Skip to content

Correct CHIP and Medicaid pregnancy income limits and effective dates - #10365

Open
hua7450 wants to merge 7 commits into
mainfrom
fix-chip-pregnant-income-limit-dates
Open

Correct CHIP and Medicaid pregnancy income limits and effective dates#10365
hua7450 wants to merge 7 commits into
mainfrom
fix-chip-pregnant-income-limit-dates

Conversation

@hua7450

@hua7450 hua7450 commented Oct 21, 2026 •
edited
Loading

Copy link
Copy Markdown
Collaborator

Fixes #10467

Summary

Correct seven pregnancy income-limit entries using state plan amendments and state rules. Stored limits include the five percentage point MAGI disregard.

For example, a pregnant Colorado citizen at 263% FPL in 2022, above the Medicaid limit, now passes the 265% CHIP income ceiling instead of being rejected by the old 260% ceiling.

This PR changes parameters, references, YAML tests and its changelog fragment. It contains no formula changes. Rhode Island's existing Medicaid ceiling is retained while the insured and postpartum pathways are addressed in stacked draft PR #10391 (issue #10390).

Parameter corrections

PDF page numbers below are file pages.

Parameter Correction Authority
Kentucky pregnancy CHIP 218% FPL from July 1, 2022 KY-21-0001, approval letter and CS8 #page 2,4; 907 KAR 4:030 §2(1)(e)1, version effective January 12, 2023 #page 1-3
West Virginia pregnancy CHIP 305% FPL from July 1, 2019 WV-19-0006, approval letter and CS8 #page 2,4
West Virginia pregnancy Medicaid 190% FPL from July 1, 2019 WV-19-0004, approval letter and S14 #page 3,14
Colorado standard pregnancy CHIP 265% FPL throughout the modeled period from 2018 CO-13-0015, CS8
Colorado FCEP 265% FPL from January 1, 2025; unavailable in 2024 CO-24-0041, amendment history; CO-25-0044, §3.1.1.1; CO-26-0047, superseding pages table (FCEP authority CO-24-0040-0002, no end date)
New Jersey standard pregnancy CHIP 205% FPL throughout the modeled period from 2018 NJ-13-0018, CS8; DMAHS 1115 amendment request to CMS, February 27, 2020 ("199% and 205% of the FPL with 5% disregard")
Rhode Island standard pregnancy CHIP 258% FPL throughout the modeled period from 2018 RI-13-023, CS8; 210-RICR-30-00-1 §1.5(B), 2018 version; §1.6(B), version effective August 17, 2025

Colorado's rule printed 250% until the amendment effective June 30, 2022 (version effective March 11, 2022), which aligned it with the existing 260% standard, as explained in the official rulemaking notice. October 30, 2023 in the Colorado amendment history applies to a different amendment, not the FCEP start. The December 31, 2026 sunset in the CO-25-0044 approval covers only the section 5121 pre-release services, so the FCEP value stays open-ended.

The parameter descriptions state that the values include the five percentage point disregard of 42 CFR 435.603(d)(4), which 42 CFR 457.315(a) applies to CHIP. KFF's January 2026 table adds the disregard only at each state's highest pregnancy level, so it shows West Virginia Medicaid at 185% and Colorado CHIP at 260%.

Modeling scope

The existing annual formulas read January 1 values. Kentucky's July 2022 change therefore first affects model year 2023, and West Virginia's July 2019 changes first affect 2020. This PR preserves the sourced effective dates and documents that annual-model limitation.

The shared pregnancy CHIP formula's existing insurance and postpartum limitations are outside this parameter-only change. General CHIP postpartum work is tracked in #10362; Rhode Island's Medicaid/CHIP routing and postpartum continuation are tracked in #10390.

Merge order: #10241, then this PR, then stacked #10391. Whichever of #10241 and this PR merges second keeps both appended blocks in is_pregnant_for_medicaid.yaml. #10363 also appends to is_chip_eligible_standard_pregnant_person.yaml and renumbers its cases after this PR's.

Tests

The three test files pin each corrected limit on both sides, in 2026 unless the table gives another year (single pregnant applicant, household of 2):

Limit Covered Not covered
West Virginia pregnancy CHIP (305%) 3.05 3.06
West Virginia pregnancy Medicaid (190%) 1.90 1.91
New Jersey pregnancy CHIP (205%) 2.05 (2022) 2.06
Rhode Island pregnancy CHIP (258%) 2.58 (2022) 2.59
Kentucky pregnancy CHIP (218%) 2.18 (2023) 2.19
Colorado FCEP (265%) 2.65 2.66

Other cases cover Kentucky's missing group in 2021 and 2022, West Virginia's in 2019, Colorado at 265% in 2019, and Colorado FCEP in 2024 (no CHIP category) and 2025 (PREGNANT_FCEP). Each person in the multi-state cases this PR adds (Cases 9, 11 and 13) has their own tax unit, SPM unit, family and marital unit.

Cases 21-22 in is_chip_eligible_standard_pregnant_person.yaml add a West Virginia hand-off pair with computed income. In 2020, a pregnant applicant earning $32,760 a year ($2,730 a month, the whole-dollar Medicaid ceiling for a household of 2) stays in Medicaid; at $32,772 she moves to standard pregnancy CHIP.

Validation

  • Merged the latest upstream/main with no conflicts.
  • make format (ruff 0.15.5) passed.
  • At head a77d2cce9a, CI passed 32 of 32 checks, and the reviewer reported that all 63 cases in the three PR test files pass locally. Commit c0d43bcc92 adds the reviewer's West Virginia hand-off pair (Cases 21-22 in is_chip_eligible_standard_pregnant_person.yaml), which passed in his run at that head; CI checks it at the new head.
  • No partner contract test files were edited.

axiom: TheAxiomFoundation/rulespec-us#1924 queued

@hua7450 @claude
Correct CHIP and Medicaid pregnancy income limit dates from state pla…
ab10812 
…n amendments

Kentucky's CHIP pregnancy group starts July 1, 2022 at 218% FPL (KY-21-0001,
907 KAR 4:030). West Virginia's CHIP group and 185% Medicaid standard start
July 1, 2019 (WV-19-0006, WV-19-0004). Colorado, New Jersey and Rhode Island
store the disregard-inclusive CHIP limits from 2018, Colorado's FCEP option
starts January 1, 2025, and Rhode Island's Medicaid pregnancy limit stays at
195% FPL.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@hua7450 hua7450 mentioned this pull request Oct 22, 2026
Open
@hua7450 hua7450 changed the title Correct CHIP pregnancy income limit dates from state plan amendments Correct CHIP and Medicaid pregnancy income limits and effective dates Oct 22, 2026
@hua7450 hua7450 mentioned this pull request Oct 22, 2026
Draft
@hua7450
hua7450 marked this pull request as ready for review October 22, 2026 21:44
This was referenced Oct 23, 2026
Open
Open
PavelMakarchuk
PavelMakarchuk requested changes Oct 23, 2026
View reviewed changes

@PavelMakarchuk PavelMakarchuk left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Program Review

PR #10365, "Correct CHIP and Medicaid pregnancy income limits and effective dates" (hua7450). Round 1, full review.

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

Paths: CP = policyengine_us/parameters/gov/hhs/chip/pregnant/income_limit.yaml, CF = policyengine_us/parameters/gov/hhs/chip/fcep/income_limit.yaml, MP = policyengine_us/parameters/gov/hhs/medicaid/eligibility/categories/pregnant/income_limit.yaml, S = policyengine_us/tests/policy/baseline/gov/hhs/chip/is_chip_eligible_standard_pregnant_person.yaml, F = policyengine_us/tests/policy/baseline/gov/hhs/chip/is_chip_eligible_fcep.yaml, M = policyengine_us/tests/policy/baseline/gov/hhs/medicaid/eligibility/categories/is_pregnant_for_medicaid.yaml. Line numbers refer to the reviewed head. Every PDF link uses the file page number, not the printed page number.

Source Documents

Branch Status

  • The branch is 3 commits ahead and 14 behind main (12b899c11a), and it is MERGEABLE. The staleness did not affect any finding.
  • Merge order: PR #10241 (same author) also edits is_pregnant_for_medicaid.yaml (M) and appends cases at the end of the file. Whichever PR merges second must rebase and keep both blocks.
  • Stacked draft PR #10391 uses this branch as its base and depends on the RI CHIP value 2.58 from 2018. Merge this PR first.

