Ghost networks start in provider data, not in the directory

Ghost networks start in provider data, not in the directory. A ghost network is a list of clinicians who appear to be available to a plan’s members but who, in practice, cannot be reached or are not taking new patients. The directory is where members find the problem. It is rarely where the problem begins.

The scale has been documented for years. When CMS reviewed Medicare Advantage online directories in 2018, it found that 48.74% of listed provider locations had at least one inaccuracy. In 2023, Senate Finance Committee staff posing as patients called 120 mental health providers listed in Medicare Advantage directories and could make an appointment only 18% of the time.

Policy is now catching up. In February 2026, Congress enacted the REAL Health Providers Act as part of the Consolidated Appropriations Act, 2026, which will require Medicare Advantage plans to verify directory information at least every 90 days and to analyze and publicly report their directory accuracy, beginning with plan year 2028. CMS has separately required plans to publish their directories on Medicare Plan Finder for the 2027 open enrollment period. This analysis looks at why directories decay and what the new rules require, then at what a plan would have to change in its provider data to meet them.

Key takeaways

• Inaccuracy is the norm, not the exception. CMS found at least one error in nearly half of Medicare Advantage directory locations it reviewed, with plan-level error rates ranging from under 5% to over 93%.
• Most errors are about where and whether. The most common errors were providers missing from the listed location or listed with a wrong phone number.
• Directories feed network adequacy. When the same provider data is used to show that a network meets access standards, a ghost listing makes an inadequate network look adequate.
• New federal rules raise the bar. Medicare Advantage plans must post directories on Medicare Plan Finder for 2027 open enrollment, and the REAL Health Providers Act adds 90-day verification and public accuracy scores starting in 2028.
• Credentialing does not keep directories current. Credentialing verifies qualifications every 3 years. Directory accuracy depends on operational facts that change every week.
• The fix is a data supply chain. Accurate directories require continuous updates from the sources that know first, with verification aimed where errors are likely.

How inaccurate directories are

CMS conducted 3 rounds of online directory reviews for Medicare Advantage between 2016 and 2018. Each round sampled providers and called the listed locations to check whether the information was right.

From CMS’s Online Provider Directory Review Report, Round 3 (November 2018). CMS noted that its methodology changed slightly between rounds.

The third round covered 52 Medicare Advantage organizations and 10,504 provider locations in 4 specialties. Error rates varied enormously across organizations, from 4.63% to 93.02%, with an average of 44.97%. That range is the most hopeful number in the report. It shows that some plans were able to keep their directories largely accurate, and that high error rates were a choice of process, not an unavoidable feature of the work.

The types of errors were consistent across plans.

From the same CMS report.

Behavioral health has drawn the most attention. The Senate Finance Committee’s 2023 secret shopper study reviewed directories from 12 Medicare Advantage plans in 6 states and called 10 listed mental health providers for each plan. A third of the listings were inaccurate or went unanswered. Staff could make an appointment only 18% of the time.

The industry’s own data points the same way. A May 2024 CAQH issue brief noted that nearly 70% of the inaccuracies CMS identified related to provider location, and reported that CAQH’s experience showed accuracy of 80% to 99% on key contact fields was achievable.

Why directories decay

Directories go wrong for structural reasons, most of which sit upstream of the directory itself.

The data changes constantly. Clinicians move between locations and groups, and they stop taking new patients or retire. Each change is small. Across a network of tens of thousands of clinicians, the changes add up to a steady stream that never stops.

Each practice reports to many plans. CAQH found that a physician practice holds, on average, 20 health plan contracts, and spends at least one full day a week on directory maintenance at a cost of almost $1,000 a month. Every plan asks for the same facts in its own format on its own schedule. Practices fall behind, and plans fill the gap with whatever they last received.

Directories are built at contracting and credentialing. Many plans populate their directories from the data collected when a clinician joins the network, such as the contract roster and the credentialing application. That data is accurate on day 1. Credentialing then revisits the clinician every 3 years, as we described in February, and it focuses on qualifications such as licensure and sanctions, not on whether the clinician still sees patients at a given address on Tuesdays.

Delegation and rosters add layers. When a plan delegates credentialing or network management to a provider group or vendor, directory data often arrives as a roster file. Rosters can lag reality by months, and they often list every clinician in a group at every location the group operates.

