Open Health Data Hub



All posts
|3 min read

Did Medicaid Providers Actually Come Back After COVID?

Medicaid provider accessCOVID Medicaid impactMedicaid workforceMedicaid billing trends 2020 2021

Across most of the country, Medicaid provider counts look roughly stable. Missouri tells a different story.

Between 2019 and 2023, Missouri lost 4,544 active Medicaid billing providers, a 43.5% decline that dwarfs every other state in the dataset. That's not a COVID dip followed by recovery. By 2022, Missouri was already down to 5,840 active providers, and 2023 brought only a marginal uptick to 5,906. The state ended the period with fewer than 6,000 providers serving a Medicaid population that hadn't shrunk.

The COVID Trough Was Real, But It Wasn't the Whole Story

The national picture in spring 2020 was alarming. Unique billing NPIs active in Medicaid claims fell from 236,134 in February 2020 to 162,138 in April 2020, a drop of nearly 74,000 providers in two months. By August 2020, that number had partially recovered to 220,412, still below the pre-COVID peak of 243,432 recorded in October 2019.

That partial recovery looked, at the time, like evidence that the system was resilient. Providers had paused billing during lockdowns, then returned. The national trend line bent back upward.

What the aggregate masked was a divergence happening at the state level. Some states recovered. Others didn't.

Missouri vs. Arkansas: A 43-Point Gap With No Obvious Explanation

The starkest comparison in the data is between Missouri and Arkansas. Arkansas had 4,140 active Medicaid providers in 2019 and 4,141 in 2023, a 0% change. Missouri started with more than twice as many providers (10,450) and ended with 5,906. Two neighboring states, similar in size and regional context, produced outcomes that are almost impossible to reconcile.

State2019 Providers2023 Providers% Change
Missouri10,4505,906-43.5%
South Carolina5,6614,002-29.3%
Tennessee8,0215,826-27.4%
Utah2,7922,088-25.2%
Virginia7,6847,664-0.3%
Arkansas4,1404,1410.0%

South Carolina fell 29.3%, from 5,661 to 4,002 providers. Tennessee dropped 27.4%, losing 2,195 providers. Virginia, by contrast, declined from 7,684 to 7,664, a 0.3% change that amounts to statistical noise. These aren't small states with idiosyncratic Medicaid programs. They're mid-to-large states with substantial Medicaid populations, and their outcomes span a 43-point range.

For patients, a 43.5% provider decline doesn't mean 43.5% fewer appointments. It means longer travel times, longer waits, and providers who remain taking on larger panels. The access gap created by provider exit is often invisible in claims data until it becomes acute.

What "Recovery" Actually Means Depends on the State

The national monthly trend through August 2020 suggested the provider workforce was bouncing back. But the 2023 state-level data shows that for Missouri, South Carolina, and Tennessee, the COVID period appears to have accelerated a departure that either didn't reverse or reversed only partially.

Missouri's 2022 count (5,840) and 2023 count (5,906) are nearly identical, which means the state wasn't still declining by 2023, but it also wasn't recovering. The floor had been set. Arkansas and Virginia, meanwhile, demonstrate that a stable provider base through the same period was achievable.

The difference between a state that lost 43.5% of its Medicaid billing providers and one that lost essentially none, over the same four-year window, is the difference between a temporary disruption and a structural change in access. Whether Missouri's Medicaid beneficiaries are now absorbing that gap through longer waits, out-of-network care, or foregone treatment entirely isn't visible in provider counts alone. But the counts make clear the gap exists, and that it didn't close between 2022 and 2023.

What state-level policy, reimbursement structure, or managed care design kept Arkansas whole while Missouri contracted by nearly half is a question the national recovery narrative has so far left unanswered.

Explore the data yourself

Run your own queries against 240M+ rows of federal health data using natural language — powered by AI.

Start analyzing