[hua7450]

Closes #10368

What and why

CHIP covers only people without other health coverage. PolicyEngine applied that test to children in every state but to pregnant people only in California's FCEP (MCAP) pathway. This PR applies has_chip_disqualifying_health_coverage to both pregnancy pathways in every state.

Pathway	Before	After	Source
Standard pregnancy option (CO, KY, MO, NJ, RI, VA, WV)	No coverage test	Excluded if covered	42 U.S.C. 1397ll(d)(2)(C): a targeted low-income pregnant woman meets 1397jj(b)(1)(C) "in the same manner as a child"
Unborn-child (FCEP) option, 24 jurisdictions besides CA from 2026 (23 before DC's option started in October 2025)	No coverage test	Excluded if covered	42 CFR 457.10 ("child" includes "the period from conception to birth"); 42 CFR 457.310(b)(2)
FCEP in California (MCAP)	Excluded if covered	Unchanged	WIC §§ 15833-15834

Whose coverage counts for the unborn child. The 2002 unborn-child final rule (67 FR 61956) says the other eligibility criteria "pertain to the unborn child" and that the option is meant to reach people "who are currently uninsured". States test the mother's coverage. For example, Missouri's Show-Me Healthy Babies requires that "The pregnant woman must be uninsured" (MO DSS memo IM-110). PolicyEngine has no unborn-child entity, so the pregnant person's coverage stands in, as the California branch already assumed.

Microsimulation effect

Pregnant people with any coverage in gov.hhs.chip.disqualifying_health_coverage lose CHIP pregnancy eligibility outside California. That list covers employer, Medicaid, Medicare, Marketplace and other direct-purchase coverage, TRICARE, VA, CHAMPVA and other means-tested coverage, plus the aggregate has_health_coverage_other_than_chip input. CHIP enrollment and spending on pregnant people fall in the 27 affected jurisdictions (26 in 2024 and 2025).

Measured in review: in the 2026 default data, CHIP pregnancy eligibility falls from 86,878 to 40,358 weighted people (-46,520, -53.5%; -46,192 in 2025). By chip_category, the standard option falls from 42,831 to about zero and FCEP from 44,047 to 40,357. Pregnancy-category CHIP spending falls by $138.3M, from $255.1M to $116.8M (total CHIP -1.1%). The losses are in Missouri (25,130), Virginia (11,636), West Virginia (6,063), Alabama (3,685) and Colorado (5). 43,564 have current employer coverage (has_esi, CPS NOW_GRP) and 2,956 have has_medicaid_health_coverage_at_interview. Nobody gains. The estimate rests on 17 CPS records, so the state split is noisy. Separately, has_medicaid_health_coverage_at_interview is mapped from CPS NOW_MCAID, which includes CHIP reporters; none of the 2,956 reports CHIP, and PolicyEngine/microcosm#1412 tracks the mapping.

API callers who send has_esi or has_health_coverage_other_than_chip for a pregnant person outside California now get no CHIP pregnancy coverage for that person.

Tests

• is_chip_eligible_pregnant.yaml Case 5: a pregnant New Jersey resident at 200% FPL with employer coverage. New Jersey has only the standard option, so this case isolates it.
• is_chip_eligible_fcep.yaml: the case "The California MCAP insurance exclusion does not reach Texas FCEP" asserted the old California-only behavior. It now asserts that Texas FCEP excludes the insured applicant (and chip_category NONE), at 205% FPL (above Texas Medicaid's 203% pregnancy limit, within the 207% FCEP limit). The comment above the California cases now says the list applies in every state.
• is_chip_eligible_pregnant.yaml Case 6: an insured and an uninsured applicant in Colorado, which offers both options; the insured one gets neither. Case 7: coverage reported through the aggregate has_health_coverage_other_than_chip input excludes a New Jersey (standard) and a Texas (FCEP) applicant.
• ca/dhcs/medi_cal/mixed_households.yaml Case 2, renamed "CHIP no-other-coverage exclusion is applied per person in a mixed-state population": the Texan is now uninsured at 205% FPL and keeps FCEP coverage, so the vector still holds an insured Californian (excluded), an uninsured Texan and an uninsured Californian (both covered). Expected values are unchanged from main.
• A static scan of the YAML suite found no other case with an insured pregnant person outside California. Partner contract tests (tests/policy/baseline/partners/**) have none and are unaffected.