Practice patterns blur the categories. A clinician may see patients at a location 1 day a month, or accept new patients for some conditions and not others. A directory field with 2 values, yes or no, cannot describe that. Members experience the mismatch as an error even when every field is technically true.

Nobody owns the whole record. In many plans, different teams own different pieces of a listing, and member services is the one that hears about the errors. Each part may be accurate on its own terms while the combined listing is wrong.

Why it matters beyond inconvenience

An inaccurate directory wastes members’ time. It can also cost them care and money, and it can distort how regulators judge a plan’s network.

Network adequacy rests on the same data. Medicare Advantage plans must meet time and distance standards and minimum provider counts for each county they serve under 42 CFR § 422.116. Plans demonstrate compliance with lists of contracted providers and their locations. If those lists include clinicians who are not at the listed site, or who do not see the plan’s members, the network can pass on paper while failing in practice. A ghost network is, in effect, a network adequacy problem hiding in a directory.

Members pay for errors. A member who cannot find an in-network clinician may go out of network or go without care. For mental health in particular, an unreachable list of providers can mean a delay at a moment when delay is dangerous.

Federal law already protects some members. For group health plans and individual market coverage, the provider directory provisions enacted in the Consolidated Appropriations Act, 2021, require plans and issuers, for plan years beginning in 2022, to verify directory information at least every 90 days and to update it within 2 business days of receiving new information from a provider. A member who relies on inaccurate network information cannot be charged more than in-network cost sharing. Those protections did not cover Medicare Advantage.

Parity analysis depends on it too. For employer and individual coverage, mental health parity rules ask whether a plan’s network standards are comparable for mental health and medical care. Federal regulators finalized more detailed requirements in 2024, including data on network composition, and announced in 2025 that they would not enforce the new provisions while they reconsidered them. Whatever the outcome, any credible parity analysis of networks starts with knowing which clinicians are actually available.

A directory that lists clinicians who cannot be reached does not just mislead members. It makes an inadequate network look adequate.

The new rules

The federal response has arrived in pieces, each on its own timeline.

Compiled from the law firm and trade press summaries linked in this section.

The Medicare Plan Finder requirement comes from the second contract year 2026 final rule, which CMS issued separately in September 2025 to give plans more lead time. Plans must directly disclose their directories on Medicare Plan Finder for the 2027 open enrollment period and update them within 30 days of becoming aware of changes. They must also attest at least annually that the information is accurate.

The REAL Health Providers Act goes further. As summarized by Defacto Health, the law, signed in February 2026, requires Medicare Advantage plans to verify and update directory information at least every 90 days and to remove departed providers within days. Plans must hold members harmless from additional cost sharing when they reasonably rely on an incorrect in-network listing, and they must analyze directory accuracy and report the results publicly. The requirements begin with plan year 2028. The law directs federal officials to issue guidance by August 2027 on maintaining accurate directories, including the use of public data sets. Much of the detail, including how accuracy will be measured, is left to implementation.

CMS has also been building a public data source. At an interoperability event in 2025, CMS showed attendees a beta national provider directory that would let Medicare beneficiaries find providers that accept Medicare, available through a free FHIR API. If it matures, it could become a shared reference point that reduces the number of times each practice has to report the same facts.

What accurate directories require

Meeting a 90-day verification standard by calling every practice every quarter would be expensive and would add to the burden on practices that already report to 20 plans. The arithmetic makes the point. A plan with 20,000 listed clinicians that verifies each one every 90 days needs about 80,000 verifications a year. If signals from claims and national registries can confirm most listings without a call, the remaining calls become a manageable list aimed at the listings most likely to be wrong. The plans that do best will treat directory accuracy as a data supply chain problem rather than a calling campaign.

• Take updates from the sources that know first. Practice systems and the plan’s own claims both carry signals about where a clinician practices. An event-driven provider record, updated whenever a source reports a change, stays current between verification cycles.
• Use claims as a check. A clinician listed at a location where the plan has seen no claims for a year is a strong candidate for verification. A clinician whose claims come from a location not in the directory points to a missing listing.
• Verify by risk, not by alphabet. Specialties with known access problems, such as behavioral health, and listings with weak signals deserve more frequent verification than a stable hospital-based practice with steady claims.
• Treat acceptance of new patients as data, not a default. A field that defaults to yes produces a directory that overstates access. A field that requires confirmation produces fewer, more honest listings.
• Label what is unverified. A member should be able to see when a listing was last confirmed. An honest label is better than a confident error.
• Measure accuracy like a quality metric. Regular secret shopper samples, scored the same way each time and broken down by specialty and region, tell a plan whether its process is working before a regulator does.
• Reduce the burden on practices. Accepting updates through shared industry sources and national data, rather than plan-specific forms, makes it easier for practices to keep every plan current at once.

