Clarify -inf sentinel in CHIP income limit parameter descriptions

[daphnehanse11]

Summary

The CHIP income limit parameters (chip/child, chip/pregnant, chip/fcep) use -.inf as a sentinel meaning the state does not operate a separate CHIP program for that category — the eligibility formulas gate on income_limit > 0. Nothing in the parameter files documented this, so the values read as missing data (an API partner recently flagged California's -.inf as an unfilled placeholder).

California, for example, covers children through CHIP-funded Medicaid expansion (M-CHIP) up to 266% FPL — captured in the Medicaid child income limit parameters — and CHIP-funded pregnancy coverage through the FCEP option at 322% FPL, so -.inf in the separate-CHIP files is correct per the CMS eligibility-levels table.

This PR adds one sentence to each of the three parameter descriptions explaining the sentinel and pointing to where the coverage for those states actually lives. No values or logic change.

Changes

• parameters/gov/hhs/chip/child/income_limit.yaml: document that -inf = no separate children's CHIP (M-CHIP states)
• parameters/gov/hhs/chip/pregnant/income_limit.yaml: document that -inf = no standard separate pregnant CHIP (coverage via Medicaid or FCEP)
• parameters/gov/hhs/chip/fcep/income_limit.yaml: document that -inf = state has not adopted the FCEP option
• Changelog fragment

🤖 Generated with Claude Code

[daphnehanse11]

Closing — documentation nice-to-have, not needed. The -inf sentinel behavior is working as intended.

[DTrim99]

Program Review

Source Documents

• No PDF value audit — documentation-only PR (no parameter values changed). Description claims checked against CMS eligibility-levels table and the CHIP eligibility formulas in-code.
• Scope: PR changes only (merge-base diff)

Branch Status

⚠ PR branch is 3031 commit(s) behind main. Consider rebasing before merging. Review was scoped to the PR's actual changes — staleness did not affect findings.

Critical (Must Fix)

None.

The new description sentences were verified as factually accurate about both the code and the policy:

• All three CHIP eligibility formulas gate on income_limit > 0 (is_chip_eligible_child.py, is_chip_eligible_standard_pregnant_person.py, is_chip_fcep_eligible_person.py), so -inf correctly means "state does not operate this separate CHIP category" — exactly as the new descriptions state.
• Every -inf state in the child file (CA, HI, MD, MI, MN, NM, OH) matches the "N/A — no separate CHIP" entries in the KFF table already cited in that file, and the Medicaid child limits reflect the M-CHIP coverage the description points to (e.g., CA 266% FPL, consistent with the PR body).
• The FCEP and pregnant descriptions restate the code gate and are consistent with the KFF/CMS/MACPAC sources already cited.
• Changelog fragment is present, correctly located, named, and typed (changed).

Should Address

1. Child description's M-CHIP claim is a categorical assertion not squarely backed by the cited sources (gov/hhs/chip/child/income_limit.yaml). The first clause ("the state does not operate a separate CHIP program for children") is directly supported by the cited KFF table ("N/A denotes state did not provide separate CHIP coverage"). The second clause — that these states "instead cover children through CHIP-funded Medicaid expansion (M-CHIP)" — is a stronger claim the cited references do not explicitly document (KFF only states the absence of a separate CHIP; the cited CMS eligibility-levels page could not be auto-verified). A state could also lack a separate child CHIP while using a combination program. Additionally, the pointer to "the Medicaid child income limit parameters" is imprecise: there is no single Medicaid child limit — child coverage spans the medicaid/eligibility/categories/{infant,young_child,older_child}/income_limit.yaml parameters. Recommended: either (a) soften to what the sources support (e.g., "children at these income levels are instead covered through Medicaid, reflected in the Medicaid infant/young-child/older-child income limit parameters"), or (b) keep the M-CHIP wording and add a reference documenting the separate-CHIP vs. M-CHIP program-type split (e.g., the CMS program-type listing or MACPAC MACStats program-type exhibit). Non-blocking — no values changed and spot-checks (CA, OH, NM) are consistent with M-CHIP.

Suggestions

1. Prose writes the sentinel as -inf while the YAML literal is -.inf (all three files). A defensible human-readable simplification; optionally align the prose to the exact literal -.inf for precision.
2. Descriptions are now two sentences (all three files). The strict style preference is a single-sentence description:, but multi-sentence descriptions are established precedent in this repo (including the HHS tree) and the added sentence is genuinely useful. No change required — noted for awareness.
3. Pre-existing, out of PR scope: gov/hhs/chip/pregnant/income_limit.yaml metadata lacks the period: key that the child and fcep files have. Not introduced by this PR; flagged only for a possible follow-up. (Separately, this PR's edit incidentally removed pre-existing trailing whitespace on the pregnant description line — a welcome cleanup.)

Validation Summary

Check	Result
Regulatory/factual accuracy of new text	0 issues (claims verified against code gates and KFF/CMS sources)
Reference Quality	1 issue (M-CHIP clause not squarely sourced)
Code Patterns / Changelog	0 issues (changelog fragment correct; 2 style notes)
Test Coverage	N/A — description-only change
PDF Value Audit	N/A — no values changed
CI Status	Passing (all checks)

Review Severity: COMMENT

Next Steps

To auto-fix issues: /fix-pr 9174

---

Automated review generated by /review-program (read-only).

[MaxGhenis]

Merging under the cross-family dual-review protocol (Claude Fable 5 + GPT-5.6-sol, independent reviews, both merge-ok).

Verified: diff is descriptions-only plus a changelog fragment (unified=0 scope check); the -.inf sentinel semantics hold in all three eligibility formulas (state_has_chip = income_limit > 0 in is_chip_eligible_child.py, is_chip_eligible_standard_pregnant_person.py, is_chip_fcep_eligible_person.py); the referenced Medicaid/FCEP parameter trees resolve; the California example is consistent with the repo's existing values (child/pregnant −inf, Medicaid child 2.66, FCEP 3.22).