relational database software can be effective labor-saving ... database management system on a microcomputer to ... system on an Apple II+ microcomputer.
MICROCOMPUTER-BASED MANAGEMENT OF A LONGITUDINAL GERIATRIC RESEARCH STUDY
Nancy E. Clapp-Channing, R.N., M.P.H.*, James A. Bobula, Ph.D.** *Duke-Watts Family Medicine Program 407 Crutchfield Street, Durham, North Carolina 27704-2799 **University of Wisconsin Department of Family Medicine and Practice
Abstract
Research Study
A model utilizing microcomputerized dBASE II generated reports to aid the execution of data collection in a large geriatric research study is examined. Multiple providers follow study subjects prospectively in the hospital at weekly intervals and then at 1 and 4 months postdischarge using a minimum of 12 forms per subject. This effort has revealed that an inexperienced staff member can be trained easily to use the computer for 2 to 3 hours a week to manage all the data collection needs in an otherwise logistically complex matrix of information; study team members can easily retrieve the information necessary to meet their responsibilities; and immediate feedback of data gathered can be vital for identifying and correcting problems as the study progresses.
Introduction The microcomputer is becoming an increasingly important instrument in the medical setting. Billing, accounting, managing clinical records, controlling instruments or storing and analyzing their outputs, word processing, financial forecasting, teaching and telecommunicating are now commonplace functions of microcomputers in medical offices and hospitals. In addition, researchers are finding that microcomputer hardware and relational database software can be effective labor-saving resources for project management, data collection, and data analysis.1-8
impact of hospitalization
on
the
Program,
residency with 9 family physician family practice residents, and 2 mid-level practitioners providing primary medical care to approximately 12,000 community patients. Study subjects are elderly patients admitted to Durham County General Hospital. At least 100 participating subjects are being randomly selected within 24 hours of admission. Independent medical and functional assessments are performed at the time of the patient's admission to the hospital, weekly during hospitalization, at discharge, and twice during the four months after discharge (Figure 1). Each medical assessment is performed by one of three physician faculty and each functional assessment by one of two mid-level practitioners. At the time of discharge a chart
faculty,
a
39
audit is done. Both assessors write a discharge summary and after the last follow-up visit they complete a final summary. The task of keeping track of data collection activities is being handled by a social worker serving as project coordinator. The project coordinator identifies potential subjects according to well-defined selection criteria; records demographic information both for the subjects who consent to participate and for those eligible but non-participating subjects; assigns a medical and a functional assessor to follow the subject throughout the study; maintains records of all scheduled and performed assessments; and monitors all deaths, readmissions and refusals of the participants. In that each of the 100 subjects is evaluated at least five times by each of two assessors and that, at any given time, subjects will be at various stages in the study design, the potential for overlooking a required assessment is great. Because activity the integrity of this study's analyses and conclusions is directly related to the completeness of the data collected, the researchers decided to use a microcomputer to keep track of each patient's progress through the study.
This paper describes the use of a relational database management system on a microcomputer to organize the logistics of data collection for a longitudinal research study of hospitalized elderly patients. The system features ease of data entry and retrieval, facilitating production of researchers' task assignment lists and of interim reports vital for timely identification and correction of study implementation problems.
CH2090-9/84/0000/0348$01.00 © 1984 IEEE
The
functional capacity (ability to perform everyday activities) of elderly patients is being studied at Durham County General Hospital, in Durham, North Carolina, a 406-bed community hospital which serves as a training site for residents from Duke University Medical Center. The study is being performed by the Duke-Watts Family Medicine
348
and reprinted the patient's master file record to keep the assessors informed of changes in the patient' s status. Generating a set of lists of assessors' pending assignments took approximately
Database Management The files and programs used for managing the study's implementation were developed using dBASE II software running under the CP/M operating system on an Apple II+ microcomputer. A "master file" contains one record of identifyng data per patient, including the patient's name, his or her study ID, demographics (age, sex, race, marital status), hospital ID, hospital room number, home address and telephone number, specific notes about the patient, etc. An "activities file" contains one record for each assessment or summary to be performed, including the patient's study ID, the assessor's ID, the type of data collection activity involved, the date when the activity is scheduled to be done, and the date when it is actually done.
15 minutes. The activities file proved to be the key to managing and evaluating study logistics by tracking the patients individually and in the aggregate. The day after assessments were performed, the coordinator spent from 15 to 45 minutes, depending upon the day's workload, recording the "done" date for each completed activity and generating a new record for subsequent activity required by the study's protocol. By listing the records of each subject who had not yet completed the data gathering process, the coordinator could review exactly what $had been completed for that subject and what was currently scheduled to be done determining how well the study protocol was being followed.
Two reporting programs were written in the dBASE II command language. The first simply prints out individual master file records, and the second merges the activity file with the master file by the patient's study ID to print each assessor's pending assignments for a specified period of time. Figure 2 shows sample output from the latter program, displaying one assessor's assignments for a two-week period. Each line in the list tells the ID and name of a patient to be seen, where the patient is, the type of activity to be performed, and date on which that activity should be performed. At a glance this output shows an assessor the due or past due activities for a time period and reminds him or her to contact patients who have left the hospital when follow-up visits need to be scheduled.
The ability to create on demand a list or patients who met a certain criterion, e.g., all patients assigned to a given assessor who were still awaiting home visits or all patients who had died, facilitated the monitoring of emerging problems and the development of strategies for maximizing the number of completed cases. For example, a report generated at the end of nine months used the master file to compare age and sex characteristics of patients who consented to participate in the study with non-participants, demonstrating the absence of systematic selection bias. A report generated from the activities file documented the average number of assessments per patient and the workload of each assessor to date; in light of this information several changes were made in the process of assigning assessors to distribute workloads more appropriately. The analysis of activities also showed that almost 20 percent of the subjects dropped out of the study, mostly during the period after discharge and before the first home visit; in light of this finding the method of arranging home visits was count of
Additional reports can be produced using dBASE II's nonprocedural query capabilities as illustrated in Figure 3. The LIST ALL FOR ID=74 query shows all activities for the patient with a study ID of 74. Each activity record consists of a record number, the patient's study ID, the initials of the assigned assessor, the type of activity to be performed, the date on which the activity should be performed, and the date on which the activity was actually done. Similar reports can be generated for other purposes, e.g., to list all pending activities for a researcher (LIST ALL FOR ASSESSOR="PP" .AND. DONE"83/O1" .AND. DONE