What is correct

All seven corrected entries match their sources. Kentucky has no CHIP pregnancy group until July 1, 2022 and 218% (213% plus the five percent cushion) from that date, so removing the 2021 entries is right. West Virginia CHIP 305% and Medicaid 190% start on July 1, 2019. Colorado 265%, New Jersey 205% and Rhode Island 258% apply from 2018, because the 2014 SPAs set 260%, 200% and 253%. Colorado FCEP is off in 2024 and 265% from January 1, 2025. The old 2023-10-30 date belonged to CO-24-0041 amendments 39 and 41, not to FCEP. The "standard plus 5" convention is correct for every changed value. 42 CFR 435.603(d)(4) applies the disregard once in each Title (78 FR 42188), so WV Medicaid 190% and WV CHIP 305% are both right. The convention is used in all three files (51 of 51 Medicaid states and 7 of 7 CHIP pregnancy states match the CMS table of July 1, 2022 plus 5), and no formula applies a second disregard. The Colorado comment about the 250% rule text is right: the 12 versions of 10 CCR 2505-3 from 10/30/2017 to 03/11/2022 print 250%. All 12 #page= anchors and the 4 extra pages listed in titles land on pages that support the value. The mid-year dates follow the repo convention, and the PR body and tests document the January 1 read. All 9 new test cases are correct by hand check. CI and every local run pass.

Critical (Must Fix)

None.

Should Address

SA1. The new tests do not pin most of the corrected limits.

  • Where: S:244-277 (Case 10, KY is tested above the limit only at 2.25), S:309-325 (Case 12, WV only at 2.50), S:198-241 and S:328-361 (Cases 9 and 13, NJ and RI only at or below the limit), M:142-158 (WV only at 1.80), F:537-552 (Case 19, CO FCEP only at 2.30).

  • Evidence: a mutation run on the 52 cases of the three PR test files. 23 of 34 limit mutants pass every case: KY 2.19, 2.23, 2.24; WV CHIP 2.60, 3.00, 3.04, 3.06, 3.10; WV Medicaid 1.85, 1.89, 1.91, 1.95; NJ 2.06, 2.10, 2.20; RI 2.59, 2.63, 2.70; CO FCEP 2.35, 2.60, 2.64, 2.66, 2.70. The survivors include the values without the disregard (WV 3.00, WV Medicaid 1.85, CO FCEP 2.60) and with a double disregard (KY 2.23, CO FCEP 2.70). These are the errors that this PR corrects. A wider run of 569 cases (all of gov/hhs/chip and gov/hhs/medicaid/eligibility/categories, plus the 22 files with a pregnant person in these states, partner files included) also passes every extreme survivor. Only CO CHIP 2.65 is pinned on both sides. The start dates are pinned on both sides. Limits: WV-19-0006 CS8 300%, WV-19-0004 S14 185%, NJ-13-0018 CS8 200%, RI-13-023 CS8 253%, KY-21-0001 CS8 213%, CO-25-0044 §3.1.1.1 260%, each plus 5 points under 42 CFR 435.603(d)(4).

  • Fix: append the 9 cases below. They pass at the head (9 passed). With the existing cases, they kill all 34 mutants. Each case has one person, so the household size is 2 and no default group is shared.

    Append to S (after Case 13):

# West Virginia's band is 300% FPL (WV-19-0006 CS8) plus the 5 percentage point
# disregard. 2026 household of 2 (FPG 21,640): 21,640 x 3.05 / 12 = 5,500.17
# -> $5,501, so 305% qualifies and 306% ($5,518.20) does not.
- name: Case 14, West Virginia pregnancy group covers income at 305% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 3.05
    households:
      household:
        members: [person1]
        state_code: WV
  output:
    is_chip_eligible_standard_pregnant_person: true

- name: Case 15, West Virginia pregnancy group excludes income at 306% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 3.06
    households:
      household:
        members: [person1]
        state_code: WV
  output:
    is_chip_eligible_standard_pregnant_person: false

# 2026 household of 2: 21,640 x 2.05 / 12 = 3,696.83 -> $3,697; 2.06 gives
# $3,714.87.
- name: Case 16, New Jersey pregnancy group excludes income at 206% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.06
    households:
      household:
        members: [person1]
        state_code: NJ
  output:
    is_chip_eligible_standard_pregnant_person: false

# 2026 household of 2: 21,640 x 2.58 / 12 = 4,652.60 -> $4,653; 2.59 gives
# $4,670.63.
- name: Case 17, Rhode Island pregnancy group excludes income at 259% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.59
    households:
      household:
        members: [person1]
        state_code: RI
  output:
    is_chip_eligible_standard_pregnant_person: false

# 2026 household of 2: 21,640 x 2.18 / 12 = 3,931.27 -> $3,932; 2.19 gives
# $3,949.30.
- name: Case 18, Kentucky pregnancy group excludes income at 219% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.19
    households:
      household:
        members: [person1]
        state_code: KY
  output:
    is_chip_eligible_standard_pregnant_person: false

Append to F (after Case 19):

# Colorado FCEP covers up to and including 260% FPL (CO-25-0044 section
# 3.1.1.1) plus the 5 percentage point disregard. 2026 household of 2:
# 21,640 x 2.65 / 12 = 4,778.83 -> $4,779, so 265% qualifies and 266%
# ($4,796.87) does not.
- name: Case 20, Colorado FCEP covers an undocumented pregnant applicant at 265% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: UNDOCUMENTED
        is_medicaid_eligible: false
        medicaid_income_level: 2.65
    households:
      household:
        members: [person1]
        state_code: CO
  output:
    is_chip_fcep_eligible_person: true

- name: Case 21, Colorado FCEP excludes an undocumented pregnant applicant at 266% FPL in 2026.
  period: 2026
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: UNDOCUMENTED
        is_medicaid_eligible: false
        medicaid_income_level: 2.66
    households:
      household:
        members: [person1]
        state_code: CO
  output:
    is_chip_fcep_eligible_person: false

Append to M (flat format, as in the file):

# WV-19-0004 S14 sets 185% FPL; with the 5 percentage point disregard the
# limit is 190%. 2026 household of 2: 21,640 x 1.90 / 12 = 3,426.33 -> $3,427.
- name: Pregnancy at the 190% FPL limit confers eligibility in WV
  period: 2026
  input:
    state_code: WV
    is_pregnant: true
    medicaid_income_level: 1.9
  output:
    is_pregnant_for_medicaid: true

# 21,640 x 1.91 / 12 = $3,444.37 per month, above the $3,427 ceiling.
- name: Pregnancy one point above the 190% FPL limit confers ineligibility in WV
  period: 2026
  input:
    state_code: WV
    is_pregnant: true
    medicaid_income_level: 1.91
  output:
    is_pregnant_for_medicaid: false

SA2. The New Jersey public-notice reference is a dead link (HTTP 404).

  • Where: CP:26-27. The New Jersey row of the PR body table uses the same URL.
  • Evidence: on 2026-10-23 the href returns HTTP 404, also without www and over http. The parent folder also returns 404, so the site was reorganized. Wayback has 23 captures from 2020-10-16 to 2025-10-31, all with the same digest. File page 1 says "199% and 205% of the FPL with 5% disregard" under SPA NJ 13-0018 (Wayback copy, #page=1). It is the only cited NJ source that prints 205%. The value 2.05 stays correct (NJ-13-0018 CS8 200% plus the disregard).
  • Fix, option A (same document, title unchanged). Replace CP:27 with:
      href: https://web.archive.org/web/20251031023321/https://www.nj.gov/humanservices/providers/grants/public/publicnoticefiles/Public%20Notice%20Amendment%20Request%20-%20NJ%201115%2012-23-19%20Final.pdf#page=1

Option B (official live copy; file page 1 has the same sentence). Replace CP:26-27 with:

    - title: New Jersey DMAHS Section 1115 Comprehensive Demonstration Amendment Request to CMS (February 27, 2020)
      href: https://www.medicaid.gov/medicaid/section-1115-demonstrations/downloads/nj-1115-request-pa.pdf#page=1

Make the same change in the PR body.