Overlap with other open work

• Extend standard pregnancy CHIP through the postpartum period #10363 (postpartum CHIP) edits is_chip_eligible_standard_pregnant_person.py too. The two edits touch different lines and git merge-file merges them cleanly in either order. Once both merge, the coverage test also applies through the postpartum period, as 1397ll(d)(2) intends.
• The new standard-option case lives in is_chip_eligible_pregnant.yaml because Extend standard pregnancy CHIP through the postpartum period #10363 and Correct CHIP and Medicaid pregnancy income limits and effective dates #10365 both append cases to is_chip_eligible_standard_pregnant_person.yaml. Correct CHIP and Medicaid pregnancy income limits and effective dates #10365 also appends to is_chip_eligible_fcep.yaml; that file merges cleanly with this PR in either order.
• Correct Rhode Island pregnancy Medicaid routing and postpartum coverage #10391 (Rhode Island) edits the same return statement in is_chip_eligible_standard_pregnant_person.py. Whichever PR merges second keeps & ~has_disqualifying_coverage and drops the Rhode Island insurance gate from CHIP, as Correct Rhode Island pregnancy Medicaid routing and postpartum coverage #10391 plans.

Not modeled

• Missouri also denies Show-Me Healthy Babies to a woman who has access to employer insurance with maternity benefits (IM-110). That needs a Missouri rule: Missouri Show-Me Healthy Babies: exclude applicants with access to employer insurance that includes maternity benefits #10476.
• California MCAP's exception for private plans without maternity benefits, or with a maternity-only deductible or copayment above $500 (former 10 CCR 2699.200(b)(1)(G)): California MCAP: keep applicants whose private plan lacks maternity benefits or has a maternity deductible above $500 #10477.
• State-funded pregnancy coverage for insured noncitizens who fail the Medicaid immigration test (for example Washington, WAC 182-505-0115(2)). Where the FCEP limit equals the Medicaid pregnancy limit (WA, IL, MN, CT, MD), such a person now has no modeled coverage: State-funded pregnancy coverage for insured noncitizens who fail the Medicaid immigration test (for example Washington) #10513.
• A plan that gives no reasonable geographic access to care does not count as coverage (42 CFR 457.310(b)(2)(ii)). PolicyEngine has no input for this.

Not changed here

• The coverage list, which children share, also holds Medicare, TRICARE, VA, CHAMPVA and other means-tested coverage. Those are public programs, not a group health plan or issuer coverage under PHSA 2791 (42 U.S.C. 300gg-91), though 42 CFR 457.320(a)(7) lets a state exclude other coverage too. Texas excludes Medicare recipients in Texas Works Handbook D-1210, and California does so by statute and regulation (Welf. & Inst. Code 15833; former 10 CCR 2699.200(b)(1)(F)). Any change to the list affects children too and belongs in its own PR: CHIP disqualifying-coverage list includes public coverage that SSA 2110(b)(1)(C) does not exclude #10475. One effect once Correct Rhode Island pregnancy Medicaid routing and postpartum coverage #10391 merges: a pregnant Rhode Islander whose only coverage is VA health care, between Rhode Island's uninsured Medicaid ceiling and its CHIP ceiling, gets neither CHIP nor Rhode Island's insured Medicaid route.

Test plan

• make format (ruff)
• Changelog fragment in changelog.d/
• Local review: statute, regulation and state manuals checked; changed expectations checked by hand
• Review round 1: 1397ll(d)(2)(C) cited; Cases 6-7 added
• CI passes (tests were not run locally)

axiom: TheAxiomFoundation/rulespec-us#1929 queued

🤖 Generated with Claude Code

[PavelMakarchuk]

Program Review

PR #10371, "Apply the CHIP no-other-coverage test to pregnancy pathways in every state" (hua7450). Round 1, full review.

IDs: C = critical, SA = should address, G = suggestion.

Paths: S = policyengine_us/variables/gov/hhs/chip/is_chip_eligible_standard_pregnant_person.py, F = policyengine_us/variables/gov/hhs/chip/is_chip_fcep_eligible_person.py, P = policyengine_us/parameters/gov/hhs/chip/disqualifying_health_coverage.yaml, TP = policyengine_us/tests/policy/baseline/gov/hhs/chip/is_chip_eligible_pregnant.yaml, TF = policyengine_us/tests/policy/baseline/gov/hhs/chip/is_chip_eligible_fcep.yaml, TM = policyengine_us/tests/policy/baseline/gov/states/ca/dhcs/medi_cal/mixed_households.yaml. Line numbers refer to the reviewed head.

