OBJECTIVE. We sought to describe population-based trends, potential risk factors, and hospital costs of readmission for jaundice for term and late preterm infants.
METHODS. Birth-cohort data were obtained from the California Office of Statewide Health Planning and Development and contained infant vital statistics data linked to infant and maternal hospital discharge summaries. The study population was limited to healthy, routinely discharged infants through the use of multiple exclusion criteria. All linked readmissions occurred within 14 days of birth. International Classification of Diseases, Ninth Revision, codes were used to further limit the sample to readmission for jaundice. Hospital discharge records were the source of diagnoses, hospital charges, and length-of-stay information. Hospital costs were estimated using hospital-specific ratios of costs to charges and adjusted to 1991.
RESULTS. Readmission rates for jaundice generally rose after 1994 and peaked in 1998 at 11.34 per 1000. The readmission rate for late preterm infants (as a share of all infants) over the study period remained at <2 per 1000. Factors associated with increased likelihood of hospital readmission for jaundice included gestational age 34 to 39 weeks, birth weight of <2500 g, male gender, Medicaid or private insurance, and Asian race. Factors associated with a decreased likelihood of readmission for jaundice were cesarean section delivery and black race. The mean cost of readmission for all infants was $2764, with a median cost of $1594.
CONCLUSIONS. Risk-adjusted readmission rates for jaundice rose following the 1994 hyperbilirubinemia guidelines and declined after postpartum length-of-stay legislation in 1998. In 2000, the readmission rate remained 6% higher than in 1991. These findings highlight the complex relationship among newborn physiology, socioeconomics, race or ethnicity, public policy, clinical guidelines, and physician practice. These trend data provide the necessary baseline to study whether revised guidelines will change practice patterns or improve outcomes. Cost data also provide a break-even point for prevention strategies.
Kernicterus continues to appear in healthy term newborns, despite the advent of Rh immunoglobulin prophylaxis, phototherapy, and exchange transfusion to treat hyperbilirubinemia. Several sources have suggested that a resurgence in cases of hyperbilirubinemia and kernicterus is because of earlier discharge from the hospital and more accepting attitudes of higher bilirubin levels in otherwise healthy term infants.1–5
More specifically, discussions focus on 2 events presumed to impact the management of hyperbilirubinemia. First, in 1994 the American Academy of Pediatrics (AAP) published a clinical practice guideline on the management of hyperbilirubinemia in healthy term newborns, but this guideline was not followed by clear changes in practice patterns of newborn care in the United States.6–11 As the management of Rh isoimmunization improved, the incidence of acute bilirubin encephalopathy and kernicterus also improved. The 1994 guideline was intended to limit unnecessary investigation and treatment for newborns with uncomplicated physiologic jaundice or breast milk-related jaundice. Second, in January of 1997, California adopted postpartum length-of-stay legislation. This was followed by the federal legislation, which was adopted in January 1998. Both pieces of legislation required insurance companies to cover a minimum hospital stay of 2 days for vaginal births and 4 days for cesarean section births. The California legislation also mandated follow-up visits in the event of an early discharge.
The impact of these events on the incidence of kernicterus is difficult to document because of rarity of occurrence and variable reporting. However, jaundice is relatively common, occurring in ∼60% of newborns, and readmission with a primary diagnosis of jaundice occurs in ∼1% of the term or late preterm (≥35 weeks) newborn population.12 Thus, the impact of these sociopolitical events may be revealed in longitudinal, population-based data of hospital readmission for jaundice. Moreover, the data would provide a baseline for tracking the impact of the more recent AAP clinical guidelines for the management of hyperbilirubinemia.13
Such data are available from the California Office of Statewide Health Planning and Development (OSHPD). These data link the OSHPD summaries of maternal delivery hospitalizations summaries, infant hospital discharges summaries, and infant rehospitalizations summaries to the infant vital statistics data (birth and death certificates). The linkage algorithm used by the OSHPD is highly accurate.14 The richness of this linked data allowed us to describe population-based trends in rehospitalization for jaundice or hyperbilirubinemia for term and late preterm infants; to examine the relationship among sociodemographic, medical, and payer characteristics and readmission; and to determine the hospital costs of readmission.