SA3. The Rhode Island rule reference is an inactive version, and the section has a new number.

  • Where: CP:30-31. The PR body links https://rules.sos.ri.gov/Regulations/part/210-30-00-1?reg_id=9690.
  • Evidence: the cited version (reg_id 9690) was in force from 06/07/2018 to 01/04/2022. The active version (effective 08/17/2025) still prints "Pregnant People | 253% | 258%" under "Income Ceiling with the five percent (5%) disregard", but in §1.6(B) on file page 7. The section moved from §1.5(B) in the 12/26/2024 version. All five later versions keep 253%/258%. So 2.58 is correct for 2018-2026. But for 2022-2026, including the target year, the only cited state source that prints 258% is a rule that is no longer in force, at a section number that no longer has this table.
  • Fix: replace CP:30-31 with:
    - title: 210-RICR-30-00-1 § 1.5(B) (version effective June 7, 2018 to January 4, 2022)
      href: https://risos-apa-production-public.s3.amazonaws.com/EOHHS/REG_9690_20180806185730.pdf#page=6
    - title: 210-RICR-30-00-1 § 1.6(B) (version effective August 17, 2025)
      href: https://risos-apa-production-public.s3.amazonaws.com/EOHHS/REG_13314_20250728104245127.pdf#page=7

In the PR body, link the 2025 S3 PDF, or https://rules.sos.ri.gov/Regulations/part/210-30-00-1, which always shows the active version. The RI key in CF (2.58) can use the same current reference.

Suggestions

G1. State the disregard convention in the three files, and cite the rule. CP:1, CP:7-33, CP:45; CF:1, CF:8-18; MP:1, MP:7-23. The cited Colorado and West Virginia sources print 260%, 300% and 185%, but the stored values are 2.65, 3.05 and 1.90. Only the KY, NJ and RI sources print the +5. The live KFF State Health Facts table, which all three files cite, changed in 2026: it shows WV Medicaid 185% and CO CHIP 260%, so a reader who checks it sees a 5-point mismatch. The values are correct under 42 CFR 435.603(d)(4) (applied to CHIP by 457.315(a)), and the KFF January 2018 and January 2022 tables (cited only in MP) print CO 265%, NJ 205%, RI 258% and WV 190%/305%. Also, CP:45 says "5% FPL disregard", which a reader can take as 5 percent of the limit. The tests and the PR body say "5 percentage point MAGI disregard". Fix:

  1. CP:1: after "as a share of the federal poverty line." add "Values include the five percentage point income disregard of 42 CFR 435.603(d)(4), which 42 CFR 457.315(a) applies to CHIP." Add to the reference: list:
    - title: 42 CFR 435.603(d)(4), five percentage point MAGI income disregard
      href: https://www.law.cornell.edu/cfr/text/42/435.603#d_4
    - title: 42 CFR 457.315(a), MAGI financial methodologies apply to CHIP
      href: https://www.law.cornell.edu/cfr/text/42/457.315#a
    - title: Medicaid and CHIP Eligibility, Enrollment, Renewal, and Cost Sharing Policies as of January 2018 | KFF, Table 4
      href: https://files.kff.org/attachment/Report-Medicaid-and-CHIP-Eligibility-Enrollment-Renewal-and-Cost-Sharing-Policies-as-of-January-2018#page=33
    - title: Medicaid and CHIP Eligibility and Enrollment Policies as of January 2022 | KFF, Table 2
      href: https://files.kff.org/attachment/REPORT-Medicaid-and-CHIP-Eligibility-and-Enrollment-Policies-as-of-January-2022.pdf#page=36
  1. CF:1: add the same sentence, and add the two CFR references.
  2. MP:1: replace with description: Maximum family income for pregnant people to qualify for Medicaid, as a share of the federal poverty line, including the five percentage point income disregard of 42 CFR 435.603(d)(4). Add the 435.603(d)(4) reference.
  3. CP:45: change "plus the 5% FPL disregard" to "plus the 5 percentage point MAGI disregard" (see G2 for the full comment).
  4. Optional, above the KFF State Health Facts reference in each file: # KFF's January 2026 table applies the disregard only at the highest level overall, so in states with CHIP pregnancy coverage it shows Medicaid without the disregard (for example, West Virginia 185%). 42 CFR 435.603(d)(4) applies the disregard in each Title.

G2. Cite the 10 CCR 2505-3 version that printed 250%. CP:16-19, CP:45-47. The comment is correct: every version from 10/30/2017 to 03/11/2022 reads "greater than 185% but not exceeding 250%" (version 10074, #page=5), and the 06/30/2022 version reads 195%/260%. But no cited page shows the old text or the June 30, 2022 date. The date is in the editor's notes on file page 37 of the cited version. Fix: add after CP:17:

    - title: 10 CCR 2505-3 § 110.1.F (version effective March 11, 2022, before the June 30, 2022 amendment)
      href: https://www.coloradosos.gov/CCR/GenerateRulePdf.do?ruleVersionId=10074&fileName=10%20CCR%202505-3#page=5

Optional: change the CP:16 title to "10 CCR 2505-3 § 110.1.F and editor's notes #page 9,37" (CLAUDE.md:154), and replace CP:45-47 with:

  # 260% FPL (CO-13-0015, effective 2014) plus the 5 percentage point MAGI
  # disregard. Until the amendment effective June 30, 2022, 10 CCR 2505-3
  # § 110.1.F printed "greater than 185% but not exceeding 250%"; the amendment
  # aligned the rule with the existing standard, so 2018 to mid-2022 also uses
  # 2.65.

G3. Say at the Colorado FCEP key that FCEP has no end date. CF:59-67, references CF:15-18. The CO-25-0044 approval "will sunset on December 31, 2026" (#page=2), but the companion letter limits the sunset to the approval "to provide pre-release services" (#page=3). The FCEP authority is CO-24-0040-0002 (CS9), effective January 1, 2025, with no end date. CMS does not publish that SPA (HTTP 404), but the CO-26-0047 superseding table of August 2026 still lists it (#page=5). The open-ended 2.65 is correct. Do not add 2027-01-01: -.inf. A note prevents that edit later. Fix: insert after CF:59:

  # FCEP authority is CO-24-0040-0002 (CS9), effective January 1, 2025, with no
  # end date. The December 31, 2026 sunset in the CO-25-0044 approval covers only
  # the section 5121 pre-release services.

and add after CF:18:

    - title: Colorado CHIP State Plan Amendment CO-26-0047, superseding pages table (CO-24-0040-0002 CS9, Coverage From Conception to Birth, effective January 1, 2025)
      href: https://www.medicaid.gov/CHIP/Downloads/CO-26-0047.pdf#page=5

To cite the companion letter, do not add a second CO-25-0044 entry (CLAUDE.md:151). Change the CF:17-18 entry to the title "Colorado CHIP State Plan Amendment CO-25-0044, companion letter and section 3.1.1.1 #page 3,6" with href #page=3. Optional: a 2027 case for Case 19's person (expected true) also guards this.

G4. Assert the downstream result of the Colorado FCEP 2024 change. F:518-535 (Case 18), F:537-552 (Case 19). 2024 is the only Colorado year in which the PR changes a result. The code probes (code_downstream2.py, code_downstream3.py) show that an undocumented or DACA pregnant CO resident at 230% FPL in 2024 goes from PREGNANT_FCEP to NONE, is_chip_eligible goes from true to false, and chip goes from $2,273.76 to $0. PREGNANT_FCEP returns in 2025. No test asserts chip_category for CO, KY, NJ, RI or WV, but F already does so for California (F:357, F:380). Fix: change the Case 18 output to:

  output:
    is_chip_fcep_eligible_person: false
    is_chip_eligible_pregnant: false
    chip_category: NONE
    is_chip_eligible: false

Use true, true, PREGNANT_FCEP, true in Case 19. Run the file after the edit. Also replace the comment at F:518-519:

# Colorado's FCEP population took effect January 1, 2025, up to 260% FPL
# (CO-24-0041 and CO-25-0044), plus the 5 percentage point MAGI disregard.
# The earlier 2023-10-30 date belonged to CO-24-0041 amendment 41, not to FCEP.
# The applicant is undocumented, so standard pregnancy CHIP does not apply and
# FCEP, which has no immigration test, is the only CHIP pathway.