Source Documents

• Reviewed head SHA: 7ed04cf
• Mode: full
• Scope: merge-base diff from 9ee14cdb43: 7 files, 61 insertions and 24 deletions. Two formulas, one parameter description, three test files, one changelog fragment.
• Federal law: 42 U.S.C. 1397ll(d)(2)(C) and (f), 42 U.S.C. 1397jj(b)(1)(C), 42 U.S.C. 1397bb, 42 CFR 457.10, 42 CFR 457.310(b)(2), 42 CFR 457.320(a)(7) and (b)(5), 45 CFR 146.113, 67 FR 61956.
• PDF: Virginia DMAS manual chapter M22, 2026-04-01 (file pages 9-10, read as images); SHO 26-001 (page 28).
• State sources: Texas Works Handbook D-1210, Missouri IM-110, Virginia FAMIS Prenatal, 12VAC30-141-740, WAC 182-505-0115.
• Data: default Populace dataset (populace-us-2024-spm-20260909), 2025 and 2026.

Branch Status

• 4 commits ahead and 19 behind main (99694203db); MERGEABLE. The 19 commits change no CHIP file. One changes the Medicaid postpartum period, which the CHIP formulas reach only through is_medicaid_eligible, an input in every changed test. No finding depends on the staleness.
• #10363 (postpartum CHIP) also edits S. The merge is clean in either order, also with the C1 line placed as shown. After both merge, the coverage test also runs through the postpartum period.
• #10391 (Rhode Island) conflicts in one hunk, the return block of S. Its body plans the resolution: keep & ~has_disqualifying_coverage and drop its Rhode Island insurance gate from CHIP. The PR that merges second applies it (G6).
• #10365 (pregnancy limits) edits TF; the merge is clean, also with the G2 comment. It keeps New Jersey at 2.05 and Texas FCEP at 2.07, so the cases here stay within the limits.
• No open sibling PR touches TP or P, so Cases 6 and 7 below can go into TP without conflict.

What the PR gets right

• The law. 1397ll(d)(2)(C) makes a targeted low-income pregnant woman meet 1397jj(b)(1)(C) "in the same manner as a child". 42 CFR 457.10 defines a child as "an individual under the age of 19 including the period from conception to birth", so 457.310(b)(2) reaches the unborn-child (FCEP) option. The PR applies the test to exactly these two pathways, in every state.
• No new concept. Both formulas reuse has_chip_disqualifying_health_coverage, as the child pathway does (is_chip_eligible_child.py:53-63).
• The mother's coverage as the proxy. The 2002 rule says the criteria "pertain to the unborn child" and target people "who are currently uninsured". Texas ("Pregnant women with any type of private health insurance are not eligible for perinatal coverage"), Missouri and Virginia test the mother. With no unborn-child entity, this is the reading the model can apply.
• Medicaid is unaffected. Probe: New Jersey, 190% FPL, has_esi, Medicaid computed: Medicaid true, CHIP false.
• California is unchanged, and the body is accurate: the before and after table, the 26 (2024-2025) and 27 (2026) affected jurisdictions, and follow-ups #10475-#10477.
• Clean code and pinned tests. No hard-coded values, correct periods and entities, vectorized, no cycles; changelog top-level fixed. Removing the standard test fails TP Case 5; the old California-only FCEP gate fails the Texas case in TF. CI 32 of 32.

Critical (Must Fix)

C1. The statute that applies the coverage test to pregnant women is not in any reference.

