Condition, prevalence, incidence and trend direction — neonatal hyperbilirubinemia home phototherapy

1. Summary

Two independent national-inpatient-database studies covering overlapping but distinct windows (1997–2012 and 2002–2017) both find that phototherapy use during the birth hospitalization rose substantially in the US while kernicterus/bilirubin-neurotoxicity discharges fell in term infants [1][2]. Hyperbilirubinemia and its treatment are not evenly distributed: Black neonates are diagnosed with hyperbilirubinemia less often than White neonates but are markedly more likely to progress to bilirubin neurotoxicity [1], and glucose-6-phosphate dehydrogenase (G6PD) deficiency — a recognized risk factor for severe hyperbilirubinemia — is modeled at roughly 17 per 1,000 US live births nationally but reaches more than twice that rate in the most racially/ethnically diverse states [3]. A hospital-system cohort that began universal G6PD screening under a state mandate found deficient infants were significantly more likely to need phototherapy, and that a risk-factor-based screening approach (rather than universal screening) would have missed 44% of affected newborns [4]. Nationally, Medicaid was the payment source for 41.5% of all US births in 2023 [6]. None of the sources located in this pass establishes the single figure this domain's dossier flags as missing — the fraction of US newborns ≥35 weeks gestation who receive phototherapy at all, in any setting.

2. What Changed

_Baseline (v1). No prior version; this establishes the starting point for future diffs._

3. Details

National trend in phototherapy use and severe outcomes

A National Inpatient Sample (NIS) analysis of 57,989,476 US newborn inpatients from 2002–2017 (91.8% term, 8.2% preterm) found that bilirubin neurotoxicity (kernicterus) decreased significantly over the period in term infants (Cochran-Armitage Z = 0.36, p = 0.03) with no significant change in preterm infants, and that "utilization of phototherapy has increased significantly over the years" — its own stated headline finding [1]. The overall bilirubin-neurotoxicity rate across the full 2002–2017 cohort was 2.4 per 100,000 live births [1]. This is consistent with an earlier HCUP Kids' Inpatient Database (KID) analysis of 1997–2012 discharges, which reported phototherapy use increasing 83% and exchange-transfusion use falling 67% over that window, alongside kernicterus discharges falling from 7 to 1.9 per 100,000 newborns [2]. The two studies use different national databases (NIS vs. KID), different years, and different research groups, and agree on both direction and general magnitude of the phototherapy-utilization and kernicterus trends.

Racial and ethnic distribution of diagnosis and severe outcomes

The same NIS 2002–2017 analysis reported that Black neonates were less likely than White neonates to be diagnosed with hyperbilirubinemia (adjusted odds ratio 0.77, 95% CI 0.77–0.78, p < 0.001) but were about three times more likely to develop bilirubin neurotoxicity (adjusted odds ratio 3.05, 95% CI 2.13–4.36, p < 0.001) [1]. The paper's own stated interpretation: "although Black newborns have less neonatal jaundice, they are at increased risk of developing kernicterus" [1]. This is a single national database and a single research group; no second, independent US source corroborating this specific disparity was located in this pass — [single-source].

G6PD deficiency: a concentrated, underscreened risk factor

A modeling study combining 2019 CDC-WONDER natality data with published race-specific G6PD prevalence estimates projected 78,010 (95% CI 76,768–79,252) US newborns with G6PD deficiency in 2019, a national median prevalence of 17.3 (interquartile range 12.4–23.2) per 1,000 live births, with five states (Washington DC, Mississippi, Louisiana, Georgia, Maryland) projected at 35–48 per 1,000 — several-fold higher than the national median, correlating with each state's population diversity index (p < 0.0005) [3]. A retrospective cohort of 5,470 infants at a New York hospital system, gathered during the first year of universal G6PD screening under that state's 2022 mandate, found empirical prevalence of 1.7% deficient and 2.4% intermediate (2.9% of male infants deficient); G6PD-deficient infants had higher bilirubin levels and were significantly more likely to require phototherapy during the birth hospitalization (p < 0.001) and to be readmitted for phototherapy (p = 0.04) than G6PD-sufficient infants, and the same cohort found that risk-factor-based screening (rather than universal screening) would have missed 44% of G6PD-deficient newborns before discharge [4]. These two figures — a modeled 17.3-per-1,000 national estimate and an empirically measured 17-per-1,000 (1.7%) rate in one hospital system — are independent, non-affiliated, and numerically consistent, though they use different methods (population modeling vs. direct enzyme testing) and different populations. A non-US cohort study of 40,305 consecutively born infants in Qatar found 2.51% incidence of G6PD deficiency, of whom 24.6% received phototherapy versus a lower rate among G6PD-sufficient infants — directional, non-US, cited here only to show the same clinical association (G6PD deficiency raising phototherapy need) replicates outside the US [single-source].

