The ADA Says 16% of Dentists Are DSO-Affiliated. Here's What the Public Record Shows.

The most-cited number in dental consolidation comes from the American Dental Association's Health Policy Institute: 16.1% of U.S. dentists were affiliated with a dental support organization as of 2024, a share that has more than doubled since 2015. Among dentists fewer than ten years out of school it's 27%; among those more than twenty-five years out, 9%.

It's a good number. It's also a survey number: self-reported, national, published once a year. That makes it the right tool for one job (sizing the trend) and the wrong tool for another (knowing which practices in a specific metro buy through corporate procurement this week). We publish the second number, and because it's built entirely from public records, anyone can check our work. Here is how it's built.

What the ADA number is, and isn't

The HPI figure comes from the ADA's annual survey of dentists. Its strengths are real: it captures an affiliation a dentist knows about even when no public record names it, and it's a consistent national time series. Its limits are just as real. It's self-reported, so it depends on how a respondent reads the word "affiliated." It's national, so it can't tell a supply rep working Phoenix from one working the Bay Area which of their practices buy centrally. And it's annual, so it can't show a transition the week it happens.

What we measure instead: a public-records floor

ProviderSignal measures something narrower and more literal: the share of practices whose DSO affiliation is visible in public records, computed per metro and refreshed weekly. The word that carries the weight is visible. If a chain renames an acquired office, files it under a personal professional corporation, and publishes nothing, the affiliation exists but leaves no trace we can point to. So our number is a floor: the verifiable minimum, not the total. As of June 13, 2026 the national figure is 7.1%.

That floor sits below the ADA's 16%, and that is exactly what you would expect: a public-records minimum should come in under a self-reported total. The value of our number isn't a bigger headline. It's that every flag is independently checkable, broken out by market, and never more than a week old. Because it's a floor, it carries one rule we never break: a practice we don't flag is one we could not see affiliated in public data, not one we know to be independent.

Six signals, four families

A practice rarely announces "we are a DSO office." The affiliation has to be inferred from the traces it leaves in public data. We run six independent signals that fall into four families:

• Name match against a brand registry. We canonicalize 200+ DSO brand names and their hundreds of state-by-state legal-entity aliases (Heartland, Aspen, Pacific Dental, Smile Brands, Western, and the long tail of regional consolidators), with an explicit exclusion list for university clinics, FQHCs, and public-health programs so they don't get mislabeled. The full list lives in our open dso_identifier.py module.
• Multi-location and shared-infrastructure patterns. The same organization name across several buildings, multiple providers clustering at one known corporate address, or three or more distinct legal entities sharing a single dispatch phone number.
• Brand-published office locators. We scrape the public "find an office" directories that 21 national DSOs publish themselves: a ground-truth set of roughly 4,400 buildings the brands claim as their own, re-scraped monthly.
• Corporate-officer clustering. For chains that publish no locator at all, a DSO's corporate credentialing team often signs as the authorized official across its whole network. One headquarters phone number on dozens of otherwise-unrelated registrations is a near-deterministic fingerprint.

Why one signal isn't enough: the PDS example

Pacific Dental Services publishes 846 offices on its own website. Name-matching alone, the first and most obvious signal, flags four of them. The other 842 file under local practice names like "Modern Dentistry" that share no text with "Pacific Dental Services."

That gap is the whole argument for running the other five signals. PDS publishes a locator, so the building-level match recovers the offices name-matching misses. A chain that publishes no locator gets caught by the corporate-officer fingerprint instead. No single signal is close to complete on its own, which is precisely why a one-signal "count the brand names" approach systematically undercounts consolidation, and why we report our number as a floor rather than a final count.

Survey and public record: two lenses, both true

The honest reconciliation is that the ADA and ProviderSignal measure different things. The ADA asks dentists and gets a national self-reported total. We read the public record and get a per-metro, weekly, externally verifiable minimum. Neither is "the real number" that invalidates the other. If you want to know whether consolidation is happening (it is), the ADA's 16% is the cleaner stat. If you need to know which practices in Minneapolis-St. Paul (the most consolidated metro we currently track) or the San Francisco Bay Area (among the least) buy through corporate procurement this week, you need a verifiable per-market floor. That is the one we publish, for every state and metro we cover.

DSO-affiliation figures are ProviderSignal's own measurements from public records (NPI registry, state corporate filings, brand-published office locators, and NPPES authorized-official data), computed as a visible-affiliation floor and refreshed weekly. The national figure cited, 7.1%, is as of June 13, 2026 and drifts as the data updates; the brand registry, exclusion list, and signal logic are in the open dso_identifier.py module. The ADA Health Policy Institute figures (16.1% of dentists DSO-affiliated as of 2024; 27% for dentists fewer than ten years out of school, 9% for more than twenty-five) come from the HPI's published practice-modalities research. Questions, or a discrepancy to report: support@providersignal.com.