• Where: S:13-21 (tuple) and S:49-51 (comment); also P:20-26.
• Evidence: the new term at S:52-54 and S:62 rests on 1397ll(d)(2)(C): a targeted low-income pregnant woman is one "who satisfies the requirements of paragraphs (1)(A), (1)(C), (2), and (3) of section 1397jj(b) ... in the same manner as a child applying for child health assistance would have to satisfy such requirements." 1397jj(b)(1)(C) sets the test for "a child ... who is not found to be eligible for medical assistance under subchapter XIX or ... covered under a group health plan or under health insurance coverage". Only 2112(d)(2)(C) carries it to a pregnant woman. The tuple has SSA 2110, the medicaid.gov levels, 8 U.S.C. 1611 and 1641, 45 CFR 155.20, SHO 12-002 and SHO 26-001 page 28 (immigration). None supports the new term. P cites only child rules, and 42 CFR part 457 has no section that applies the test to pregnant women. The citation is only in the comment.
• Why critical: a missing reference is critical under the review criteria, and CLAUDE.md asks to "cite specific regulation sections in variable reference". SSA 2110 does not cover it, because 2110(b)(1) defines the test for "a child".
• Fix (run at the head; tests pass): in S, add the line directly after the medicaid.gov line. Directly after the 2110 line it conflicts with #10363, which inserts 1397ll#d_2_A there.

        "https://www.medicaid.gov/medicaid/national-medicaid-chip-program-information/medicaid-childrens-health-insurance-program-basic-health-program-eligibility-levels",
        "https://www.law.cornell.edu/uscode/text/42/1397ll#d_2_C",

In P, replace P:20-26 with this block. It also carries G2 and G3, and every anchor exists.

  reference:
    - title: 42 CFR § 457.310(b)(2) - No other coverage standard
      href: https://www.ecfr.gov/current/title-42/section-457.310#p-457.310(b)(2)
    - title: 42 CFR § 457.320(a)(7) - Other eligibility standards
      href: https://www.ecfr.gov/current/title-42/section-457.320#p-457.320(a)(7)
    - title: 42 CFR § 457.320(b)(5) - Other eligibility standards
      href: https://www.ecfr.gov/current/title-42/section-457.320#p-457.320(b)(5)
    - title: 42 U.S.C. § 1397ll(d)(2)(C) - Targeted low-income pregnant woman
      href: https://www.law.cornell.edu/uscode/text/42/1397ll#d_2_C
    - title: 45 CFR § 146.113(a)(1) - Creditable coverage
      href: https://www.ecfr.gov/current/title-45/section-146.113#p-146.113(a)(1)

Should Address

SA1. The changed parameter description uses the acronym "CHIP".

• Where: P:1, which the PR rewrote.
• Evidence: the PolicyEngine parameter guide (policyengine-model-development/references/parameters.md:35) says "description: one sentence, spell the program out fully (no acronyms)". Sibling files spell it out (fcep/income_limit.yaml:1, pregnant/income_limit.yaml:1).
• Why should address: a pattern violation on a changed line.
• Fix (run at the head):

description: Health coverage sources that disqualify a child or a pregnant person from Children's Health Insurance Program eligibility.

SA2. No test covers an insured applicant in a state with both pregnancy options.

• Where: TP:41-65 (New Jersey, standard only) and TF:452-473 (Texas, FCEP only).
• Evidence: Colorado, Missouri, Rhode Island and Virginia offer both options. There, 1397ll(d)(2)(C) closes the standard option and 457.310(b)(2) closes FCEP: the child "must not be— ... Covered under a group health plan or under health insurance coverage". Two wrong formulas pass all 213 related cases (gov/hhs/chip, TM, ca/dhcs/cchip, partner healthcare/chip): a standard test that runs only where the state has no FCEP, and an FCEP test that runs only where the state has no standard option. Missouri and Virginia carry 79% of the 2026 loss.
• Why should address: a missing edge case for variables the PR already tests.
• Fix (passes at the head; fails on both wrong formulas): append to TP.

- name: Case 6, an insured pregnant applicant in a state with both CHIP pregnancy options gets neither option.
  period: 2025
  input:
    people:
      # person1 has employer coverage; person2 is the uninsured control.
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.3
        has_esi: true
      person2:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.3
    tax_units:
      tax_unit1:
        members: [person1]
      tax_unit2:
        members: [person2]
    households:
      household1:
        members: [person1]
        state_code: CO
      household2:
        members: [person2]
        state_code: CO
  output:
    # Colorado offers the standard and FCEP options up to 265% FPL; 230% FPL
    # is above its 200% Medicaid pregnancy limit and within both CHIP limits.
    is_chip_eligible_standard_pregnant_person: [false, true]
    is_chip_fcep_eligible_person: [false, true]
    is_chip_eligible_pregnant: [false, true]
    chip_category: [NONE, PREGNANT_STANDARD]

Suggestions

G1. Outside California, the pregnancy pathways are tested only with has_esi.

