An approach to health data validation
โ Scribed by Lewis B. Wolfenson; Norman Worth
- Book ID
- 103052491
- Publisher
- Elsevier Science
- Year
- 1980
- Tongue
- English
- Weight
- 611 KB
- Volume
- 13
- Category
- Article
- ISSN
- 0010-4809
No coin nor oath required. For personal study only.
โฆ Synopsis
To identify occupational disease and injury patterns, case records from existing computer systems were assembled into a data base. A single computer program detects errors in all of these diversly formatted files. To detect errors in a data file (i.e., to validate the data), the user supplies specifications defining a set or range of possible values or normal values. Numerous controls detect erroneous specifications by the user. The program produces a list of errors in the data and a file to be used for semiautomatic correction (i.e., editing) of the errors. The program has broad application.
To identify those accidents and illnesses which are most common in agriculture, workers compensation data were collected. To date, eight states have provided these data in machine-readable form. It is anticipated that each year, more states will automate these data and will become eligible for the study. As Burkart and deGroot (I) imply, conflicting coding schemes and geographical maldistribution prevent the data from being pooled. Thus, analysis was done on a state-by-state basis.
The analytical tool selected was SPSS [Statistical Package for the Social Sciences (2)]. As SPSS is batch oriented, a magnetic tape file management system was designed for the data. Burkart, deGroot, and Wolfenson describe the file management system in an earlier work (3). Validation-the identification of errors in the data-is one of the functions of the system. It was required that the file management system handle at least two types of data-possibly describing diseases, injuries, or fatal accidents-arriving from each of 50 states. Thus, it was expected that eventually there would be data in 100 diEerent formats. It was desirable to avoid having to write one validation program for each file because it would then be necessary to document and maintain each program. It was preferable to write a single program capable of validating all of the files.
THEORY
Validation is appropriately performed after each of several events. Upon the arrival of data to be added to the data base, the new data should be checked * Work performed under Contract CDC-99-74-110 for the National Institute for Occupational Safety and Health.
๐ SIMILAR VOLUMES