G5. Name the January 1 lag in the test names, and use one term for the WV Medicaid limit. S:198, S:281, M:151. Kentucky's group started July 1, 2022 and West Virginia's July 1, 2019. Cases 9 and 11 are false only because the YEAR variable reads January 1, but their names state it as policy ("before Kentucky's group starts", "does not apply in 2019"). The repo already uses "in annual simulations" for this (policyengine_us/tests/policy/baseline/gov/hhs/medicaid/eligibility/is_medicaid_lawfully_residing_option_eligible.yaml:489). M:151 says "185% FPL standard", but the case tests the stored 190% limit. Fix:

- name: Case 9, pregnancy bands in Colorado, New Jersey and Rhode Island include the disregard, and Kentucky's July 2022 group does not apply in 2022 annual simulations.
- name: Case 11, West Virginia's July 2019 pregnancy group does not apply in 2019 annual simulations, and Colorado excludes income above 265% FPL.
- name: Pregnancy under the 190% FPL limit (185% plus the disregard) confers eligibility in WV

G6. Close two small gaps: Kentucky 2021 and the Case 11 Colorado person. The PR changes KY 2021 from 2.13 to -inf, but only 2022 is tested. The S:279-280 comment says "Colorado's 265% limit already applied in 2019", but the CO person at 2.66 (S:292-296) is ineligible under the old 2.60 and the new 2.65, so the case does not prove it. Fix: append to S (hand-checked, not run; the old file gives the opposite result in both):

# Kentucky's group starts July 1, 2022 (KY-21-0001), so 2021 has no group.
- name: Case 19, Kentucky has no CHIP pregnancy group in 2021.
  period: 2021
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.1
    households:
      household:
        members: [person1]
        state_code: KY
  output:
    is_chip_eligible_standard_pregnant_person: false

# 2019 household of 2: 16,910 x 2.65 / 12 = 3,734.29 -> $3,735, so 265%
# qualifies (the old 2.60 limit gave $3,664).
- name: Case 20, Colorado pregnancy group covers income at 265% FPL in 2019.
  period: 2019
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        is_medicaid_eligible: false
        medicaid_income_level: 2.65
    households:
      household:
        members: [person1]
        state_code: CO
  output:
    is_chip_eligible_standard_pregnant_person: true

G7. Do not share one default tax unit and marital unit across separate households. Cases 9, 10, 11 and 13 (S:198-241, S:244-277, S:281-305, S:328-361) declare only households. policyengine-core then puts all 3-4 people in one tax unit, SPM unit, family and marital unit. CLAUDE.md:216-217 says that a marital unit has 1 or 2 people, and the shared unit makes medicaid_household_size larger than 2. The outcomes do not change today, because each tested point is exactly at the limit or at least 0.01 away. The existing Cases 7-8 use the same pattern. Fix: write single-person cases, as in SA1, or add one tax_units, spm_units, families and marital_units entry for each person.

G8. Add one West Virginia hand-off case with computed income. Every new CHIP case sets is_medicaid_eligible: false, and every new case sets medicaid_income_level. So no case shows the hand-off between the two WV limits that this PR moves to the same date, or the whole-dollar monthly ceiling. The M:95-137 Missouri cases already use employment_income for this. A probe at the head gives WV 2020 Medicaid true at 1.90 and false at 1.91, and CHIP true at 1.91. Fix: add a WV 2020 single-person pair with employment_income: 32_760 ($2,730 a month; Medicaid eligible) and 32_772 (not eligible), with a one-person tax_units entry as in M:108-110. Check the medicaid_category result, and run the cases before you add them.

G9. Explain the 2018 start and the January 1 read at the keys. CP:100-101 (NJ), CP:116-117 (RI), CP:71-73 (KY), CP:136-138 (WV), MP:132-134 (WV). The NJ and RI SPAs are effective January 1, 2014, and the keys use the file's first date, 2018-01-01. Behavior is the same, because the first value applies to earlier years. The CO key has a comment, but NJ and RI do not. The mid-year lag is explained only in tests and the PR body. Fix, for example:

NJ:
  # NJ-13-0018 has applied since January 1, 2014; 2018-01-01 is this file's first date.
  2018-01-01: 2.05
KY:
  2018-01-01: -.inf
  # Effective July 1, 2022 (KY-21-0001). YEAR-defined formulas read January 1,
  # so this value first applies in 2023.
  2022-07-01: 2.18

Use the same RI comment (RI-13-023), and the KY-style comment for WV in CP and MP ("first applies in 2020").

G10. Use one 907 KAR 4:030 link, and give the rule's version date. CP:22-23. The YAML links the engrossed PDF, and the PR body links https://apps.legislature.ky.gov/law/kar/titles/907/004/030/. Both are live and have the same text. The cited version took effect January 12, 2023 (#page=3), after the 2022-07-01 key. KY-21-0001 covers July 1, 2022 to January 11, 2023, so the SPA date is the right key. Fix: use one href in both places. If you keep the PDF, change the title to "907 KAR 4:030 § 2(1)(e)1 (version effective January 12, 2023) #page 1-2".

Investigated and cleared

  • Page lists in four reference titles (CP:20, CP:22, CP:32, MP:22). They follow CLAUDE.md:151-155: one entry per document, the href opens the first page, and the quoted title ends with the page list. Every listed page is correct. Splitting them would break line 151. No change.
  • Disregard applied per Title, and only once. WV Medicaid 1.90 (Title XIX) and WV CHIP 3.05 (Title XXI) are both correct under 435.603(d)(4) and 78 FR 42188. medicaid_income_eligible (policyengine_us/variables/gov/hhs/medicaid/income/medicaid_income_level.py:23-76) only converts the stored ratio to a whole-dollar monthly ceiling. No formula adds or subtracts a disregard.
  • Colorado +5 with no Colorado text that states it. This is not a missing reference. The cited CS8 page corroborates the 260% base, the +5 is the federal rule and the repo convention, and a cited KFF table prints 265%. G1 covers the documentation gap.
  • Colorado FCEP end date. There is none. The CO-25-0044 sunset covers only the pre-release services. Keep 2025-01-01: 2.65 open-ended (G3).
  • Colorado 250% claim. It is confirmed in the 12 rule versions from 10/30/2017 to 03/11/2022 (G2 adds the citation).
  • Kentucky 2021 removal and the 2022-07-01 key. The SPA was submitted June 25, 2021 and is effective July 1, 2022. The CMS table shows N/A as of July 1, 2021, and KFF January 2022 shows no KY group. The 907 KAR version of January 12, 2023 is later than the federal date, and the SPA date is the right key.
  • Mid-year keys with the January 1 read. This is the convention in these files (AK 2024-02-01, ND 2023-10-01, NV 2025-07-01, WA 2024-11-01, DC 2025-10-01), and the PR documents it.
  • Colorado FCEP 2024 effect. Only undocumented and DACA pregnant CO residents change in 2024. This matches the law. -inf gives no NaN and no warning. The 2025 start lines up with OmniSalud's covers_undocumented_children turning false on 2025-01-01.
  • Kentucky child CHIP -inf from 2022-07-01 (policyengine_us/parameters/gov/hhs/chip/child/income_limit.yaml:55-57, outside the diff). It is correct. KY-22-0003-CHIP moves all children to Medicaid expansion CHIP effective July 1, 2022.
  • Household size for a pregnant applicant. medicaid_household_size (tax household plus current_pregnancies) matches 907 KAR 4:030 §2(2), 210-RICR §1.5(A)(1)(d) and 42 CFR 435.603(b).
  • CO-24-0040-0002 approval date (January 15, 2025 against 11/15/25). Both sources give the same effective date, so the stored date does not change.
  • MP:134 1.90 and the unnumbered M case names. Both match the existing style of the file. No change.
  • Severity disputes settled by the verifiers. The Kentucky upper bound is not pinned (2.19 to 2.24 survive), so it is part of SA1. Colorado FCEP 2.65 is also in SA1: the PR adds the line 2025-01-01: 2.65, two references for it and a comment that states the value, and no case proves it. The disregard citation gap is a suggestion (G1), not a critical item. The RI version gap is SA3, not a critical item, because the cited version prints the stored value for its own period.

