Oracle Objects Vs Relational Tables

We have a database (Oracle 7.3.x) that is used to collect medical research data for statistical analysis. A common scenario is that a medical study opens (40 - 100 per year) and specific data is collected for the analysis. Approximately 75% of the data is common among all studies (e.g., demographic data), but that last 25% is starting to cause problems. A good example would be the patient eligibility category of elements. We ask approx. 15 questions to determine if the patient is eligible to go on a study and usually about 5 of those are common across studies. Our first approach was a separate eligibility table for each study (a patient can be on multiple studies), but the number of tables started to grow quickly and data integrity was becoming difficult to maintain. Our second approach was to combine them into a single large table. This seems to be working, but the number of data elements is growing quickly. We are considering upgrading to Oracle 8i this year and turning tables like the eligibility table into a series of objects (a "study" eligibility object would inherit common elements from a
"parent" eligibility object).

The "object relational" model has been heavily advertised by Oracle, I was wondering if anyone had any experience going from relational to
"object relational" and what you thought of the process. I would also
be interested in any opinions that anyone has about my "dilemma". I'm sure that I'm not the only one to ever face this scenario.

Thanks in advance

Bill
--
Bill Sanns
POG Statistical Office
104 North Main Street, Suite 600
Gainesville, FL 32601-3330
Tel: (352) 392-5198 ext 309 -- Fax: (352) 392-8162 bill@pog.ufl.edu -- http://www.pog.ufl.edu Received on Wed Feb 03 1999 - 07:19:13 CST