Linked data were obtained for the 1991–2000 California birth cohorts. In this study, 98% of maternal hospitalizations were linked with infant birth certificates, and >99% of the infant discharge abstracts were linked with the birth certificates. Hospital discharge abstracts were the source of information on discharge diagnoses, lengths of stay, and hospital charges.
Trends in Readmission for Jaundice
The study population was limited to well infants. This was achieved through multiple exclusion criteria. Gestational age was confirmed by using the Kotelchuck imputation, and infants were required to be ≥34 weeks’ gestation and ≤42 weeks’ gestation. Length of stay was computed as the total number of hospital days until the first discharge to home. Infants delivered vaginally were limited to those with lengths of stay ≤3 days. Infants delivered by cesarean section were limited to those with lengths of stay ≤5 days. Infants transferred to another hospital were ineligible. Infants with any indication of neonatal illness during birth hospitalization were also ineligible. Infants who may have received phototherapy before discharge or at home were not excluded.
All of the linked readmissions of the study population occurred within 14 days of birth. International Classification of Diseases, Ninth Revision (ICD-9), codes were used to further limit the sample to readmission for jaundice or a related diagnosis. Table 1 shows the acceptable related diagnoses for inclusion in the main sample. Four related diagnoses were excluded from the main sample: 243.0 with congenital hypothyroidism, 775.5 with dehydration neonatal, 779.3 with feeding problems in newborn, and 783.3 with feeding difficulties and mismanagement. All of the readmissions were screened for unrelated diagnoses as the cause of readmission (eg, hypoplastic left heart), and, if an unrelated cause was found, the case was excluded.
Risk of Readmission Based on Sociodemographic and Medical Characteristics
Logistic regression analysis was used to obtain odds ratios (ORs) for the variables related to readmission, controlling for maternal and infant demographic characteristics. In the logistic regression model, records with missing values for any of the predictor variables were excluded from the analysis. The following maternal and infant variables were used in the regression analysis: (1) gestational age (34–42 weeks); (2) type of delivery (vaginal or cesarean); (3) birth weight (<2500 g, 2500–4000 g, or ≥4000 g); (4) gender (male or female); (5) length of delivery hospital stay (<2 days or ≥2 days); (6) first trimester prenatal care; (7) mother's age; (8) mother's education (at or less than eighth grade, ninth to 11th grade, or ≥12th grade); (9) insurance type covering delivery (health maintenance organization [HMO]/preferred provider organization [PPO], Medicaid, private, self-pay, or other); (10) race or ethnicity (white, Asian, black, Hispanic, or other); (11) mother's place of birth (US-born or foreign-born); and (12) year of infant's birth.
The insurance variable “private” included indemnity plans and/or any other private insurance plan that was not categorized as managed care (HMO/PPO). OSHPD data did not include the PPO variable until after 1994. The variable “self-pay” included mothers who were uninsured and/or chose to pay directly for a service.
Costs of Readmission
The hospital charges reported in the OSHPD discharge data represent total charges, not department-specific charges. To determine hospital costs for each readmission, we multiplied the charges for each hospital stay by a hospital-specific cost-to-charge ratio derived from annual hospital financial data compiled by the OSHPD. Costs per hospital day were then computed by dividing total adjusted costs by total length of stay. All of the costs were adjusted to December 2001. Cases without cost data and cases with outlying or improbable adjusted costs per day were identified and excluded from the analyses. We excluded any infant readmission case with an adjusted cost per day less than $300 or more than $10 000.
Trends in Readmission for Jaundice and Related Conditions
Table 2 displays trends in readmission rates for jaundice and related conditions for the study population in California from 1991 to 2000. Selected trends are also shown in Fig 1. In 1991, the readmission rate for all healthy, routinely discharged term or late preterm infants was 9.34 per 1000. Readmission rates did not decrease after the release of the 1994 AAP guidelines. In general, the rate increased through 1998 to a peak of 11.34 per 1000 and only then started to decrease, although there was no change from 1999 to 2000.
Trends in readmission rates for other stratified groups of infants mirrored the overall trend for the most part. The readmission rate for late preterm infants (as a share of all infants) over the study period remained at <2 per 1000, as did the rates for infants delivered by cesarean section. Readmission rates of infants with length of stay of <2 days, with white race, with Medicaid, and with mothers <20 years old peaked 2 years earlier in 1996. Readmission rates for infants of privately insured mothers decreased steadily over the study period.