Out of scope / follow-up

  • Rhode Island Medicaid pregnancy 2.58 from 2023-01-01 (MP:109-111, pre-existing since #4563). No source supports a step on that date, and it leaves the RI separate CHIP band empty from 2023. Issue #10390 tracks it, and stacked draft #10391 removes it. This PR does not change any 2023+ result.
  • No coverage test in CHIP pregnancy outside California. An insured pregnant teen in WV or CO gets PREGNANT_STANDARD. Issue #10368 tracks this.
  • West Virginia's age 19-55 range. WV-19-0006 CS8 limits the group to ages 19 to 55 (#page=4), but is_chip_eligible_standard_pregnant_person.py:23-55 has no age test. The date move adds model year 2020 to this small gap. No open issue covers it. Open a low-priority issue: "Apply West Virginia's age 19-55 range to standard pregnancy CHIP (WV-19-0006 CS8)". KY and CO select "no age restriction".
  • CHIP postpartum coverage. Issue #10362 tracks it.
  • Other states' FCEP series. 48 of 51 keys end at 2024-01-01. The last value carries forward, and changes in 2025-2026 were not checked. The KY, NJ and WV FCEP -inf values have no positive source, but no source contradicts them.
  • Metadata. CP and MP have no metadata.period: year, and the MP description does not follow the template (G1 item 3 fixes the description).
  • Other pre-existing timing. ND Medicaid pregnancy reads 1.52 until 2021-01-01, but the CMS table shows 157% as of October 1, 2020. Indiana 2018 is 2.13 in the repo and 218% in KFF January 2018 (state-specific disregard).
  • Review tooling. The installed skill (4.0.1) and the Mode B checklist still flag page lists in titles. They will stop when PolicyEngine/policyengine-skills#96 merges.

PDF Audit Summary

Source group Documents read (file pages) Result
CHIP SPAs, KY and WV KY-21-0001 pp.2-5, 11, 21, 28; WV-19-0006 pp.2-5 (scanned); KY-22-0003-CHIP p.2 KY 213% from 2022-07-01; WV 300% from 2019-07-01. Match
Medicaid SPA, WV WV-19-0004 pp.3, 5, 14-16 185% from 2019-07-01. Match
CHIP SPAs of 2014, CO, NJ, RI CO-13-0015 pp.2, 3, 8-10; NJ-13-0018 pp.2, 8; RI-13-023 pp.2, 8, 10 (all scanned, read as images) 260%, 200%, 253% from 2014-01-01. Match
Colorado FCEP CO-24-0041 pp.2, 3, 5; CO-25-0044 pp.2-6; CO-26-0047 p.5; CO-26-0045 p.5 260% from 2025-01-01, no end date. Match
State rules 10 CCR 2505-3 current pp.8, 9, 37 and 14 earlier versions p.5; CO notice p.1; 907 KAR 4:030 pp.1-3; 210-RICR-30-00-1 2018 pp.5, 6, 24 and 2025 pp.4-7 Match. RI cited version inactive (SA3)
NJ notice DMAHS notice p.1 (Wayback); 1115 request to CMS p.1 205% with the disregard. Live link dead (SA2)
Secondary tables KFF January 2018 pp.33-34; KFF January 2022 pp.36-37; KFF State Health Facts (January 2026); MACPAC Exhibit 35 pp.1-2; CMS levels (8 captures); data.medicaid.gov CSV Agree. The live KFF table now omits the disregard below the highest level (G1)
Federal 42 CFR 457.315(a), 435.603(d)(4); 78 FR 42187-42188 Disregard per Title. Supports every +5 value

Tally: 17 values and dates match on page images, 0 mismatches, 0 missing from the repo. 12 of 12 #page= anchors and 4 of 4 title pages are correct. 1 dead link (SA2), 1 inactive rule version (SA3).

Changed entries (7)
Entry Stored Source value and date (file page) Result
KY CHIP pregnant, CP:71-73 -inf to 2022-06-30; 2.18 from 2022-07-01 KY-21-0001 letter p.2, CS8 p.4: 213%, July 1, 2022; 907 KAR 4:030 pp.1-2: +5 cushion Match; upper side unpinned (SA1)
WV CHIP pregnant, CP:136-138 3.05 from 2019-07-01 WV-19-0006 letter p.2, CS8 p.4: 300%, July 1, 2019 Match; value unpinned (SA1)
WV Medicaid pregnant, MP:132-134 1.63, then 1.90 from 2019-07-01 WV-19-0004 letter p.3, S14 p.14: 185%, July 1, 2019 Match; value unpinned (SA1)
CO CHIP pregnant, CP:44-48 2.65 from 2018 CO-13-0015 CS8 p.8: 260%, January 1, 2014; 10 CCR §110.1.F p.9: 260% Match
CO FCEP, CF:59-67 -inf in 2024; 2.65 from 2025-01-01 CO-24-0041 p.3; CO-25-0044 p.6: 260%, January 1, 2025 Match; value unpinned (SA1)
NJ CHIP pregnant, CP:100-101 2.05 from 2018 NJ-13-0018 CS8 p.8: 200%; DMAHS notice p.1: 205% with the disregard Match; link dead (SA2)
RI CHIP pregnant, CP:116-117 2.58 from 2018 RI-13-023 CS8 p.8: 253%; 210-RICR 2018 p.6 and 2025 p.7: 258% Match; inactive version cited (SA3)

Validation Summary

Check Result
CI 27 pass, 3 skipping, 0 fail at 5e0bb06b86
Local tests gov/hhs/chip 152/152; gov/hhs/medicaid/eligibility/categories 244/244; the three PR test files 52/52
Partner contract tests No partner file touched. Read-only run of partners/analytics_coverage/edge_cases/federal/healthcare: 70/70 pass
Regulatory 0 critical. All 7 entries match the SPAs, the state rules and 42 CFR 435.603(d)(4)
References 13 new entries: 12 #page= anchors and 4 title pages correct; 1 dead link (SA2); 1 inactive rule version (SA3); disregard rule not cited (G1)
Code patterns No Python in the diff. No duplicate keys, ascending dates, two-decimal values. The changelog fragment is top-level, typed fixed, and named after the branch
Tests 9 new cases (17 assertions), all correct by hand check. Every changed state and year except KY 2021 has a case that fails under the old values (G6)
Mutation 23 of 34 limit mutants pass the 52 PR cases. A wider run of 569 cases catches none of the extreme survivors. The 9 SA1 cases pass at the head and kill all 34
Probes January 1 reads: KY -inf in 2022, 2.18 in 2023; WV CHIP -inf in 2019, 3.05 in 2020; WV Medicaid 1.63 in 2019, 1.90 in 2020; CO FCEP -inf in 2024, 2.65 in 2025. 2026 values do not change
Branch 3 ahead / 14 behind main; MERGEABLE. Overlap with #10241 in M; #10391 is stacked on this branch
Other reviews None on the PR

Review Severity: REQUEST_CHANGES

No critical item is open. Three should-address items (SA1, SA2, SA3) are open.

Next Steps

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

  1. Apply SA1: append the 9 pre-run cases to S, F and M.
  2. Apply SA2: replace the NJ notice href in CP and in the PR body.
  3. Apply SA3: replace the RI reference in CP and update the PR body link.
  4. Take the suggestions you agree with. G1 removes the most reader confusion. G4 and G6 are test YAML. G2, G3, G9 and G10 are reference or comment edits.
  5. Run make format, then the three PR test files and the gov/hhs/chip and gov/hhs/medicaid/eligibility/categories folders.
  6. Merge this PR before #10391. PR #10241 also appends to M, so the PR that merges second must rebase and keep both blocks.
  7. Not assessed by the review delegates: the PR body has no axiom: line. The repo CLAUDE.md section "Axiom Parity (required for policy changes)" asks for one line on every policy change: axiom: <legal id> encoded-correct | <rulespec PR> encoded | <rulespec issue> queued | n/a: <reason>. An external contributor who cannot do the Axiom work uses axiom: needed. Add the line when you update the body.

hua7450 and others added 2 commits October 23, 2026 22:24
@hua7450 @claude
Fix issues from review: boundary tests, New Jersey and Rhode Island s…
300d6bf 
…ources

- Add boundary cases that pin the corrected WV CHIP, WV Medicaid, NJ, RI,
  KY and Colorado FCEP limits on both sides, plus KY 2021 and CO 2019.
- Replace the dead New Jersey notice link with the medicaid.gov 1115
  amendment request and add the current Rhode Island rule version.
- State the five percentage point disregard (42 CFR 435.603(d)(4)) in the
  descriptions, cite the older Colorado rule text and CO-26-0047, and
  explain the January 1 read and the Colorado FCEP end date.
- Assert the Colorado FCEP downstream CHIP result, rename three cases, and
  give each person in Cases 9-11 and 13 their own units.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@hua7450
Merge remote-tracking branch 'upstream/main' into fix-chip-pregnant-i…
a77d2cc 
…ncome-limit-dates
PavelMakarchuk
PavelMakarchuk approved these changes Oct 26, 2026
View reviewed changes

@PavelMakarchuk PavelMakarchuk left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Program Review

PR #10365, "Correct CHIP and Medicaid pregnancy income limits and effective dates" (hua7450). Round 2, incremental review of the changes since my round-1 review.