• Where: TP:51 and TF:462.
• Evidence: a standard test or a non-California FCEP test that reads only has_esi passes all 213 related cases. The PR body names has_health_coverage_other_than_chip as an API input that now removes coverage, but no pregnancy case outside California sets it.
• Why a suggestion: has_chip_disqualifying_health_coverage.yaml tests the list source by source, so this is a wiring check.
• Fix (passes at the head; fails on both wrong formulas; uninsured controls are eligible): append after Case 6.

- name: Case 7, other coverage reported through the aggregate input excludes pregnant applicants from both CHIP pregnancy options.
  period: 2026
  input:
    people:
      # person1: New Jersey (standard option only); person2: Texas (FCEP only).
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2
        has_health_coverage_other_than_chip: true
      person2:
        age: 30
        is_pregnant: true
        is_medicaid_eligible: false
        medicaid_income_level: 2.05
        has_health_coverage_other_than_chip: true
    tax_units:
      tax_unit1:
        members: [person1]
      tax_unit2:
        members: [person2]
    households:
      household1:
        members: [person1]
        state_code: NJ
      household2:
        members: [person2]
        state_code: TX
  output:
    is_chip_eligible_standard_pregnant_person: [false, false]
    is_chip_fcep_eligible_person: [false, false]
    is_chip_eligible_pregnant: [false, false]
    chip_category: [NONE, NONE]

Optional: the Texas case at TF:452-473 does not assert chip_category, unlike its California neighbors (TF:358 to TF:450). Add chip_category: NONE, the value Case 7 gives for an insured Texan at 205% FPL.

G2. The new comment at TF:336-337 credits the whole list to 42 CFR 457.310(b)(2).

• Evidence: 457.310(b)(2)(ii) excludes only coverage "under a group health plan or under health insurance coverage, as defined in section 2791 of the Public Health Service Act". The list also holds Medicare, TRICARE, VA, CHAMPVA and means-tested coverage. 457.320(a)(7) lets a state adopt standards on "Access to, or coverage under, other health coverage", and Texas ("Do not certify a Medicare recipient for ... CHIP or CHIP perinatal"), California (WIC 15833) and Virginia (M22) use it.
• Why a suggestion: wording only; these entries remove nobody in the 2026 default data.
• Fix (run at the head; merges cleanly with #10365): replace TF:336-337 with the lines below. The 457.320(a)(7) reference is in the C1 block.

# coverage. The CHIP disqualifying coverage list applies to FCEP in every
# state and carries these exclusions: 42 CFR 457.310(b)(2) excludes group
# health plan and health insurance coverage, and 42 CFR 457.320(a)(7) lets a
# state also exclude other coverage, such as Medicare.

G3. Two reference titles in P are wrong.

• Where: P:21 and P:25 (older lines in a touched file).
• Evidence: 457.310 is "Targeted low-income child", and (b)(2) is "No other coverage standard", not "Health benefits coverage". 146.113 is "Rules relating to creditable coverage", and (a)(1) is "Creditable coverage", not "Excepted benefits" (those are in 146.145(b)).
• Why a suggestion: titles only; the hrefs are correct.
• Fix: the C1 block corrects both.

G4. Insured noncitizen pregnant people in Washington now have no modeled coverage.

• Where: F:90-100, where the FCEP limit equals the Medicaid limit, so FCEP reaches only people who fail the Medicaid immigration test.
• Evidence: under WAC 182-505-0115(2), a noncitizen pregnant person "is eligible for apple health pregnancy coverage if they meet countable income standards", with no insurance condition. Title XXI cannot pay for an insured unborn child, so the PR is right for CHIP; the state pays, and the model has no variable for it. Probe, Medicaid computed: Washington, undocumented, 200% FPL, has_esi: Medicaid false, FCEP false (FCEP true without has_esi). Illinois, Minnesota, Connecticut and Maryland give the same model result (state law not checked). Oregon is not affected: the model gives Medicaid there.
• Why a suggestion: no undocumented person is among the 2026 losers (46,515 citizens, 5 DACA).
• Fix: open a tracking issue and add to "Not modeled": "State-funded pregnancy coverage for insured noncitizens who fail the Medicaid immigration test (for example Washington, WAC 182-505-0115(2)). Where the FCEP limit equals the Medicaid pregnancy limit (WA, IL, MN, CT, MD), such a person now has no modeled coverage."

