A March spike in Medicare enrollment deactivations thinned provider supply in shortage areas

Most months, Medicare enrollment deactivations in PECOS are a quiet administrative trickle: retirements, relocations, providers letting a lapsed enrollment expire. March 2026 was not most months. The deactivation count ran 28% above the trailing twelve-month average — a spike large enough to clear our snapshot-to-snapshot alerting threshold on the first pass.

□Figure not yet availableSource: CMS PECOS · snapshot 2026-04

A national spike is interesting. Where it lands is what matters for access.

Mapping deactivations onto shortage areas

We geocoded each deactivated enrollment to its practice location and joined it to HRSA's Health Professional Shortage Area (HPSA) designations. Two findings stood out.

First, the spike was not uniform. Deactivations in already-designated shortage areas grew 41% against trend, against 19% in non-shortage areas. The places with the least slack lost the most supply.

Second, the deactivations were concentrated among primary-care and behavioral- health enrollments — exactly the specialties that drive HPSA scoring in the first place.

A deactivation in a saturated metro specialty market is absorbed without a patient noticing. A deactivation of the second of two primary-care physicians in a rural county is a different event entirely.

The county-level concentration is what turns a national curiosity into an access story. When we ranked counties by net primary-care enrollment loss, the top decile was almost entirely non-metropolitan and overwhelmingly already HPSA-designated. In several of those counties the March deactivations represent a double-digit percentage of the active primary-care panel — the kind of move that does not average out, because there is nothing nearby to average against.

A deactivation is not necessarily a provider leaving practice. It can reflect a billing reorganization or a move between group enrollments. We flag the net change in active enrollments per county to avoid double-counting moves.

Why a snapshot beats a survey here

Workforce-supply estimates usually arrive from annual surveys, months or years after the fact. By then the access gap has either been backfilled or hardened into a permanent desert, and the survey cannot tell you which. The PECOS snapshot, frozen monthly, catches the change in the month it posts. Joined to the HPSA file, it does not just say supply fell — it says supply fell where supply was already short, which is the only version of the finding that should move a policymaker.

This is the third study in a row where the leading signal lives in a federal snapshot and the official designation lags it. The SNF quality slide showed up before the star rating; the exclusion clustering showed up before the LEIE listing; here, the supply contraction shows up before the next HPSA redesignation cycle.

Reproducibility

The enrollment snapshot is cms-pecos/2026-04; the shortage file is hrsa-hpsa/2026-05. The geocoding pipeline and the net-enrollment calculation are versioned with the rest of our methodology at /quality/methodology. Every county figure here resolves to the specific rows that produced it.

We will revisit these counties when the next PECOS snapshot posts to see whether the March spike reflects a transient reorganization or the leading edge of a durable contraction.

-- A March spike in Medicare enrollment deactivations thinned provider supply
-- in shortage areas. Snapshot: cms-pecos/2026-04. Maps deactivations to HRSA
-- HPSA shortage designations by county.

with deactivations as (
  select npi, enrollment_state, county_fips, deactivation_date
  from pecos_snapshot
  where dataset_id = 'cms-pecos'
    and snapshot_date = date '2026-04-15'
    and deactivation_date >= date '2026-03-01'
    and deactivation_date <  date '2026-04-01'
),
hpsa as (
  select county_fips,
         max(hpsa_score) as hpsa_score,
         bool_or(is_designated) as shortage_area
  from hrsa_hpsa_snapshot
  group by county_fips
)
select
  coalesce(h.shortage_area, false) as shortage_area,
  count(*)                         as deactivations,
  round(avg(h.hpsa_score), 1)      as avg_hpsa_score
from deactivations d
left join hpsa h using (county_fips)
group by 1
order by 1 desc;

pecos.county_fips = hrsa_hpsa.county_fips
deactivation window: 2026-03-01 ≤ deactivation_date < 2026-04-01
shortage_area = any HPSA designation active in county