IDs: C = critical, SA = should address, G = suggestion. The round-1 IDs (SA1-SA3, G1-G10) keep their numbers. New items start at G11.

Paths: CP = policyengine_us/parameters/gov/hhs/chip/pregnant/income_limit.yaml, CF = policyengine_us/parameters/gov/hhs/chip/fcep/income_limit.yaml, MP = policyengine_us/parameters/gov/hhs/medicaid/eligibility/categories/pregnant/income_limit.yaml, S = policyengine_us/tests/policy/baseline/gov/hhs/chip/is_chip_eligible_standard_pregnant_person.yaml, F = policyengine_us/tests/policy/baseline/gov/hhs/chip/is_chip_eligible_fcep.yaml, M = policyengine_us/tests/policy/baseline/gov/hhs/medicaid/eligibility/categories/is_pregnant_for_medicaid.yaml. Line numbers refer to the reviewed head. Every PDF link uses the file page number, not the printed page number.

Source Documents

Branch Status

  • The branch is 5 commits ahead and 31 behind main (0e7ced1747), and it is MERGEABLE. The staleness did not affect any finding. The merge of main into the branch did not touch the PR files.
  • Merge order: PR #10241 (same author, open) also appends cases at the end of M. Whichever PR merges second must rebase and keep both blocks.
  • Stacked draft PR #10391 uses this branch as its base and depends on the RI CHIP value 2.58 from 2018. Merge this PR first.

What round 2 fixed

Thank you for the careful update. Round 2 fixes all three should-address items and every suggestion except G8.

  • Tests (SA1). All 9 proposed cases are in the files word for word (S Cases 14-18, F Cases 20-21 and the two WV cases at M:160-179). Every corrected limit is now pinned on both sides. A mutation run kills 30 of 30 limit mutants: each changed limit at +-0.01 and +-0.05, which includes the values without the disregard and with a double disregard, plus the two -inf entries. In round 1, 23 of 34 mutants survived. All 34 round-1 mutants are now killed. A revert of any entry to main (7 of 7), or a start-date move that changes the January 1 result (7 of 7), also fails a case.
  • Sources (SA2, SA3). The New Jersey entry now cites the live medicaid.gov copy of the 1115 request. File page 1 prints "199% and 205% of the FPL with 5% disregard". The Rhode Island entries cite the 2018 rule version with its date range and the active August 17, 2025 version at § 1.6(B), file page 7.
  • Documentation (G1, G2, G3, G9, G10). The three descriptions state the disregard and cite 42 CFR 435.603(d)(4). The CHIP files also cite 457.315(a). The KFF comment explains the 5-point gaps in the 2026 KFF table. The Colorado rule history, the FCEP "no end date" note and the effective-date comments at the five keys are in place.
  • Test structure (G4-G7). F Cases 18-19 assert the downstream CHIP result, three test names now state the January 1 read and the 190% limit, KY 2021 and CO 2019 have cases, and the multi-state cases give each person their own units.
  • Checks. 15 of 15 new or changed references and 11 of 11 comments with factual claims are correct. I read every PDF page as an image and the two Cornell anchors as HTML. The three PR test files pass 63 of 63 cases. CI passes 32 of 32 checks. A read-only run of the 70 partner healthcare tests passes.

Round-1 status

ID Status Evidence
SA1 RESOLVED All 9 cases added word for word (S:475-568, F:563-599, M:160-179). 30 of 30 limit mutants are killed. The "Tests" table in the PR body is accurate.
SA2 RESOLVED CP:40-41 uses option B. File page 1 is dated February 27, 2020 and prints 205% with the disregard. The live href works, and the PR body uses the same href.
SA3 RESOLVED CP:44-47 cite the 2018 version ("June 7, 2018 to January 4, 2022") and the active August 17, 2025 version, § 1.6(B), page 7 (253%/258%). The PR body links both. The optional CF RI reference is not added, which is fine.
G1 RESOLVED Items 1-5 are applied (CP:1, CP:18-25, CP:61, CF:1, CF:19-22, MP:1, MP:22-23, and the KFF comment in all three files). The 457.315(a) clause is in a reference title, not in the description. That is acceptable.
G2 RESOLVED CP:30-31 cites version 10074 #page=5 (the quote is verbatim). The CP:28 title ends "#page 9,37". The CP:61-65 comment is the proposed text.
G3 RESOLVED The CF:70-72 comment, the CF:27-28 CO-26-0047 reference and the CF:25-26 retitle with #page=3 are correct. The optional 2027 case is not added. See the optional note after G13.
G4 RESOLVED F Case 18 asserts false, NONE and false (F:537-541). Case 19 asserts true, PREGNANT_FCEP and true (F:557-561). A revert of CO FCEP to main fails Case 18.
G5 RESOLVED S:198, S:345 and M:151 use the proposed names.
G6 RESOLVED S Cases 19-20 (S:570-604) are added. They fail under the pre-PR KY 2021 and CO 2019 values.
G7 RESOLVED Cases 9, 10, 11 and 13 give each person one tax_units, spm_units, families and marital_units entry. Every tested person has medicaid_household_size 2. Cases 7-8 are outside the diff and still share default units.
G8 STILL OPEN No change and no reply. No test uses dollar income for the WV hand-off from Medicaid to CHIP. Details below.
G9 RESOLVED Comments at CP:91-92 (KY), CP:121-122 (NJ), CP:139-140 (RI), CP:162-163 (WV) and MP:140-141 (WV). The five dates match the SPA approval letters.
G10 RESOLVED The CP:36 title gives the version date, and the PR body uses the same engrossed-PDF href. G13 adds the page that prints the date.

Tally: 12 RESOLVED, 1 STILL OPEN (G8, a suggestion), 0 PARTIAL, 0 DECLINED.

Critical (Must Fix)

None.

Should Address

None. All three round-1 items are resolved.

Suggestions

G8 (round 1, still open). Add one West Virginia hand-off pair with computed income. Every new CHIP case sets is_medicaid_eligible: false, and every new case sets medicaid_income_level. So no case shows the hand-off between the WV Medicaid limit (1.90) and the WV CHIP limit (3.05), which this PR moves to the same date. The current cases already pin every value, so this pair adds no new value pin. Its value is the computed path: dollar income to ratio, the unborn child in the household size, the whole-dollar monthly ceiling, and the Medicaid-before-CHIP order with is_medicaid_eligible computed. The pair also fails under the pre-PR WV values and at WV Medicaid 1.89 and 1.91. It is fine to leave it out. If you add it, append to S after Case 20 (the pair passes at the head):

# West Virginia's Medicaid pregnancy limit is 190% FPL (185% plus the 5
# percentage point disregard), and its separate CHIP pregnancy band runs above
# it to 305%. Both start July 1, 2019, so 2020 is the first annual year. 2020
# household of 2 (FPG 17,240): 17,240 x 1.90 / 12 = 2,729.67 -> $2,730 a month
# ($32,760 a year). At $2,730 a month the applicant stays in Medicaid; at
# $2,731 a month ($32,772 a year) she moves to standard pregnancy CHIP.
- name: Case 21, West Virginia pregnant applicant at the $2,730 Medicaid ceiling stays in Medicaid in 2020.
  period: 2020
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        employment_income: 32_760
    tax_units:
      tax_unit:
        members: [person1]
    households:
      household:
        members: [person1]
        state_code: WV
  output:
    medicaid_household_size: 2
    is_pregnant_for_medicaid: true
    medicaid_category: PREGNANT
    is_medicaid_eligible: true
    is_chip_eligible_standard_pregnant_person: false
    chip_category: NONE

- name: Case 22, West Virginia pregnant applicant $1 above the Medicaid ceiling moves to standard pregnancy CHIP in 2020.
  period: 2020
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: CITIZEN
        employment_income: 32_772
    tax_units:
      tax_unit:
        members: [person1]
    households:
      household:
        members: [person1]
        state_code: WV
  output:
    medicaid_household_size: 2
    is_pregnant_for_medicaid: false
    medicaid_category: NONE
    is_medicaid_eligible: false
    is_chip_eligible_standard_pregnant_person: true
    chip_category: PREGNANT_STANDARD

G11. Update the PR body line that says the new cases were not run. The third "Validation" bullet says: "The 11 cases added in the review round were checked by hand and not run locally; CI validates them." They pass. The three PR test files pass 63 of 63 cases locally at the head, and CI passes 32 of 32 checks. Fix: replace the bullet with "All 63 cases in the three PR test files pass locally and in CI."

