The finding. Among US newborns who were full‑term, normal birth weight and not twins — a group defined to be low‑risk — the share recorded as admitted to newborn intensive care rose from 3.84% in 2016 to 4.50% in 2024. That is a 17.2% relative increase. Over the same nine years the rate among babies born before 37 weeks rose only 5.0%. The increase is largest in the group least likely to need intensive care.
The one thing this cannot settle. The birth certificate is known to under‑record newborn intensive care, and to under‑record it worst in exactly the babies this finding is about. A rise in the record is therefore either a rise in admissions or an improvement in recording, and this source cannot distinguish the two. Everything below is written so you can see which parts of the finding survive that doubt and which do not.
The question
Newborn intensive care is the most expensive routine service in a maternity hospital, and it is billed by the day. If admissions were being driven by billing rather than by need, the signature would be specific: admission rising among babies who are not sick — born at term, at normal weight, singleton — and rising faster there than among premature babies, whose need is not discretionary.
That signature is what the national birth records show. Whether the cause is billing, changing practice, or changing paperwork is a separate question this figure does not answer, and the section on limitations explains why.
What was measured
Every US birth certificate since the 2003 revision carries a checkbox for admission to NICU. CDC publishes it, tabulated, through CDC WONDER. Because the 2003 revision only reached all 50 states and DC in 2016, the series starts in 2016 — so no part of the trend comes from new states joining the file partway through.
Cohorts
Each cohort is a filter applied inside CDC WONDER, not a group re‑derived afterwards. Holding gestational age, birth weight and plurality fixed is what keeps a shift in the mix of babies from masquerading as a shift in admission.
All newborns — no restriction.
Born early — obstetric‑estimate gestational age under 37 weeks.
Full‑term, normal weight — 37 weeks or more, birth weight 2,500–3,999 g, and a single birth.
Strictest low‑risk — as above but 39–40 weeks only.
Rate
Rate = babies recorded Yes ÷ (recorded Yes + recorded No), × 100. Certificates where the box was left blank are reported separately and are excluded from the denominator, so blank forms cannot inflate or deflate the rate. Blanks are small and not trending: 0.11% of all births in 2016 and 0.19% in 2024.
The numbers
Admission to newborn intensive care as recorded on the US birth certificate, per 100 babies with the box filled in. Source: CDC WONDER, Natality 2016–2024 expanded (dataset D149), variable "NICU Admission"; pulled 8 August 2026. 33,511,275 births across all nine years.
Cohort
2016
2024
Change
Babies (2024)
All newborns
8.74%
9.88%
+13.1%
3,621,981
Born early (under 37 weeks)
49.06%
51.52%
+5.0%
376,805
Full‑term, normal weight, single
3.84%
4.50%
+17.2%
2,857,670
Strictest low‑risk (39–40 weeks)
3.26%
3.54%
+8.7%
1,778,741
Held at the 2016 rate, the 2024 low‑risk cohort would have produced 109,729 admissions. It produced 128,560 — about 18,800 more than the old rate predicts. That figure is an estimate: it is arithmetic on the two measured rates, not a separately observed count, and it inherits every limitation below.
Does it hold up?
It is not one or two states
Measured state by state on the low‑risk cohort, the rate rose in 46 of 51 jurisdictions between 2016 and 2024 and fell in five (Massachusetts, Delaware, Idaho, Minnesota and New York). A national artifact created by one or two large states re‑coding their forms would not look like this.
It is not confined to one payer
On the low‑risk cohort the rate rose for every payer, and it rose most for privately insured births — the ones that pay the most per NICU day.
Low‑risk cohort only, by expected source of payment for the delivery as recorded on the birth certificate. Source: CDC WONDER dataset D149, variable "Source of Payment for Delivery".
Payer
2016
2024
Change
Private insurance
3.63%
4.46%
+22.7%
Medicaid
4.19%
4.69%
+12.1%
Self‑pay
3.09%
3.38%
+9.5%
Other
3.57%
4.43%
+24.1%
This is suggestive and it is not proof. Payer here is the payer for the delivery, not for the newborn's stay, and the payer mix itself shifted over the period.
The internal arithmetic checks out
The 51 state figures sum to the published national figure exactly, for all nine years and both cohorts. The rate is computed from counts held in the database, not copied from a summary.
Limitations
1. The birth certificate under‑records NICU admission — this is the serious one
Linking birth certificates to hospital discharge records for 2,470,917 births in Pennsylvania and South Carolina, 2006–2018, Hughes and colleagues found that 38% of babies genuinely admitted to a NICU had no NICU admission recorded on the birth certificate, and that birth certificates missed more than half of NICU admissions among term infants, including babies with conditions considered severe.
Two consequences, and they cut differently:
The levels here are floors. True admission among low‑risk babies is higher than 4.50%, probably much higher. Nothing on this page should be read as the real rate.
The trend is only a trend in admissions if recording quality held steady from 2016 to 2024. If hospitals got better at completing the box, the recorded rate would rise with no change in care — and because the box is missed most often for term babies, better recording would raise the low‑risk cohort fastest. That is the same pattern as the finding. This data source cannot separate the two, and no amount of further slicing of it will.
2. Admission is not severity
The checkbox records that a baby entered a NICU. It does not record why, for how long, or at what level of care. A four‑hour observation admission and a two‑week stay are the same tick.
3. What this figure does not show
Cost. We hold hospital‑level Medicare cost reports, but they carry whole‑hospital margins, not nursery‑versus‑NICU revenue centres. What a NICU day bills against what it costs is not answerable from what is currently in the database.
Whether NICU margins exceed maternity margins in the same hospital. Same reason. It needs the revenue‑centre worksheets of the Medicare cost report, which have not been ingested.
County‑level detail. CDC WONDER suppresses small sub‑state cells on this dataset, so state is the finest grain this source supports.
4. Deliberately not used
The datasets that would settle the recording question — discharge‑level data such as HCUP KID, NIS and SID, and T‑MSIS Medicaid claims — all require a signed data‑use agreement. None were pulled. Confirming whether this rise is admissions or paperwork requires one of them, or a repeat of the Hughes linkage design on recent years.
Confidence
High
That the recorded rate rose 17.2% in the low‑risk cohort and 5.0% among babies born early. Complete national count data, arithmetic verified against the state breakdown.
High
That the pattern is broad — 46 of 51 states, every payer group.
Moderate
That a real increase in low‑risk admissions is part of what this shows. The direction is consistent across every cut, but improved recording is not excluded.
Not established
That billing incentives are driving it. That is a hypothesis this figure is consistent with, not a finding. It would need cost and claims data we have not pulled.
Sources
CDC WONDER, Natality 2016–2024 expanded (dataset D149), variable "NICU Admission", with gestational age, birth weight, plurality and source of payment. Public; standard click‑through only, no data‑use agreement. Pulled 8 August 2026. wonder.cdc.gov/natality-expanded-current.html
Hughes CS, Lorch SA, Schmitt S, Passarella M, Phibbs CS. Birth certificate data substantially misrepresent actual NICU admissions, including among most vulnerable. Journal of Perinatology 2026;46(7):1307–1312. doi:10.1038/s41372-026-02726-6 (via PubMed)