Risk of Readmission Based on Sociodemographic and Medical Characteristics
Table 3 displays the results of the regression analysis to determine the odds of readmission based on several sociodemographic and medical characteristics. The factors most strongly associated with increased likelihood of readmission for jaundice were gestational age 34 to 37 weeks (OR: 2.33–3.18), gestational age 38 weeks (OR: 1.64), Medicaid or private insurance (OR: 1.45), and Asian race (OR: 1.53). Infants with birth weight of <2500 g or male gender also had slightly higher odds of readmission (OR: 1.24 and 1.37, respectively).
The factors most strongly associated with a decreased likelihood of readmission for jaundice were cesarean delivery (OR: 0.44) and black race (OR: 0.34). Several maternal factors were associated with slightly decreased likelihood of rehospitalization, including prenatal care before the sixth month, <12th grade education, Hispanic ethnicity, and foreign born.
Costs of Readmission
The mean hospital length of stay was 2.5 days, with a median of 2.0 days and a range of 0.0–155.0 days. The adjusted mean cost for each readmission was $2764, with a median cost of $1594 and a range of $300 to $756 868. The mean cost per day was $991, with a median of $819 and a range of $300 to $9727. The variance in length of stay and total cost seemed to be because of a small number of outliers. When the 4 additional related diagnoses (243.0 with congenital hypothyroidism, 775.5 with neonatal dehydration, 779.3 with feeding problems in the newborn, and 783.3 with feeding difficulties and mismanagement) were included in the sample, length of stay remained unchanged, but adjusted mean cost per readmission rose 10% to $3028, as cost per day rose 3.3% to $1025.
This study is one of the first to show population-based trends in readmissions for jaundice. Over the 10-year period from 1991 to 2000, the overall readmission rate increased through 1998. When stratified by sociodemographic, medical, and payer characteristics, most groups followed a similar trend.
Two major sociopolitical events occurred in temporal relation to inflection points on the trend line. First, in October 1994, the AAP released the practice parameter Management of Hyperbilirubinemia in the Healthy Term Newborn.6 This guideline was intended to limit unnecessary investigation and treatment for newborns with uncomplicated physiologic jaundice or breast milk-related jaundice. Studies of practice patterns after release of the guideline showed wide variation, with neonatologists initiating phototherapy and exchange transfusions at lower serum bilirubin concentrations than office-based pediatricians. Practices of office-based pediatricians more closely approximated the guideline.15 Studies also showed a decrease in the use of phototherapy and exchange transfusion after the publication of the guideline.16 This more relaxed approach to jaundice converged with 2 other trends: a push toward early postnatal discharge of term and late preterm infants and initiatives to promote breastfeeding. Our population-based data suggest that these 3 factors had a synergistic effect, ultimately resulting in an overall increase in readmissions for jaundice.
Risks related to early postnatal discharge were addressed by the second event, the adoption of postpartum length of stay legislation in January 1997. Early discharge limits a provider's ability to ensure adequate feeding and elimination, to recognize early jaundice, and to ensure close follow-up. The increase in jaundice readmissions before 1998 was attributable to term newborns delivered vaginally and discharged early, suggesting that the postpartum stay legislation was appropriately targeted. Recent studies of the impact of postpartum stay legislation have indeed shown a decrease in jaundice-related admissions in 2 and 3 years after passage of the law.17 Our post-1998 trend also reflects this decrease.
Despite the trend data, our regression analysis did not reveal increased odds of readmission with length of stay of <2 days. One previous study of discharge timing and newborn readmission also showed that infants delivered vaginally with 1 night of hospital stay and adequate prenatal and postnatal care outside the hospital did not have an increased risk of readmission.18 Interestingly, women and newborns who are Hispanic, have low incomes, or are insured by Medicaid are significantly more likely to be discharged earlier than recommended.11 In our regression equation, similar variables (ethnicity, education, and Medicaid) may have decreased the effect attributable to shorter length of stay.
