admitting diagnosis of rule-out kernicterus could be coded as a final diagnosis of kernicterus. If the coder entered 774.7 (kernicterus not due to isoimmunization) only 1 time in a thousand in which they intended to enter 774.6 (unspecified fetal and neonatal jaundice) the incidence of the diagnosis of kernicterus would be significantly overestimated. It is not possible to test the validity of diagnosis according to well-defined criteria for acute bilirubin encephalopathy or kernicteric sequelae,2 nor is it possible to ascertain clinical values such as total serum bilirubin levels. For these reasons, it is possible that the observed decline in hospitalizations with a diagnosis of kernicterus reflects changes in coding practice rather than actual changes in the incidence of kernicterus.
In our initial analyses using only HCUP diagnosis codes, we found an incidence of the diagnosis of kernicterus ∼5 times what would be expected based on the incidence of extreme hyperbilirubinemia (30 mg/dL), which is expected to occur in 1 in 10000 newborns7,8,43–46 with a minority of these cases developing kernicterus. Thus, we adopted a case finding rule requiring an ICD-9-CM diagnosis code of kernicterus and an ICD-9-CM procedure code for either phototherapy or exchange transfusion. This case finding rule generated a dramatically decreased estimate (1.5 cases per 100000 births) more in line with expectations for the most recent study periods.
The ecological nature of this study also should be acknowledged. Although we observe findings consistent with adherence to the 1994 guideline in this study, it is likely a number of unobserved factors contributed to the decline in the diagnosis of kernicterus. One explanation for the decline is that shorter hospital length of stays among newborns has led to an increase in the rate of diagnosis of jaundice. In addition, better coding may make it seem that the rate of diagnosis of jaundice has increased. Thus, rather than guideline adherence driving the increased recognition of jaundice observed in the HCUP data, changes in the use of health services may be accountable. In addition, because we do not have individual identifiers, it is possible that rates of diagnosis of jaundice and kernicterus observed in the HCUP data are overstated because of double-counting. We examined changes in length of stay over the study period for newborns and did not find a substantial reduction, which may be attributable to the increase in Cesarean and premature births. Others have not found a relationship between length of stay in newborns and hyperbilirubinemia.18 The use of a 30-day window for selecting cases over all study years likely reduces double-counting but cannot eliminate it. This problem is potentially more severe for rates of jaundice than kernicterus, as infants diagnosed with jaundice can be admitted more than once in a 30-day time frame, whereas children diagnosed with kernicterus are older in terms of day of life at admission and have longer lengths of stay after admission.
No registry exists for kernicterus in the United States, nor is there any mandate to report the diagnosis of kernicterus. Such a registry, if it existed, would represent the best tool for describing trends of the incidence pattern of kernicterus, and would provide more accurate data than is available through the HCUP database. Despite the limitations of the HCUP database, it is the most powerful source of trend data currently available for studying rare conditions of childhood, such as kernicterus. It provides a cost-effective system to assess current trends in the diagnosis and management of jaundice and the incidence of kernicterus. We demonstrate how by combining nationally weighted hospital discharge data from the HCUP into distinct time-periods, we were able to detect a statistically significant change in hospitalizations for kernicterus despite an average incidence of 2.7 cases in 100000 births. Consistent use of selection criteria, diagnosis codes, and procedure codes over time suggests that the number of newborns diagnosed with jaundice and treated with phototherapy increased after publication of the guidelines and that the incidence of the diagnosis of kernicterus declined over the study period.