G5. The geographic-access exception is not listed under "Not modeled".

• Evidence: 457.310(b)(2)(ii): "A child is not considered covered under a group health plan or health insurance coverage if the child does not have reasonable geographic access to care under that plan." Virginia (M2220.200 B, 12VAC30-141-740 E.2.a) applies the same idea. No input exists, for children or pregnant people.
• Fix: add to "Not modeled": "A plan that gives no reasonable geographic access to care does not count as coverage (42 CFR 457.310(b)(2)(ii)). PolicyEngine has no input for this."

G6. "Overlap with other open work" does not mention the #10391 conflict.

• Evidence: a three-way merge of S with #10391 (1c78751418) gives one conflict in the return block: & ~has_disqualifying_coverage here against & ri_insurance_eligible ) | person("ri_chip_postpartum_eligible", period.first_month). Item (c) of the #10391 body records the resolution.
• Fix: add: "#10391 (Rhode Island) edits the same return statement in is_chip_eligible_standard_pregnant_person.py. Whichever PR merges second keeps & ~has_disqualifying_coverage and drops the Rhode Island insurance gate from CHIP, as #10391 plans."

G7. The PR body says "The population effect has not been measured."

• Evidence: two separate runs on the default data (merge-base formulas against the head) agree on the numbers below. No default changes, so this is the intended legal effect, not a biased default.
• Fix: replace the sentence with: "In the 2026 default data, CHIP pregnancy eligibility falls from 86,878 to 40,358 weighted people (-46,520, -53.5%; -46,192 in 2025). By chip_category, the standard option falls from 42,831 to about zero and FCEP from 44,047 to 40,357. Pregnancy-category CHIP spending falls by $138.3M, from $255.1M to $116.8M (total CHIP -1.1%). The losses are in Missouri (25,130), Virginia (11,636), West Virginia (6,063), Alabama (3,685) and Colorado (5). 43,564 have current employer coverage (has_esi, CPS NOW_GRP) and 2,956 have has_medicaid_health_coverage_at_interview. Nobody gains. The estimate rests on 17 CPS records, so the state split is noisy."

Investigated and cleared

• Medicaid coverage at interview. We checked whether has_medicaid_health_coverage_at_interview holds Medicaid only, as its label and the microcosm column notes say. It does not. In the default data it equals CPS NOW_MCAID, the union of NOW_CAID, NOW_PCHIP (CHIP) and NOW_OTHMT (10,622 + 400 + 159 records). So all 400 CHIP-reporter records (956,895 weighted) fail the CHIP coverage test; a New Jersey child at 250% FPL with the flag gets no CHIP. The effect on this PR is zero: the 44,244 weighted pregnant CHIP reporters have no CHIP pregnancy category before or after, and none of the 2,956 people who lose through this flag reports CHIP. Pre-existing, not counted (follow-up below).
• Coverage beyond group health plans. Medicare, VA, TRICARE, CHAMPVA and means-tested coverage go beyond the federal floor, but 457.320(a)(7) permits them (G2). They remove nobody in the default data, and #10475 tracks the list.
• Indian Health Service. Not on the list, so 457.320(b)(5) holds for pregnant people.
• Crowd-out. 1397bb(b)(3)(C) is a substitution rule, not the current-coverage test; nothing here depends on it.
• Test placement. Case 5 sits in TP because three sibling PRs append to the standard-option test file; reasonable for now. TM Case 2 still shows per-person application.
• Partner contract tests. None edited. The partner FCEP case (Texas, 2026) still gives chip 1,456.51 and PREGNANT_FCEP.

Out of scope / follow-up

• CHIP reporters excluded from CHIP (pre-existing, not counted). Open a data issue to map has_medicaid_health_coverage_at_interview from NOW_CAID alone. It affects the children's pathway. Related: #8577 (closed), #8495, #10475.
• #10475. Add the 457.320(a)(7) point (G2), so any narrowing is state-specific. Note a Virginia conflict there: M22 counts Medicare as creditable coverage, but 12VAC30-141-740 E.2.b says "Health insurance does not include Medicare, Medicaid, FAMIS" or HIPP-paid insurance.
• #10476 and #10477 (Missouri access to employer coverage; California MCAP maternity exception): already tracked.