Reasons for decreased risk of readmission in Hispanic infants are unclear and demand additional investigation. Hispanic infants evaluated for jaundice were shown elsewhere to have higher total bilirubin levels compared with other groups.19 Fewer readmissions despite higher bilirubin levels suggests the existence of protective factors. The epidemiologic paradox is well described for infants of foreign-born mothers, and our data showed that these newborns also have a lower risk for readmission for jaundice. Some have suggested that the epidemiologic paradox applies only to certain subsets of the population, for example, Mexican immigrants.20 Because California has a large number of Mexican immigrants, this could explain a portion of the protective effect of being Hispanic.
In contrast, late preterm gestation, low birth weight, male gender, and Asian ethnicity increased the risk of readmission, with late preterm infants having ≤3 times the risk. Later and higher peaks in serum bilirubin, increased breastfeeding rates, and “mistimed” follow-up visits combine to put the late preterm population at higher risk. Recent literature highlights greater risk for respiratory distress, hypothermia, hypoglycemia, and jaundice.21 The emergence of late preterm gestation as an independent risk factor for hyperbilirubinemia and readmission is a notable avenue for additional research.
Timing of follow-up may also help to explain why infants of both publicly and privately insured mothers had slightly increased odds of readmission. The 2 groups of infants may have become jaundiced for the same reason, lack of timely follow-up, but this does not explain why the privately insured were at higher risk than those with managed care. A second possible explanation is that the 2 groups of infants received disparate care: the publicly insured group admitted at higher bilirubin levels secondary to poor follow-up and the privately insured readmitted at lower levels secondary to close follow-up and conservative management. Linked laboratory values would be required to evaluate physician practice at this level, and this is another avenue for future research.
The risk profiles and sheer number of births encourage us to consider the cost-effectiveness of preventive strategies. In California, ∼1% of all well infants were readmitted for jaundice, and the mean cost per readmission was approximately $3000. Thus, the break-even point for preventing 1 readmission is approximately $30 per infant. The costs of 3 recommended kernicterus prevention strategies have been estimated to be between $7.80 (predischarge transcutaneous bilirubin) and $29 (universal office visit).22 However, estimating true cost-effectiveness of prevention strategies would be difficult given practice variability, unknown incidence of kernicterus, and unknown costs related to policies, information exchange, and technology.
This population-based study of linked data has certain limitations. First, these data are compiled with limited clinician review. Clinicians reviewed the hospital discharge diagnoses and made a judgment as to whether the primary reason for hospitalization was because of jaundice or a related cause. Second, these judgments are based on diagnostic codes assigned at the time of the hospitalization, and coding errors may have occurred during data collection. Third, even when correct, ICD-9 codes provide no detail with regard to the hospitalization, resulting in a limited ability to trace outcomes, such as complications or jaundice-induced injuries (chronic encephalopathy, developmental or motor delay, and deafness). Moreover, the data did not include incidence of kernicterus for the study population. Finally, the data provide no clinical information to relate to clinician adherence or the use of clinical guidelines, such as risk factors for hyperbilirubinemia or total bilirubin values. Thus, the threshold and criteria for readmission were not known through the study period.
A key assumption in this study is that the rate of readmission for hyperbilirubinemia represents the frequency of severe hyperbilirubinemia (itself a surrogate for kernicterus). This assumption is true only if the criteria for diagnosis, testing, readmission, and treatment for hyperbilirubinemia remain unchanged throughout the study period and if home phototherapy rates remain unchanged. Otherwise, readmission rates may simply represent shifts in thresholds for testing and treating hyperbilirubinemia or in the use of home phototherapy. Similar to the discussion above regarding insurance coverage, more readmissions may represent either earlier diagnosis and treatment or missed cases that required hospitalization.
The key strength of this study is that it is population based. The use of linked data allowed us to generate very large sample sizes, which are good for examining rare events. These methods will allow us to continue to longitudinally follow trends in readmission, as well as incidence of more serious jaundice-related injuries. This study also illustrates the interrelationship and influence of practice guidelines and public policy on hospital readmission.
