The letters every plan writes policy against

The California Department of Health Care Services (DHCS) directs Medi-Cal managed care plans through All Plan Letters (APLs), numbered by year and sequence, such as APL 22-012. When DHCS revises a requirement it issues a new letter and states in the title which letter it supersedes. Plans then revise the policies that implement it.

The 17 member plans of the Local Health Plans of California (LHPC) publish their policies on public websites. We collected those documents into a policy knowledge base, extracted every APL reference from their text, and joined each reference to DHCS’s own supersession chain. The result is a citation graph: 1,983 plan documents, 315 letters referenced in them, and 2,525 places where a plan policy cites a letter.

Reading that graph gives a simple measure of how current a policy collection is: what share of its APL citations point to a letter DHCS has since replaced.

One in four citations, and no mention of the replacement

Across all plans, 2,525 citations reference an APL. 937 of them, 37%, cite a letter that DHCS has superseded.

Some of those are lineage references: a policy that names both APL 17-006 and its replacement, APL 21-011, is tracking the requirement’s history, not missing an update. Removing every citation whose document also cites a later letter in the same chain leaves 654 citations, 26% of the total, that name a replaced letter and nothing that came after it.

The range across plans is wide. At CalOptima Health, the largest corpus in the collection, 41 percent of APL citations name a replaced letter and nothing later. At Central California Alliance for Health, 15 percent do, and a further 14 percent sit in documents that also cite the newer letter. Five plans with fewer than 40 citations are omitted from the chart and shown in the table below; their rates are not reliable at that size.

One letter accounts for a fifth of the problem

The stale citations are concentrated. The most-cited replaced letter is APL 22-012, which implemented the Governor’s executive order moving Medi-Cal pharmacy benefits from managed care plans to the state’s Medi-Cal Rx program. DHCS superseded it with APL 25-013. It is still cited 142 times across eight plans, 136 of them without the replacement, more than the next three letters combined.

The rest of the list is the core of Medi-Cal managed care administration: provider credentialing, grievances and appeals, network certification, immunizations, delegation, and provider suspensions. All ten were replaced between 2021 and 2026.

Why the backlog accumulates

DHCS has replaced letters at a faster pace in recent years. Counting supersession events by the year of the replacing letter, the department retired 69 letters from 2021 through the first eight months of 2026, including 20 in 2023 alone, against 53 across the whole of the 2010s. Each replacement obliges every plan to find and revise the policies that cite the old letter, across 17 plans and collections of several hundred documents each.

What this brief does and does not show

• Descriptive, from a public-document corpus. The unit is a citation found in the text of a published policy. Nothing here evaluates whether a policy’s substance matches current DHCS requirements.
• A flag, not a finding. A policy can cite a replaced letter deliberately, as history or because the replacement did not change the cited provision. The old-letter-only measure removes documents that also cite a later letter in the chain, but it cannot read intent.
• Coverage is uneven. The corpus holds what each plan posts publicly and what our collectors could reach on 6 September 2026. CalOptima, Partnership, and Central California Alliance contribute more than 400 documents each and Kern 235; Contra Costa Health Plan, CalViva Health, Santa Clara Family Health Plan, and Health Plan of San Mateo contribute 10 or fewer; and three plans (Contra Costa, CalViva, and Community Health Group) have no citing documents at all. A plan’s rate describes its collected documents, not its full policy library.
• Extraction is automated. APL references are mined from document text with a pattern matcher; supersession is read from DHCS’s own letter titles. Both can miss cases: a letter replaced without a “supersedes” note in the title is counted as current, and a citation written in an unusual form is not counted.
• Point in time. Plans revise continuously. Two letters in the top ten, APL 26-001 and 26-002, replaced their predecessors in 2026, and plans may still be inside their revision window.

Data and sources

The plan policy documents were collected from each plan’s public website; the DHCS All Plan Letters and their supersession statements from the DHCS APL library. Both were indexed into the Causalytics LHPC policy knowledge base on 6 September 2026 and modelled as a graph of plans, documents, letters, implements edges mined from document text, and supersedes edges read from DHCS titles.

The extracted tables are published as CSV: APL citations by plan, most-cited replaced letters, and replacements by year. The charts are drawn from those files when this page is rendered. The export script that produces them from the graph is scripts/export_apl_currency.py in the knowledge base repository.

Send a correction or suggest the next question →

The one group that grew

Each March, the National Resident Matching Program (NRMP) publishes who matched into every residency specialty, broken down by applicant type: senior students at U.S. MD and DO schools, prior-year U.S. graduates, U.S. citizens who trained at medical schools abroad, and non-U.S. citizens who trained abroad. The last two groups are international medical graduates, or IMGs.

Reading the family medicine rows across the five most recent reports produces the chart below.

