|
The Challenge
The NARCOMS research database began its life in the
mid-90’s at Yale University as a small Macintosh
database. Over the years, it underwent innumerable revisions
by different database administrators. It was migrated
from the Mac to Microsoft MS Access in the late 90’s
and finally in the year 2001 it was ported to Microsoft
SQL server. By early 2003, the database resembled an
archeological exploration site with generations of incredibly
complex rules and exceptions buried into each table.
Even though the data itself was of exceptional quality
and it resided in an enterprise class database, the
actual data model had become so convoluted that reliable
queries were a challenge to construct. Late in 2003,
Emerge.MD was asked to help simplify the database and
to unify the data model. In addition, the NARCOMS online
registry was designed to be a copy of the paper questionnaire
without harnessing the power of interactive online forms
(eg. both men and women were asked if they had ever
been pregnant). Therefore, the registry was unnecessarily
long and patient compliance with data entry was low
as a result. Emerge.MD was also asked to completely
revamp the online enrollment process to leverage the
strength of interactive web-based data entry.
Our Solution
After substantial analysis of the NARCOMS database and
registry web site, the Emerge.MD team proposed a bi-directional
and multi-phased approach to achieve its objectives.
We began both prospective and retrospective processes.
Prospectively, we designed a new interactive patient
registry web portal from the ground up. Each question
the patient answered had an impact on the subsequent
questions being asked. Unnecessary questions were therefore
quickly eliminated, thereby simplifying the registry
enrollment process. The data was then captured into
a comprehensive research data model with clear scalability
in mind so as to avoid the problems of the past. Retrospectively,
we begin to recondition prior NARCOMS data into the
same data model. This process involved investigating
all prior assumptions that had been used to store prior
data. These assumptions were gathered from notes or
reverse engineered out of software code and SQL scripts.
The cleaned and reconditioned retrospective data was
then joined to the prospectively gathered data to provide
a unified and simplified environment for research. Today,
we continue to service the database needs of NARCOMS,
helping them to find a cure for MS.
|