G12. Give the MP:1 description a verb. CP:1 and CF:1 use the template form "State governments limit ... to these incomes". MP:1 is a noun phrase: "Maximum family income for pregnant people to qualify for Medicaid, ...". This is my own round-1 wording (G1 item 3), so the gap is mine. Optional fix:

description: State governments limit Medicaid pregnancy eligibility to these incomes, as a share of the federal poverty line, including the five percentage point income disregard of 42 CFR 435.603(d)(4).

G13. Add file page 3 to the 907 KAR page list. CP:36 reads "907 KAR 4:030 § 2(1)(e)1 (version effective January 12, 2023) #page 1-2". § 2(1)(e)1 is on pages 1-2, but "eff 1-12-2023" is printed only on file page 3. This also comes from my round-1 G10 text. Fix: change the title to "907 KAR 4:030 § 2(1)(e)1 (version effective January 12, 2023) #page 1-3". The href does not change.

Optional, not counted (G3 follow-through). The CF:70-72 comment guards the open-ended Colorado FCEP value, but no test does: a 2027-01-01: -.inf edit passes all 63 PR cases. If you want a test guard, append to F after Case 21 (it passes at the head; the ratio input is far below the limit, so the projected 2027 FPG does not matter):

# Colorado's FCEP authority (CO-24-0040-0002, CS9) has no end date; the
# December 31, 2026 sunset in the CO-25-0044 approval covers only the section
# 5121 pre-release services. So FCEP still applies in 2027.
- name: Case 22, Colorado FCEP still covers an undocumented pregnant applicant in 2027.
  period: 2027
  input:
    people:
      person1:
        age: 30
        is_pregnant: true
        immigration_status: UNDOCUMENTED
        is_medicaid_eligible: false
        medicaid_income_level: 2.3
    households:
      household:
        members: [person1]
        state_code: CO
  output:
    is_chip_fcep_eligible_person: true

Investigated and cleared

  • No value or date changed in round 2. A YAML parse of the three parameter files at both heads gives equal series for all 51 states, and equal non-reference metadata.
  • Exact-limit cases. The ratio inputs are float32 (2.65 is stored as 2.6500001), so the cases at the limit pass because of the whole-dollar monthly ceiling in medicaid_income_eligible. This is the documented behavior of the helper, not a test defect. All 20 hand-check rows (the 11 new cases and the changed S Cases 9-13 and F Cases 18-19) match the model.
  • One case for each side of each limit. Each side rests on one case. That is enough.
  • The -inf pins. S Case 19, S Case 11 (WV) and F Case 18 prove only that the limit is below the tested ratio. The realistic regressions (the pre-PR values) all fail a case, so this is not a gap.
  • Start-date moves inside one year. A move to another date after January 1 of the same year gives the same annual result under the January 1 read. No annual test can catch it, and none needs to.
  • KFF comment. I compared the live KFF January 2026 data with the repo's latest values. Every FCEP value matches. The only 5-point gaps are CO CHIP and the Medicaid limits in CO, KY, MO, NJ, VA and WV, all lower levels, as the comment says.
  • Page-list titles (CP:28, CP:34, CP:36, CP:48, MP:28, CF:25). They are quoted, end with the page list, and the href opens the first page (CLAUDE.md:151-155). The two 10 CCR entries and the two 210-RICR entries are different version PDFs, not one document split by page.
  • CO-25-0044 anchor in the PR body. The PR body links #page=6 under the label "§3.1.1.1", which is the right page for that label. The YAML uses #page=3 because its title lists "#page 3,6". Both are correct.
  • Colorado FCEP "no end date" comment. It agrees with CO-25-0044 pages 2-5 (the sunset covers only the section 5121 pre-release services) and with CO-26-0047 page 5 (approved August 1, 2026), which still lists CO-24-0040-0002 CS9 from January 1, 2025.

Out of scope / follow-up

  • Iowa Medicaid pregnancy limit (new; pre-existing, outside this diff). MP:55-56 keeps IA: 2018-01-01: 3.80 (375% plus 5) with no later step, and main has the same entry. IA-25-0019 S28, file page 14 sets the pregnant women standard at 215.00% FPL, effective April 1, 2025. The approval letter, file page 2 (March 28, 2025) gives the April 1, 2025 date, and the summary, file page 5 says "from 375% to 215%" under SF 2251. IA-25-0018, file page 14 keeps 375% for January to March 2025, so a 2025-01-01 step would be wrong. With the disregard, the value is 2.20 from 2025-04-01, and KFF January 2026 shows 220%. Iowa has no CHIP pregnancy group (CP IA is -inf). So from model year 2026, the repo passes pregnant Iowans above 220% and up to 380% FPL on the Medicaid income test. Under the January 1 read, annual 2025 stays at 3.80, so tests should use 2026. No open issue or PR covers this. Closed #9183 covered only the Iowa infant and Nevada pregnancy limits, and #10241 is postpartum only. Proposed issue: "Iowa Medicaid pregnancy income limit should drop to 2.20 (215% FPL plus the 5-point disregard) from 2025-04-01 per SPA IA-25-0019". Fix sketch: add 2025-04-01: 2.20 and one reference, "Iowa Medicaid State Plan Amendment IA-25-0019, approval letter and S28 #page 2,14".
  • Iowa infant side note (needs a check). IA-25-0019 file page 10 shows the infant Medicaid standard at 300.00% FPL from April 1, 2025 (3.05 with the disregard). policyengine_us/parameters/gov/hhs/medicaid/eligibility/categories/infant/income_limit.yaml has IA: 2025-01-01: 2.20, set by #9201 (which closed #9183) on the reasoning that infants above 215% move to Hawki. The SPA differs in both the value and the date. The CHIP SPA IA-25-0021 was not read, so check this in the same issue before any change.
  • Rhode Island Medicaid pregnancy 2.58 from 2023-01-01 (MP, pre-existing since #4563). KFF January 2026 shows 190%. Issue #10390 tracks it, and stacked draft #10391 removes it.
  • Carried forward from round 1, unchanged:
    • No coverage test in CHIP pregnancy outside California. Issue #10368 tracks it.
    • West Virginia's age 19-55 range (WV-19-0006 CS8, file page 4). is_chip_eligible_standard_pregnant_person.py has no age test. There is still no issue. Proposed low-priority issue: "Apply West Virginia's age 19-55 range to standard pregnancy CHIP (WV-19-0006 CS8)".
    • CHIP postpartum coverage. Issue #10362 tracks it.
    • Other states' FCEP series. 48 of 51 keys end at 2024-01-01, and changes in 2025-2026 were not checked.
    • Metadata. CP and MP have no metadata.period: year.
    • Other pre-existing timing. ND Medicaid pregnancy (157% from October 1, 2020 in the CMS table) and Indiana 2018 (2.13 in the repo, 218% in KFF January 2018).
    • Review tooling. PolicyEngine/policyengine-skills#96 (page lists in titles) is still open.
References checked in round 2 (15)
Entry Href (file page) Result
CP:18-19, CF:19-20, MP:22-23 42 CFR 435.603(d)(4) Correct. Anchor exists; 5 percentage points for the highest group in each Title
CP:20-21, CF:21-22 42 CFR 457.315(a) Correct. Applies 435.603(b)-(i) to CHIP
CP:22-23 KFF January 2018, Table 4 p.33 Correct. CHIP CO 265%, NJ 205%, RI 258%
CP:24-25 KFF January 2022, Table 2 p.36 Correct. CHIP CO 265%, NJ 205%, RI 258%, WV 305%; WV Medicaid 190%
CP:28-29 10 CCR 2505-3, version 12687 pp.9, 37 Correct. 260% on p.9; "eff. 06/30/2022" on p.37
CP:30-31 10 CCR 2505-3, version 10074 p.5 Correct. "greater than 185% but not exceeding 250%"
CP:36-37 907 KAR 4:030 pp.1-3 Correct. Version date on p.3 only (G13)
CP:40-41 NJ 1115 request p.1 Correct. February 27, 2020; 205% with the disregard
CP:44-45 210-RICR 2018 p.6 Correct. 253%/258%; in force 06/07/2018 to 01/04/2022
CP:46-47 210-RICR 2025 p.7 Correct. § 1.6(B), 253%/258%; effective 08/17/2025
CF:25-26 CO-25-0044 pp.3, 6 Correct. Sunset only for pre-release services; 260% in § 3.1.1.1
CF:27-28 CO-26-0047 p.5 Correct. CO-24-0040-0002 CS9, effective January 1, 2025, no end date