Between 2022 and 2026 family medicine programs offered 575 more positions, an 11.7 percent increase. Matches rose by only 141. Where those matches came from is the story:

• U.S. MD seniors: 1,541 to 1,492, down 49.
• U.S. DO seniors: 1,494 to 1,403, down 91.
• Prior-year U.S. MD and DO graduates: 181 to 150, down 31.
• U.S. citizen IMGs: 777 to 585, down 192.
• Non-U.S. citizen IMGs: 457 to 962, up 505.

Non-U.S. citizen IMGs were 10.3 percent of matched family medicine positions in 2022 and 20.9 percent in 2026. Unfilled positions still rose from 465 to 899, because the growth in one applicant group did not cover both the new positions and the decline in every other group.

That figure is simple subtraction under a stated assumption. It is not a causal estimate. Other applicants might have ranked family medicine differently if fewer IMGs were competing, and programs might have offered fewer positions. It does show the scale of the dependence.

The applicants filling the gap matched at a five-year low

Match rate is the share of active applicants who matched to a first-year position. Non-U.S. citizen IMGs are the only applicant type whose rate fell in 2026: 56.4 percent, down from 58.0 percent in 2025 and the lowest since 2021. U.S. citizen IMGs reached 70.0 percent, their highest on record, and U.S. DO seniors reached 93.2 percent.

The NRMP’s announcement of the 2026 results adds a split the report tables do not: among foreign-born IMGs, those who required visa sponsorship matched at 54.4 percent, a five-year low, while those who did not require sponsorship, mainly U.S. permanent residents, matched at 67.9 percent, a five-year high. The gap between an applicant who needs a visa and one who does not was 13.5 percentage points.

The report does not say why. Active non-U.S. citizen IMG applicants increased by 479 to 11,944, so a falling rate is consistent with more applicants competing for a similar number of positions. It is also consistent with programs ranking visa-requiring applicants lower. The data here cannot separate those explanations.

A primary care pattern, not a family medicine one

Family medicine is not unusual. Across the 18 PGY-1 specialties that filled at least 200 categorical positions in 2026, the four with the highest share of IMG matches are all primary care as the NRMP defines it: internal medicine, family medicine, primary internal medicine, and pediatrics. Pathology is the one non-primary-care specialty in that range.

Internal medicine is the largest specialty in the Match and the most IMG-dependent: 4,508 of its 10,657 filled categorical positions, 42.3 percent, went to IMGs, and 3,448 of those to non-U.S. citizens. At the other end, orthopaedic surgery filled 963 positions and 10 went to IMGs.

Why this matters

Residency is the only path to licensure for an IMG, and family medicine and internal medicine are the residencies most likely to produce the primary care physicians that shortage-area designations are built around. A physician workforce policy that changes visa availability, J-1 waiver capacity, or the willingness of programs to sponsor is, in practice, a primary care supply policy. The figures above put a number on how much of the current pipeline that touches.

What this brief does and does not show

• Descriptive. These are counts and shares from published tables. Nothing here estimates what caused the shift, and the counterfactual figure is arithmetic under a stated assumption.
• Match, not workforce. Matching into residency is not the same as completing it, obtaining a visa, or practicing in a shortage area. Attrition and post-residency location are outside this data.
• Main Match only. The match rates exclude positions filled through the Supplemental Offer and Acceptance Program (SOAP). Some of the 899 unfilled family medicine positions were filled in SOAP; the NRMP’s report includes SOAP totals but not by specialty and applicant type.
• Citizenship, not visa status. “Non-U.S. citizen IMG” is the NRMP’s category. It includes permanent residents who need no sponsorship. The visa split is available only as two aggregate rates in the NRMP’s announcement.
• Applicant supply moved too. Some of the decline among U.S. DO seniors and U.S. citizen IMGs reflects changes in how many applied and where they ranked, not only program choices.

Data and sources

Every figure in this brief comes from the NRMP’s Results and Data: 2026 Main Residency Match (May 2026): Table 2 for 2026 matches by specialty and applicant type, Table 6 for match rates 2017–2026, and Tables 7A–7F for family medicine positions filled by applicant type 2022–2026. The visa sponsorship rates come from the NRMP’s 2026 Match results announcement (March 2026). The 2022 to 2025 family medicine unfilled counts are derived as positions offered minus positions filled by all applicants. One 2022 match in the NRMP’s small “Others” category (Canadian and Fifth Pathway applicants) explains why the five group changes sum to 142 while total matches rose by 141.

The extracted tables are available as CSV files, with the source table recorded on every row: family medicine matches, 2022–2026, IMG share by specialty, 2026, and PGY-1 match rates, 2017–2026. The charts are drawn from those files when this page is rendered.

Send a correction or suggest the next chart →