Validation Summary

Check	Result
CI	32/32 pass at 7ed04cf7b2; 3 skipped checks are from a superseded run
Local tests	Changed files 30/30; gov/hhs/chip 146/146; ca/dhcs/medi_cal 12/12; ca/dhcs/cchip 47/47
Proposed fixes	All applied together: 232 YAML cases pass. Reference and parameter checks 161 pass; 1 fails the same way on an unchanged head copy (local environment)
Partner contract tests	None edited; partner healthcare folder 70/70 (read-only)
References	1 added at F:25, correct. 1397ll(d)(2)(C) missing (C1); two wrong titles (G3)
Regulatory	Both pathways must apply the test in every state, and the PR does so; the proxy follows 67 FR 61956 and state practice
Code patterns	Clean; description acronym (SA1)
Mutation	13 formula changes: 9 caught; Cases 6 and 7 catch the other 4
Merge trials	#10363 clean; #10391 one conflict on S; #10365 clean
Microsimulation (2026)	-46,520 people (-53.5%) and -$138.3M pregnancy-category CHIP; 17 records; nobody gains
Branch	4 ahead / 19 behind main; MERGEABLE
PR body	axiom: line present (rulespec-us#1929 queued), process note only. Edits in G4-G7
Other reviews	None

Review Severity: REQUEST_CHANGES

One critical item (C1, a missing statute reference) and two should-address items (SA1, SA2) are open; the model change itself is legally correct.

Next Steps

To auto-fix issues: run the fix-pr workflow for this PR.

1. Add the 1397ll(d)(2)(C) line to S after the medicaid.gov line, and replace the references in P (C1, with G2 and G3).
2. Spell out the program name at P:1 (SA1).
3. Append Cases 6 and 7 to TP (SA2, G1).
4. Replace the comment at TF:336-337 (G2).
5. Update the PR body: measured effect (G7), two "Not modeled" bullets (G4, G5), the #10391 overlap (G6).
6. Run make format, then gov/hhs/chip, TM and ca/dhcs/cchip again.
7. Optional, outside this PR: open the data issue for has_medicaid_health_coverage_at_interview and the G4 tracking issue.

[hua7450]

Thanks for the review. Round-1 changes are in 5bd740e289, and 3440a77744 merges main (4c65c82191). Dispositions:

Item	Disposition
C1 1397ll(d)(2)(C) reference	Done. The Cornell 1397ll#d_2_C line now sits after the medicaid.gov line in is_chip_eligible_standard_pregnant_person.py, and the reference block in disqualifying_health_coverage.yaml is replaced with yours. git merge-tree against #10363 (3079297387) and #10365 (ae4dc7a828) is still clean.
SA1 description acronym	Done: "Children's Health Insurance Program".
SA2 both options	Done: Case 6 (Colorado, 230% FPL, insured gets neither option, uninsured control gets the standard option).
G1 aggregate input	Done: Case 7 (New Jersey and Texas, has_health_coverage_other_than_chip), plus chip_category: NONE on the Texas FCEP case.
G2 comment	Done: the FCEP test comment now names 457.310(b)(2) for group health plans and health insurance coverage, and 457.320(a)(7) for other coverage such as Medicare.
G3 reference titles	Done through the C1 block.
G4 insured noncitizens	Done. Filed #10513 and added it to "Not modeled".
G5 geographic access	Done: added to "Not modeled".
G6 #10391 overlap	Done: added to "Overlap with other open work". The merge-tree check still shows the one return-block conflict with #10391 (1c78751418).
G7 measured effect	Done: the body now gives your 2026 figures (-46,520 people, -$138.3M) with the state split and the 17-record caveat.
Out of scope: NOW_MCAID	Filed PolicyEngine/microcosm#1412 (map has_medicaid_health_coverage_at_interview from NOW_CAID alone).
Out of scope: #10475	Commented there with the 457.320(a)(7) point and the Virginia M22 versus 12VAC30-141-740 E.2.b conflict.

TheAxiomFoundation/rulespec-us#1929 now describes this head, with a Colorado companion case and the 457.320(a)(7) point. No tests were run locally; I checked each new expected value by hand against the Colorado, New Jersey and Texas limits.