Validation Summary

Check Result
CI 32 pass, 0 fail at a77d2cce9a
Local tests gov/hhs/chip 161/161; gov/hhs/medicaid/eligibility/categories 246/246; the three PR test files 63/63 (52 original, 11 new)
Partner contract tests No partner file touched. Read-only run of partners/analytics_coverage/edge_cases/federal/healthcare: 70/70 pass
Regulatory 0 critical. No stored value or date changed in round 2. The new descriptions and comments agree with 42 CFR 435.603(d)(4), 457.315(a) and the SPAs
References 15 of 15 new or changed entries correct (PDF pages read as images). 11 of 11 comments with factual claims correct. The round-1 anchors are unchanged, except CO-25-0044, which moved from page 6 to page 3 and keeps page 6 in the title
Code patterns No Python. Descriptions, comments and page-list titles follow the repo rules. MP:1 has no verb (G12)
Tests 11 new cases and the changed cases are correct by hand check (20 of 20 rows). Each tested person has medicaid_household_size 2
Mutation 30 of 30 requested limit mutants killed. 14 of 14 reverts and start-date moves killed. All 34 round-1 mutants killed. One CO FCEP 2027 end-date mutant survives (optional G3 case)
Branch 5 ahead / 31 behind main; MERGEABLE. Overlap with #10241 in M; #10391 is stacked on this branch
Other reviews None on the PR except my round-1 review

Review Severity: APPROVE

No critical or should-address item is open. Four suggestions are open (G8, G11, G12, G13), and they are optional.

Next Steps

To apply the optional suggestions automatically: run the fix-pr workflow for this PR.

  1. No change is required before merge. The suggestions are optional.
  2. Optional, PR body only: apply G11 (replace the "not run locally" bullet).
  3. Optional, YAML: apply G12 and G13 (one line each), the G8 pair and the G3 2027 case. Both test blocks ran at the head and pass.
  4. If you change any YAML, run make format, then the three PR test files.
  5. Merge this PR before #10391. PR #10241 also appends to M, so the PR that merges second must rebase and keep both blocks.
  6. Process note, not counted as a finding: the PR body still has no axiom: line. Round 1 asked for it, and it is still missing. The repo CLAUDE.md section "Axiom Parity (required for policy changes)" asks for one line on every policy change: axiom: <legal id> encoded-correct | <rulespec PR> encoded | <rulespec issue> queued | n/a: <reason>. A contributor who cannot do the Axiom work uses axiom: needed. Add the line when you next edit the body.

@PavelMakarchuk

PavelMakarchuk commented Oct 26, 2026

Copy link
Copy Markdown
Collaborator

@hua7450 Process note before merge: the PR body still has no axiom: line.

The repo CLAUDE.md section "Axiom Parity (required for policy changes)" and CONTRIBUTING.md (#axiom-parity) require one line on every policy change. My approval stands, but please add the line before this PR merges. Other reviewers held #10319, #10333 and #10317 until a valid line was added.

Valid forms for this PR:

  • n/a: <reason> does not fit, because the PR changes policy output (the pregnancy and CHIP income limits and their effective dates).
  • axiom: needed is for external contributors only.
  • axiom: <legal id> encoded-correct fits only if a rulespec-us module and a companion case in its .test.yaml cover the same state limits and dates.
  • Otherwise, open a dispatch-ready rulespec-us issue labeled pe-parity for the state limits: the module path, the corpus citation, the verbatim law, the required outputs and companion tests with expected values from the SPAs (not from policyengine-us output). Record the blocker in the issue, then use:
axiom: TheAxiomFoundation/rulespec-us#<issue> queued

For reference, TheAxiomFoundation/rulespec-us#1185 tracks stale state Medicaid/CHIP FPL limits for AK, IA, ND, NV and WA. It has no pe-parity label. You can extend it or open a new issue.

@hua7450 @claude
Apply approving-review suggestions: West Virginia hand-off pair, desc…
c0d43bc 
…ription verb, 907 KAR pages

Add the West Virginia Medicaid-to-CHIP hand-off pair with computed income
(Cases 21-22), give the Medicaid pregnancy limit description a verb, and add
file page 3 (the version date) to the 907 KAR 4:030 page list.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@hua7450

hua7450 commented Oct 26, 2026

Copy link
Copy Markdown
Collaborator Author

Thanks for the approval and the worked suggestions. Commit c0d43bcc92 (pushed right after this comment) applies them:

Item Disposition
G8 Applied. Your West Virginia hand-off pair is now Cases 21-22 in is_chip_eligible_standard_pregnant_person.yaml, with computed income ($32,760 stays in Medicaid; $32,772 moves to standard pregnancy CHIP). I rechecked the arithmetic: 2020 FPG for two is $17,240, and 17,240 x 1.90 / 12 = 2,729.67, which rounds up to $2,730 a month.
G11 Applied in the PR body. The validation bullet now states the CI result at the reviewed head and your local 63-case result, and marks the two new cases as yours.
G12 Applied. The Medicaid pregnancy limit description now reads "State governments limit Medicaid pregnancy eligibility to these incomes, ...".
G13 Applied. The 907 KAR 4:030 title now ends #page 1-3; the href is unchanged.
G3 (optional 2027 FCEP case) Not added. The parameter comment documents the open-ended Colorado value.

Process note: I'm filing a dispatch-ready pe-parity issue in TheAxiomFoundation/rulespec-us for the state limits and dates this PR sets. Once it's filed, I'll add the axiom: ... queued line to the body.

Merge order: #10241 first (it also appends to is_pregnant_for_medicaid.yaml), then this PR, then #10391, which is stacked on this branch.

This was referenced Oct 26, 2026
Open
Open
Open
Open
@hua7450 @claude
Merge upstream main into fix-chip-pregnant-income-limit-dates
ad1ceed 
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
hua7450 added a commit to hua7450/policyengine-us that referenced this pull request Oct 26, 2026
@hua7450 @claude
Fix issues from review: Virginia implementation date and stale comments
407fa2b 
- Case 10 and Case 15 comments give Virginia's FAMIS MOMS extension as
  approved November 18, 2021 and implemented July 1, 2022, matching the
  Medicaid date in PolicyEngine#10241.
- Case 8 comment dates Colorado's FCEP option from January 1, 2025, as
  in PolicyEngine#10365.
- Case 9 comment states the Colorado Medicaid and CHIP pregnancy limits
  before the 5 percentage point MAGI disregard.
- Case 12 also asserts medicaid_category [PREGNANT, NONE].

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This was referenced Oct 26, 2026
Open
Open
Open
Open

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@PavelMakarchuk PavelMakarchuk PavelMakarchuk approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

Correct CHIP pregnancy income limit dates and values (KY, WV, CO, NJ, RI) and the West Virginia Medicaid pregnancy start date

2 participants

@hua7450 @PavelMakarchuk