The power of the HCUP data to study important childhood conditions and evaluate the effectiveness of public health interventions should be recognized. In this study, we combined the NIS and the KID by using KID data where available. The KID has a superior sampling methodology for studying rare childhood conditions relative to the NIS, yet is produced only once every third year compared with every year for the NIS. Additional power to address important, but rare childhood conditions could be achieved with more frequent production of the KID database. Such efforts would not replace the value of registries to obtain clinical data to provide more refined tools for predicting at-risk children.16
The 1994 AAP guideline on the diagnosis and management of neonatal hyperbilirubinemia has likely succeeded in its aims. Since its introduction, infants hospitalized with a diagnosis of jaundice and treated with phototherapy have increased. Consistent with a “kinder and gentler” approach to the evaluation and management of jaundice,11 the use of exchange transfusions has also decreased. The data also indicates a substantial reduction in hospitalizations with a diagnosis of kernicterus. Although it is not possible to attribute the decline in hospitalizations to the AAP guideline, the findings provide support for the view that the guidelines did not lead to a resurgence of kernicterus. Given that a new guideline for the management of hyperbilirubinemia was issued in 2004, additional monitoring of jaundice and kernicterus in newborn infants seems warranted. It is important to note that the HCUP databases provide a cost-effective system for monitoring kernicterus and other uncommon newborn conditions and should be recognized as a powerful tool for public health systems research.
This work was supported by the Centers for Disease Control and Prevention through Association of American Medical Colleges (AAMC) grant U36/CCU319276, AAMC ID number MM-0636-04/04, as well as the Arkansas Children's Hospital Research Institute/Arkansas Biosciences Institute (grant ACHRI/ABI). The Arkansas Biosciences Institute is the major Arkansas research component of the Tobacco Settlement Proceeds Act of 2000.
- Accepted May 23, 2008.
- Address correspondence to John M. Tilford, PhD, Center for Applied Research and Evaluation, Department of Pediatrics, 800 Marshall St, Slot 512-26, Little Rock, AR 72202-3591. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
Kernicterus is a neonatal condition with high morbidity and mortality that is still present among US newborns. Efforts to prevent kernicterus include the development of guidelines for the systematic evaluation and management of jaundice in newborns.
What This Study Adds
Examination of trends in neonatal jaundice and kernicterus do not find evidence of a reemergence of kernicterus. The incidence of kernicterus declined early in the study period and has since remained constant at approximately 1.5 per 100 000 infants.
- ↵Halamek LP, Stevenson DK. Neonatal jaundice and liver disease. In: Fanaroff AA, Martin RJ, eds. Neonatal-Perinatal Medicine: Diseases of the Fetus and Neonate. 7th ed. St Louis, MO: Mosby; 2002:1324
- ↵American Academy of Pediatrics, Subcommittee on Hyperbilirubinemia. Management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation [published correction appears in Pediatrics. 2004;114(4):1138]. Pediatrics.2004;114 (1):297– 316
- ↵Ip S, Chung M, Kulig J, O'Brien R, et al. An evidence-based review of important issues concerning neonatal hyperbilirubinemia. Pediatrics.2004;114 (1). Available at: www.pediatrics.org/cgi/content/full/114/1/e130
- ↵American Academy of Pediatrics, Provisional Committee for Quality Improvement and Subcommittee on Hyperbilirubinemia. Practice parameter: management of hyperbilirubinemia in the healthy term newborn. Pediatrics.1994;94 (4 pt 1):558– 565
- ↵Sgro M, Campbell D, Shah V. Incidence and causes of severe neonatal hyperbilirubinemia in Canada. CMAJ.2006;175 (6):587– 590
