April 1, 1999
Most clinical settings collect hundreds of data points on each client or patient. What are known as "data warehousing" strategies can be used to develop a rich and valuable data archive to improve treatments, inform decisions about program development, prepare annual reports of program activity, and plan and implement research.
A comprehensive data warehousing plan should address all the uses you have for archived information: clinical, legal, administrative, and research. This article offers suggestions for creating a data warehouse for research purposes from information that has already been collected, with suggestions for planning a new data warehousing system.
Creating a Data Warehouse
1. Make an honest appraisal of the data that have already been collected. Good research questions can usually be asked of data that are already available, provided that the user has confidence in the quality of the data. Questions to ask about your archive include:
- Are the measures reliable and valid for the client population?
- Are there lots of missing data?
- Are the data in a format that is easy to interpret and enter?
- Which portions of the data appear to address issues that are of research interest?
The answers to these questions may suggest the need for changes in the clinical data collection process. It may be necessary to add more measures to answer certain research questions. Consider ways to streamline the data collection process to reduce the burden on clients and therapists and reduce the expense of collecting and entering data.
2. Make a list of the data desired for warehousing and note their current location. Clinical data are stored in many forms and sometimes in many locations. Demographic and medical history may be located in the hospital chart while intake measures are stored in the therapists' filing cabinets. Some information may already be stored on computer. However, information from computerized administrations of clinical instruments typically is not transferable to other computer programs (See comments below). A careful appraisal of what data are actually needed will reduce the burden of data entry.
3. Make an appraisal of the resources you have for storing and analyzing data. Garden-variety desktop computers are more than adequate for creating a data warehouse and analyzing data for research. Statistical software packages such as SPSS will run on both Macintosh and IBM-compatible computers. The graphical user interface makes the process of entering, storing, and analyzing data simple. In addition, SPSS can import data from other formats, such as Microsoft Excel spreadsheets. Many clinical settings allow practicum students and interns to use some of the data for their own research in exchange for help in collecting and entering data.
If you have the programming and hardware resources for computerized assessments, then clients and staff can enter data directly into the computer (or onto scan sheets) for later exportation to statistical software. Some clinical settings have written their own software to computerize all the self-report measures in their assessment battery. Two major issues in the use of online assessments concern compatibility with statistical software and whether the measures have been validated for computer -- rather than paper-and-pencil -- administration.
4. Plan for the protection of human subjects. Safeguard patient confidentiality by entering identification numbers rather than names into your data warehouse. In addition, in some places the use of archival data for research requires prior subject consent, usually obtained with the consent for treatment.
The good news is that maintaining a data warehouse is easier than creating one. A carefully planned warehouse can meet the data needs of many clinical researchers and students in addition to serving important programmatic and administrative functions.
This report is sponsored by the ISTSS Special Interest Group on Research Methodology. If you are interested in becoming a member of this group, contact co-chairs Daniel and Lynda King, National Center for PTSD (116B-2), Boston VA Medical Center, 150 S. Huntington Ave., Boston, MA 02130. E-mail: firstname.lastname@example.org or lking@