First Databank, the Hearst-Owned Drug Knowledge Layer Sitting Inside Almost Every US EHR and Pharmacy System: How the Moat Actually Works, What It Costs, and Where AI Can Finally Pry It Loose
Video Preview
🎧 Podcast episode for paid subscribers only. Also available on Spotify.
To listen to paid episodes in Apple or Spotify, link your Substack subscription via the show settings on those platforms (instructions inside the Substack app under Subscriptions → Podcast).
Table of Contents
What FDB actually is, and why Hearst owns it
The product is not the data, it is the maintenance
How the money works: OEM, modules, and the renewal ratchet
The switching cost math nobody writes down
Where the public data actually lives
The competitive set, and the antitrust accident that created it
What AI actually changes, and what it does not
Four attack surfaces that are real
The liability problem, which is the whole ballgame
What a credible challenger looks like in practice
Scorecard: who wins, who gets bought, who burns five years
Abstract
FDB (First Databank) is a Hearst-owned company that sells the machine-readable medication knowledge layer embedded in US EHRs, pharmacy systems, PBMs, e-prescribing, and claims adjudication. MedKnowledge is the flagship. Prizm is the newer platform.
The raw inputs are mostly free. FDA structured product labeling, the NDC directory, NADAC, ASP, FUL, and NLM’s RxNorm all cost zero dollars. The product is normalization, curation, proprietary identifiers, clinical rules, and daily maintenance.
The real moat is identifier lock-in. GCN_SEQNO, HICL_SEQNO, and MEDID at FDB, GPI and DDID at Medi-Span. Formularies, order sets, allergy records, interaction overrides, and analytics get keyed to those codes and then never move.
Regulation creates a demand floor. ONC certification requires drug-drug and drug-allergy checking. OBRA ‘90 forced prospective DUR in pharmacy. Nobody is opting out.
Duopoly by accident. DOJ forced Hearst to divest Medi-Span in 2001 after its 1998 acquisition, with a roughly $19 million disgorgement, which is why the second player exists at all.
AI collapses curation cost by one to two orders of magnitude but does not touch liability, distribution, or identifier gravity.
Four viable attacks: the crosswalk liberation layer, the tuning layer sold on top of incumbents, greenfield licensing to AI-native buyers priced out of enterprise OEM deals, and vertical content nobody bothered to encode.
Constraint that decides everything: FDA’s 2022 CDS guidance requires an independently reviewable basis. Black box output fails. Citation-grounded output passes. Build accordingly.
What FDB actually is, and why Hearst owns it
First Databank is a roughly fifty-year-old company out of the Bay Area that sells the medication knowledge layer nobody sees and everybody uses. When a physician types a drug name into Epic and gets an interaction warning, when a pharmacy system flags a duplicate therapy, when a PBM adjudication engine classifies an NDC into a therapeutic class to apply a step edit, there is a decent chance the logic underneath came from a licensed FDB dataset. The flagship is MedKnowledge, formerly known as National Drug Data File Plus, and FDB has been shipping a newer knowledge platform under the Prizm name for a while now. The company also sells patient-friendly medication instructions, an e-prescribing service aimed at specialty medications, and a set of pricing and reimbursement files that state Medicaid programs and PBMs use to decide what things cost.
The ownership is the fun part. Hearst owns it, and has since around 1980. Yes, that Hearst. The magazine company. The one that publishes Cosmopolitan and owns a chunk of ESPN also quietly owns a piece of infrastructure that touches nearly every prescription written in the United States. Hearst Health is a real operating group, not a hobby, and over the years it has included care guidelines, order sets, and home health software alongside FDB. This is the sort of thing that makes newcomers to healthcare data assume the industry is a prank. It is not a prank. It is just what happens when a family-controlled media conglomerate makes one extremely good acquisition in 1980 and then does absolutely nothing to jeopardize it for four decades.
Nobody outside the building knows FDB’s revenue, because Hearst is private and does not break it out. Directionally, the customer footprint suggests a business in the low hundreds of millions with very high gross margin and renewal rates that would make a SaaS board weep with joy. The point is not the size. The point is the shape. This is reference-data economics, closer to a Bloomberg terminal or a bond pricing service than to a healthcare software company, and it has all the pleasant characteristics that implies.