- ↵Joint Commission on Accreditation of Healthcare Organizations. Kernicterus threatens healthy newborns. Sentinal Event Alert.2001;18 :1– 4
- ↵Newman TB, Maisels MJ. Evaluation and treatment of jaundice in the term newborn: a kinder, gentler approach. Pediatrics.1992;89 (5 pt 1):809– 818
- ↵Brown AK, Seidman DS, Stevenson DK. Jaundice in healthy, term neonates: do we need new action levels or new approaches? Pediatrics.1992;89 (5 pt 1):827– 829
- ↵Suresh GK, Clark RE. Cost-effectiveness of strategies that are intended to prevent kernicterus in newborn infants. Pediatrics.2004;114 (4):917– 924
- ↵Bhutani V, Johnson LH. Kernicterus: lessons for the future from a current tragedy. NeoReviews.2003;4 (2):e30– e32
- ↵Newman TB, Klebanoff MA. Neonatal hyperbilirubinemia and long-term outcome: another look at the Collaborative Perinatal Project. Pediatrics.1993;92 (5):651– 657
- ↵Maisels MJ, Newman TB. Surveillance of severe neonatal hyperbilirubinemia: a view from south of the border. CMAJ.2006;175 (6):599
- ↵Raju TN, Higgins RD, Stark AR, Leveno KJ. Optimizing care and outcome for late-preterm (near-term) infants: a summary of the workshop sponsored by the National Institute of Child Health and Human Development. Pediatrics.2006;118 (3):1207– 1214
- ↵Blackmon LR, Fanaroff AA, Raju TN. Research on prevention of bilirubin-induced brain injury and kernicterus: National Institute of Child Health and Human Development conference executive summary, 2003. Pediatrics.2004;114 (1):229– 233
- ↵Healthcare Cost and Utilization Project (HCUP). The HCUP Nationwide Inpatient Sample (NIS). Rockville, MD: Agency for Healthcare Research and Quality; 2004
- ↵Healthcare Cost and Utilization Project (HCUP). The HCUP Kids' Inpatient Database (KID). Rockville, MD: Agency for Healthcare Research and Quality; 2005
- ↵Holman RC, Curns AT, Belay ED, Steiner CA, Schonberger LB. Kawasaki syndrome hospitalizations in the United States, 1997 and 2000. Pediatrics.2003;112 (3 pt 1):495– 501
- Healthcare Cost and Utilization Project. 2003 Kids' Inpatient Database (KID) Technical Documentation. Rockville, MD: Agency for Healthcare Research and Quality; 2005
- ↵Healthcare Cost and Utilization Project. 2002 Nationwide Inpatient Sample Technical Documentation. Rockville, MD: Agency for Healthcare Research and Quality; 2004
- Robbins JM, Tilford JM, Bird TM, Cleves MA, Reading JA, Hobbs CA. Hospitalizations of newborns with folate-sensitive birth defects before and after fortification of foods with folic acid. Pediatrics.2006;118 (3):906– 915
- ↵Penn AA, Enzmann DR, Hahn JS, Stevenson DK. Kernicterus in a full term infant. Pediatrics.1994;93 (6 pt 1):1003– 1006
- ↵MacDonald MG. Hidden risks: early discharge and bilirubin toxicity due to glucose 6-phosphate dehydrogenase deficiency. Pediatrics.1995;96 (4 pt 1):734– 738
- ↵Brown AK, Johnson L. Loss of concern about jaundice an the reemergence of kernicterus in full term infants in the era of managed care. In: Fanaroff AA, Klaus M, eds. The Yearbook of Neonatal and Perinatal Medicine. St Louis, MO: Mosby; 1996:17– 28
- ↵Clancy CM. Getting to “smart” health care. Health Aff (Millwood).2006;25 (6):w589– w592
- ↵Steiner LA, Bizzarro MJ, Ehrenkranz RA, Gallagher PG. A decline in the frequency of neonatal exchange transfusions and its effect on exchange-related morbidity and mortality. Pediatrics.2007;120 (1):27– 32
- ↵Jackson JC. Adverse events associated with exchange transfusion in healthy and ill newborns. Pediatrics.1997;99 (5). Available at: www.pediatrics.org/cgi/content/full/99/5/e7
- Newman TB, Liljestrand P, Escobar GJ. Infants with bilirubin levels of 30 mg/dL or more in a large managed care organization. Pediatrics.2003;111 (6 pt 1):1303– 1311
- ↵Manning D, Todd P, Maxwell M, Jane PM. Prospective surveillance study of severe hyperbilirubinaemia in the newborn in the UK and Ireland. Arch Dis Child Fetal Neonatal Ed.2007;92 (5):F342– F346
- Copyright © 2009 by the American Academy of Pediatrics