Thursday, September 28, 2006

FolioViews and ICD-9: a billion dollar tale never told

The more decrepit one becomes, the more "odd stories never told" one accumulates. This is one of those stories, a tale of a seemingly minor technical decision made 20 years ago -- with vast economic, social and scientific consequences that almost nobody recognizes today. It's a story about what happens when data moves into software and how proprietary data formats can shape a world -- unrecognized by anyone. It's a story of the mutability and immortality of software, and about data lock. It's boring, obscure, and I really don't know it all -- but I'm one of the few people that can guess some pieces of it. So, I'll tell it now, and hope corrections in the comments section.

In the 1980s a company named Folio Corporation produced a DOS information management software packaged called "FolioViews". They called it an InfoBase, and it was one of several semi-structured information managers that, in today's terms, were a cross between a database, a textbase, a content management system and and ontology editor. I was quite fond of that class of software, the most famous of which was probably Lotus Agenda. There were several similar, immensely powerful, and long forgotten applications on Mac Classic.

I remember that product because I used to write software reviews for the Journal of Family Practice in the 80s, and I wrote on FolioViews/DOS. (Back then JFP was still a scientific journal, not a throwaway.) FolioViews made a rocky transition to Windows 3.x, I wasn't as impressed with the Windows version, but, oddly enough, I have all the manuals. The software is gone.

That's where I thought the story ended, around 1993. It didn't though.

FolioViews was picked up by a number of governments, government contractors, publishers, CD-ROM vendors and legal firms around the world. It became a vertical market product. Somewhere along the way it was purchased by NextPage and buried (a search on their web site yields nothing about FolioViews). It probably lives a zombie existence there as a potential source of software patent litigation; many of these 1980s software packages implemented designs and ideas that were later patented independently, the old software can invalidate later patents.

Ahh, but what about the economic implications? That's a twisted tale. Somewhere along the line FolioViews became the standard means of editing and publishing ICD-9 and ICD-10, which started life as data sets used by epidemiologists and healthcare researchers. (For all I know it's used for CPT also, but that's murkier.) Here's the catch. ICD-9, in the US at least, became, by default, the only "standard" way to talk about diseases, disorders, and patient conditions. It became the flawed foundation of health care payment rules (along with CPT, DRG, etc etc), clinical decision support systems and EHRs (including one I helped create).

ICD-9 is the primary source of everything we know about what health issues Americans have, a fundamental constraint on the accuracy and capabilities of decision support systems, and a major obstacle to smarter/better healthcare systems. ICD-9 is also built and maintained in FolioViews -- but it's not distributed that way. It's distributed as a paper book (or a PDF, same difference). Fragments of ICD-9 are also published as database-like tables -- but that's only a fraction of it. The complete ICD-9, which exists only in FolioViews, is a rich and baroque mixture of classification, ontology and document -- mostly we don't have it.

Fundamentally, ICD-9 is locked away in the FolioViews Infobase somewhere in the offices of a federal contractor. My 1990s FolioViews manual tells me FolioViews can export wordprocessing documents or the proprietary "Folio Flat File" -- but that's it. Data Lock. Big time. FolioViews had its own peculiar and powerful way of managing data, and even if FolioViews had some sort of useable export facility it would be a non-trivial job to recreate it. It's not clear that there's an equivalent modern software environment that could recreate FolioViews. It might be easier to hire people to translate the WordPerfect output into a purpose-built environment -- except that really we need to dump ICD-9 and ICD-10 and do a SNOMED-derivedICD-11 (but that's another story).

So here we are, twenty years later, living with the consequences of a modest decision made when DOS was king. In a sense, a large portion of American healthcare is a captive of FolioViews, software who's fundamentals, including file formats and data structures, are lost in the mists of ancient computing. Something to think about the next time you wonder why software vendors struggle mightily to produce reliable and interoperable systems to support both clinical practice and financial obligations. The reasons may be more mysterious than you can imagine ...


Update 12/13/06: Interesting twist. A renamed descendant of Folio Views is used by the Mormon church (LDS) to distribute census/genealogy data. FolioViews was indeed a groundbreaking piece of software in its DOS incarnation, so unique it cannot be quite killed. Incidentally, I found this usenet discussion as a side-effect of how I track my usenet postings and follow-up. I add the semi-unique string 'jfaughnan' as metadata to my postings, and I have standard google query embedded in a my news page that retrieves postings. Since 'jfaughnan' appears in the URL for this blog, my standing query turned up the usenet posting referencing the blog. A very new age sequence of circumstances.


Anonymous said...

platsocr said...


We've worked with the ICDs and SNOMEDs in English, French and a few others, since the late 1980s.

We know well the frustrations of working with FolioViews.

Is it possible to get in touch with the author of this document and share experiences ?

Thank you and have a good day,

Michel A. Boisvert
MedSight Informatique Inc.