In 1994, the AAP published guidelines for the management of hyperbilirubinemia in the healthy term newborn.6 Our study shows that risk-adjusted readmission rates for jaundice rose after the 1994 guidelines and peaked in 1998 after changes in postpartum stay legislation. Despite slowing, in the year 2000, the readmission rate was still 6% higher than in 1991. These results highlight the complex relationship among newborn physiology, socioeconomics, race or ethnicity, public policy, clinical guidelines, and physician practice. Linking population-based data with risk factors for hyperbilirubinemia and total serum bilirubin values is critical to further elucidate physician practice patterns and behaviors. Nonetheless, this study provides the necessary baseline to study whether revised guidelines will change practice patterns or improve outcomes.
The work of Drs Phibbs and Schmitt was supported by National Institute of Child Health and Human Development grant HD36914.
- Accepted September 6, 2007.
- Address correspondence to Anthony E. Burgos, MD, MPH, 750 Welch Rd, Suite 325, Palo Alto, CA 94304. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
What's Known on This Subject
Practice patterns became less conservative after the release of the 1994 hyperbilirubinemia guidelines, and early discharge after delivery is considered a risk factor for hyperbilirubinemia. Kernicterus is a rare outcome, but jaundice readmission is more common and quite costly.
What This Study Adds
This study describes 10-year trends in jaundice readmission in relation to 2 key sociopolitical events and provides the baseline rates and cost data needed to study changes in practice patterns or outcomes and cost of prevention strategies.
- ↵Braveman P, Egerter S, Pearl M, Marchi K, Miller C. Problems associated with early discharge of newborn infants. Early discharge of newborns and mothers: a critical review of the literature. Pediatrics.1995;96 (4 pt 1):716– 726
- Britton JR, Britton HL, Beebe SA. Early discharge of the term newborn: a continued dilemma. Pediatrics.1994;94 (3):291– 295
- Watchko JF, Oski FA. Bilirubin 20 mg/dL = vigintiphobia. Pediatrics.1983;71 (4):660– 663
- ↵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
- ↵American Academy of Pediatrics, Provisional Committee for Quality Improvement. Practice parameter: management of hyperbilirubinemia in the healthy term newborn [published correction appears in Pediatrics. 1995;95(3):458–461]. Pediatrics.1994;94 (4 pt 1):558– 565
- Maisels MJ, Kring E. Early discharge from the newborn nursery and effect on scheduling of follow-up visits by pediatricians. Pediatrics.1997;100 (1):72– 74
- Parisi Eaton A. Early postpartum discharge: recommendations from a preliminary report to Congress. Pediatrics.2001;107 (2):400– 404
- ↵Galbraith AA, Egerter SA, Marchi KS, Chavez G, Braveman PA. Newborn early discharge revisited: are California newborns receiving recommended postnatal services? Pediatrics.2003;111 (2):364– 371
- ↵Escobar GJ, Greene JD, Hulac P, et al. Rehospitalisation after birth hospitalisation: patterns among infants of all gestations. Arch Dis Child.2005;90 (2):125– 131
- ↵American Association of Pediatrics Subcommittee on Hyperbilirubinemia. Management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation. Pediatrics.2004;114 (1):297– 316
- ↵Gartner LM, Herrarias CT, Sebring RH. Practice patterns in neonatal hyperbilirubinemia. Pediatrics.1998;101 (1 pt 1):25– 31
- ↵Datar A, Sood N. Impact of postpartum hospital-stay legislation on newborn length of stay, readmission, and mortality in California. Pediatrics.2006;118 (1):63– 72
- ↵Danielsen B, Castles AG, Damberg CL, Gould JB. Newborn discharge timing and readmissions: California, 1992–1995. Pediatrics.2000;106 (1 pt 1):31– 39
- ↵Engle WD, Jackson GL, Sendelbach D, Manning D, Frawley WH. Assessment of a transcutaneous device in the evaluation of neonatal hyperbilirubinemia in a primarily Hispanic population. Pediatrics.2002;110 (1 pt 1):61– 67
- ↵Wang ML, Dorer DJ, Fleming MP, Catlin EA. Clinical outcomes of near-term infants. Pediatrics.2004;114 (2):372– 376
- ↵Suresh GK, Clark RE. Cost-effectiveness of strategies that are intended to prevent kernicterus in newborn infants. Pediatrics.2004;114 (4):917– 924
- Copyright © 2008 by the American Academy of Pediatrics