The Agent Orange relational databases

The Agent Orange relational databases

Shortly after the Agent Orange case was filed in January 1979, the office became inundated with voluminous unorganized medical records some of them weighing 8 pounds. At that time the conventional way to review medical records in personal injury litigation was to hire a number of retired nurses and have them abstract the records then index them. The Vietnam combat veterans did not have the resources to support that kind of effort.

The need for meaningful data

Because the Veterans Administration steadfastly refused to consider the medical conditions of the Vietnam combat veterans as service-connected unless there was a record of a battlefield injury “in country,” it became critically important to conduct what amounted to a general epidemiological study of the health of Vietnam combat veterans.

Mainframe versus minicomputer

At that time the chemical company war contractors who manufactured dioxin contaminated phenoxy herbicides were using leased state-of-the-art IBM-370/175 mainframe computers and the attorney for the Vietnam combat veterans had only a used PDP 11/34 minicomputer from Digital Equipment Corporation (DEC). The PC had not yet been invented and the primary commercial computing languages were COBOL and FORTRAN. The minicomputer was able to run BASIC also.

The databases

Victor and Carol Yannacone designed a relational database system which would manage the medical records of the veterans and their children, many of whom were suffering from catastrophic polygenetic birth defects, and permit the kind of complex statistical analysis of the data that had yet to be used in medical epidemiological research. Dr. Robert W. Liquori, a brilliant young applied mathematician, computer scientist, and educator wrote the code that performed the feat in BASIC.


The entire database, CHAOS (Case Histories of Agent Orange Survivors) was built from two medical records specific relational databases which at the time were unheard of because computers were still being thought of as nothing more than high speed adding machines. HOSPIT (Helpless Overmedicated Sick People In Trouble) was developed to handle fixed field data from the well-defined portions of conventional hospital records, while DOOM (Doctors Orders and Other Mistakes) handled random length medical reports and all of the other free form data about the veterans including the contemporaneous notes of Carol Yannacone taken during telephone interviews of the veterans and their families.

What made the Agent Orange databases unique

The Agent Orange databases were unique because they contained geotemporal data in addition to the conventional kinds of data the commmercial databases of that time handled. It became important to add searchable geotemporal data to the databases because when the veterans were exposed and where they were exposed to dioxin contaminated herbicides was critical in determinining whether their illnesses might be related to their exposure. These were new problems in epidemiology that had never been formally addressed in prior studies.
Soon after the databases were loaded and operating, during a telephone call Carol Yannacone was able to identify the location where and the year during which a veteran served “in country” after hearing a description of his symptoms and conversely she was able to accurately predict what illnesses would be diagnosed from the date and location of the veteran’s service in country.