Payer mix at the national level

Medicaid was the source of payment for 41.5% of all US births in 2023, up from 41.3% in 2022, per the National Vital Statistics Reports final natality report for that year [6]. This is a national complement to the domain dossier's existing single-state (Louisiana) DMEPOS fee-schedule finding — it establishes the national order of magnitude for how much of the birth-hospital and downstream newborn-care population is Medicaid-financed, but it is a birth-level payer statistic, not a payer-mix figure specific to newborns who go on to need phototherapy — [single-source].

4. Sources

[1] Neonatal hyperbilirubinemia and bilirubin neurotoxicity in hospitalized neonates: analysis of the US Database — Pediatric Research (published 2022-06, day not given in the PubMed record; accessed 2026-09-01). PMID 34429513 — https://pubmed.ncbi.nlm.nih.gov/34429513/ [peer-reviewed] [2] Trends in hospitalizations of newborns with hyperbilirubinemia and kernicterus in United States: an epidemiological study — The Journal of Maternal-Fetal & Neonatal Medicine (published 2022, month not given in the PubMed record; accessed 2026-09-01). PMID 34470114 — https://pubmed.ncbi.nlm.nih.gov/34470114/ [peer-reviewed] [3] Georacial Epidemiological Estimates of Glucose-6-Phosphate Dehydrogenase Deficiency among Newborns in the United States — American Journal of Perinatology (published 2024-05, day not given in the PubMed record; accessed 2026-09-01). PMID 37105226 — https://pubmed.ncbi.nlm.nih.gov/37105226/ [peer-reviewed] [4] Outcomes of Universal Newborn G6PD Deficiency Screening in a Large Urban Cohort — Pediatrics (published 2026-01-01; accessed 2026-09-01). PMID 41325981 — https://pubmed.ncbi.nlm.nih.gov/41325981/ [peer-reviewed] [5] Glucose-6-phosphate dehydrogenase deficiency and neonatal indirect hyperbilirubinemia: a retrospective cohort study among 40,305 consecutively born babies — Journal of Perinatology (published 2024, month not given in the PubMed record; accessed 2026-09-01). PMID 38480787 — https://pubmed.ncbi.nlm.nih.gov/38480787/ [peer-reviewed] [6] Births: Final Data for 2023 — National Vital Statistics Reports, Vol. 74 No. 1, National Center for Health Statistics (published 2025-03-18; accessed 2026-09-01). DOI 10.15620/cdc/175204 — https://www.ncbi.nlm.nih.gov/books/NBK618136/ [federal-registry]

5. Sourcing & Gaps

Well established: The direction of the national trend — phototherapy utilization rising and kernicterus/bilirubin-neurotoxicity discharges falling in term infants over the 2000s–2010s — rests on two independent national inpatient-database studies using different data sources and different study windows, and is stated plainly in Sections 1 and 3 [1][2].

Thin: The racial disparity in hyperbilirubinemia diagnosis versus bilirubin-neurotoxicity risk rests on one national database study and no independent US corroboration was located [single-source] [1]. The G6PD national-prevalence estimate is a model, not a direct count [single-source] [3], though it is numerically consistent with the one empirical hospital-system cohort located [4]. The Qatar G6PD cohort [5] is non-US and used only as directional corroboration that the G6PD-phototherapy association replicates outside the US population this dossier covers. The national Medicaid birth-payer share [6] is a single source and is a birth-level, not phototherapy-specific, figure.

Rescoped from class 3: None in this report. Every proposition above is a class-2 measured quantity from a named cohort or database, not a question about what a person would do.

Out of scope: CPT-coded professional-service data, veterinary datasets, and any figure requiring a fda.gov guidance document were not pursued, consistent with this domain's dossier. State-by-state birth-volume ranking (which states carry the most newborn volume, and therefore the most Medicaid-financed births) was searched via WebSearch only; the results pointed to secondary news coverage (e.g., a Newsweek summary) of CDC birth data rather than a primary NCHS table this pass could open and quote directly, so no state-level birth-count figure is recorded here — a search result is not a source of record.

Not searched vs. not found: The population-wide fraction of US newborns ≥35 weeks receiving phototherapy in any setting was searched for directly in this pass (multiple literature queries on population-based phototherapy incidence) and not found — this is a "not found," not a "not searched." CDC WONDER natality microdata were not queried directly in this pass (no connector reaches CDC WONDER); the G6PD prevalence model [3] used CDC-WONDER-derived birth statistics from 2019 as an input, but this report did not independently verify that input against CDC WONDER itself — "not searched" for this dossier's own use.

[inference] The numerical similarity between the modeled national G6PD prevalence (17.3 per 1,000, [3]) and the empirically measured rate at one New York hospital system (17 per 1,000 deficient, i.e. 1.7%, [4]) is treated in Section 3 as two independent sources agreeing, which is the writer's own synthesis across a model and a direct measurement rather than a finding either paper states about the other — a reader should weigh this as suggestive, not as two replications of the same measurement.

6. Claim Candidates