A Database for Analysis of Clinical Research Findings

6th Annual Conservative Health Science Conference
Old Brookfield, NY. October 9-11, 1987

Edward F. Owens, Jr. M.S., D.C.
Life Chiropractic College
Marietta, GA

ABSTRACT

A new data base program has been designed at Life College to help analyze the data resulting from clinical studies. The program consists of modules for entry of patient information gleaned from the history and physical exams, as well as the chiropractic exam and adjustment series.

The system has been developed on a personal computer in a special database language called QPRO-4. In this language, the data files can be indexed in a number of ways to allow quick access to information. Each patient record has been broken into 8 separate data files, each of which can be indexed and sorted separately. In this way, search times of data for analysis are decreased and flexibility of the system increased. Several analysis programs are in the planning stages to evaluate subject demographics, and to generate a time series analysis of the effects of treatment on the severity of positive findings.

INTRODUCTION

Data base management systems have been used for several years in health science to store patient information in a useable and selectively retrieveable form. Data base programs have developed in stages along with the computer industry. In 1980, data base management systems for medical applications were used on minicomputers such as the PDP 11/34, at a system cost of approximately $15,000 for the minimal computer configuration [1]. The development costs for such a database program were in excess of $100,000.

In contrast, today's microcomputers and application software can perform the same data base operations at a fraction of the cost of the older systems. It is now economically feasible to develop data base software for chiropractic clinical research projects.

In ongoing clinical studies at Life Chiropractic College we collect data from physical and chiropractic examinations of subjects and look for the effects of adjustment on the positive findings. To facilitate the process of data compilation and analysis we have designed a clinical database system which is being programmed in-house. The system has been developed on a personal computer (IBM PC compatible), in a data base language called QPRO-4 (QNE International)[2]. To date, the input modules of the system are in use, collecting data for later analysis. The analysis modules are still under development.

DATABASE ARCHITECTURE

As designed, the system is divided into three main parts: an input module for subject information, a reports and analysis module and a utility module.

DATA INPUT: The subject information module accepts the following types of information for each subject:

SUBJECT INFORMATION MODULE

The data entry screens for each of the above modules, along with their accompanying help screens are shown in Appendix A.

To allow the quickest and most flexible search and analysis of the above data, the system stores each different type of information into a separate data file. All of the data files are tied to the subject entrance data file by a subject number.

Wherever possible, the information is stored in coded form. Using codes is, in a sense, a draw-back because it limits the possible range of entries. On the other hand, it conserves storage space and lends a uniformity to the data, both of which will ultimately make analysis quicker and easier. For example, the chief complaint is described by codes for its location, duration, intensity, and modifying factors. The codes are available to the system operators in the form of Code Help screens that can be displayed during data entry (see Appendix A). During analysis, the data file can be quickly searched for the presence of certain codes and their correlation to other findings.

To store subject's medical history and the results of physical and chiropractic examinations, only positive findings are recorded. Each finding has been assigned a numeric code which is used during data entry. Where appropriate, a grade code ranging from 0 to 5 is also applied. Affixing a grade to a finding will lend greater resolution to a time series analysis of the effects of adjustment.

The grading scales are not always consistent from test to test and many tests could only be graded as positive or negative. For instance, muscle reflex tests have a standard grade ranging from 0 to 4 where 0 represents no response and 4 denotes an extreme response. Posture analysis, on the other hand, is scaled differently to reflect the range of postural deviations. Using a universal protractor to measure shoulder tilt could result in readings that might vary from -15 to +15 degrees, where negative numbers represent a left tilt and positive represents a right tilt. Fitting this range to the 0 - 4 scale produces a scale for which 2 is normal. The Physical Examination form on which the Positive Findings Codes and grading scales are based is shown in Appendix B.

In Q-PRO4, the data files can have more than one data item in the key. The key is the data item or set of items that must be unique for each record stored in the data file. For example, this means that a particular finding can be stored more than once for a particular patient, since the key contains the patient code, the finding code and the date.

The visit record file stores the results of pre- and post-adjustment chiropractic evaluation. The information on posture, palpation and procedure performed is stored in coded form.

DATA RETRIEVAL

Information stored in a relational data base such as this one can be retrieved in a number of different ways. The simplest report is a list of some data file. More complex reports can be generated by performing searches for the incidence of certain findings or conditions. Reports may also involve the analysis of data for correlation of findings, or trends that represent changes in a condition following a course of treatment.

The report module of the clinical data base is still under development. The following menu shows the proposed reports:

                           REPORT MENU
    Individuals:             Groups:                 Lists:
P. Problem List     D. Subject Demographics       F. Finding Codes     
V. Visit Record     I. Incidence of Complaints    M. Mailing
C. Psychological                                  S. Subject #'s
T. Time Series       

Of these reports, the Psychological and Time Series reports require the most complex analysis. The Psychological report is based on information gathered by a computerized administration of the Minnesota Multiphasic Personality Inventory (MMPI). The input module for the test prompts the subject to choose true or false to each of 566 statements. The resulting data is analysed to produce the number of questions that apply to each of the 13 scales used in the MMPI analysis. At this point, the psychological test is being used to screen patients in a low back study for hysteria, which might confuse the outcome of the study.

The time series report is being developed as a way to determine the effectiveness of a course of treatment on the positive findings of examination. The report will generate a plot of the grade of selected findings as a function of time for comparison with the treatment schedule. One drawback of the program development software being used is the inability to search and manipulate data files interactively. This means that a separate, dedicated program needs to be written for every report or investigation of the data. In essence then, the report module is being completed and expanded as new reports are needed.

UTILITY MODULE

Programs that are used to generate backup copies of the program and to define codes and data files are grouped into the utility module. The Menu for this module is shown below:

                         UTILITY SCREEN
From this screen you can:
     A. Backup the Program             E. Restore the Program
     B. Backup the Data Files          F. Restore the Data Files 
     C. Define Finding Codes           G. Setup Psychological Test
     D. Define Diagnosis Codes         H. Define Procedure Codes

Eventually, any data base system will fill it's storage space. For this data base, each patient record occupies an average of 12 Kilobytes on the storage disc, allowing maintenance of 500 patient files on 6 MegaBytes of a Hard disc. This system has been setup to allow selected groups of patient files to be "backed-up", or removed from the hard disc and copied onto floppy discs for permanent storage. Recovery of the data is also possible if a later analysis is needed. An added benefit of this design is the transportability of patient files from computer to computer. Field doctors or students who use the system could contribute clinical data to a common database of practical or technical information.

CONCLUSION

A clinical data base program has been designed and implemented at Life Chiropractic College. The system is being used to store and analyse the data generated in clinical studies performed in the research department. Eventually, the system will be used to compile results from the student clinic and field doctors into a data base demonstrating the effectiveness of chiropractic care for specific complaints and disorders.

Bibliography

  1. Loyd, S.C. Programming systems for medical data base management,Proceedings of Fourth Annual Symposium on Computer Applications in Medical Care, Nov, 1980, Washing. D.C.,pg 815-821.

  2. Quick-N-Easi Products, Inc., 136 Granite Hill Court, Langhorn, PA 19047