None of these steps is exotic. Most plans already have the data. What is missing is usually the connection between the systems that hold it and the directory that displays it.

Where Redlake stands

We build tools health plans use in utilization review, and network status shows up in our work more often than people expect. Whether a clinician is in network and can take a new patient affects requests for out-of-network exceptions and continuity of care. A member who cannot find an available in-network specialist may have a valid reason for an out-of-network request, and a reviewer should be able to see that.

Those requests are also a signal worth using. When members repeatedly ask for out-of-network exceptions in a specialty and region where the directory shows plenty of in-network options, the directory is probably wrong. A plan that routes that pattern back to its provider data team learns about ghost listings from the members who found them.

We think the same principles that govern a good review apply to provider data. Facts should be recorded with their source and the time they were confirmed, and changes should arrive as events rather than at the next cycle. Every decision should also be reconstructable, including what the plan’s provider data said on the day a decision was made.

Directory accuracy will soon be scored and published. Plans that build the data supply chain now will be able to show their numbers with confidence. Plans that wait will discover their accuracy score at the same time as their members.

What to watch in 2026

• Medicare Plan Finder. How plans prepare their directory submissions for the 2027 open enrollment period, which begins in October 2026, and what CMS publishes about data quality.
• Implementation of the REAL Health Providers Act. Early signals from federal officials on how accuracy will be measured and displayed.
• The national provider directory. Whether CMS’s directory moves from beta to production, and whether plans can use it as a verification source.
• Commercial market rules. Whether federal regulators finally issue regulations for the directory provisions that took effect for commercial plans in 2022, and what happens to the paused parity provisions on network composition.
• State action. Whether states adopt their own accuracy standards or secret shopper programs for plans they regulate.

About this analysis

Basis. The accuracy evidence comes from CMS’s 2018 directory review report and the Senate Finance Committee’s 2023 secret shopper study, supplemented by a 2024 CAQH issue brief. The rules are described from published summaries of each law and rule, along with reporting on CMS’s national provider directory prototype.

Data notes. CMS’s directory reviews used slightly different methods across rounds, so their results are not strictly comparable. The Senate study was small, with 120 calls, and was designed to illustrate a problem rather than estimate a national rate. Details of the REAL Health Providers Act are drawn from published summaries, and implementation details are subject to future guidance and rulemaking. The verification arithmetic is an illustration.

Sources. CMS, Online Provider Directory Review Report, Round 3 (November 2018). Senate Finance Committee (May 2023). CAQH, Provider Directories, What Impacts Accuracy (May 2024). Ballard Spahr (February 2022). Healthcare Dive (September 2025). Defacto Health (February 2026). Fierce Healthcare (2025). 42 CFR § 422.116.

Frequently asked questions

What is a ghost network?

A ghost network is a set of providers listed in a health plan’s directory who are not actually available to members, often because they cannot be reached or are not accepting new patients.

How accurate are Medicare Advantage provider directories?

In CMS’s 2018 review, 48.74% of the provider locations listed in Medicare Advantage directories had at least one inaccuracy, and error rates by plan ranged from 4.63% to 93.02%.

What does the REAL Health Providers Act require?

Beginning with plan year 2028, it requires Medicare Advantage plans to verify directory information at least every 90 days and to publicly report their directory accuracy. It also protects members from extra cost sharing when they rely on errors.

When will Medicare Advantage directories appear on Medicare Plan Finder?

Under CMS’s September 2025 final rule, plans must disclose their provider directories on Medicare Plan Finder for the 2027 open enrollment period, which begins in October 2026.

Does credentialing keep directories accurate?

Not by itself. Credentialing verifies qualifications such as licensure and board certification, usually every 3 years with monthly checks for sanctions. Directory accuracy depends on operational details, such as location and whether a clinician is accepting new patients, that can change at any time.