Date of Award

6-13-2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Information Science

First Advisor

John R.Talburt

Second Advisor

Meredith Zozus

Abstract

Measuring and managing data quality in healthcare has remained largely uncharted territory. Rule-based data quality assessment in healthcare was explored through compilation of 63,397 data quality rules. This work applied a design science framework to design, demonstrate and test a scalable framework with which data quality rules can be managed, executed, and used in healthcare facilities for data quality assessment and monitoring. Rules were acquired by multiple methods including iterative design meetings with stakeholders. The system was validated using a mixed methods approach. System function was validated through iterative testing to assure that the system was working properly. Rule performance was assessed by checking that rule output was within expected ranges through quantification of discrepancies identified by each rule. Deficiencies in system and rule operation were corrected during development and testing. The challenge however lies in organizational and individual acceptance of the approach. Thus, the following two questions were pursued (1) whether the approach can identify data errors of concern to clinicians, system owners, hospital administrators, and researchers, and (2) whether the findings prompt corrective action. These questions around organizational acceptance of the approach were validated through interviews with primary data users (physicians), information system owners, and secondary data users (researchers). The study found that stakeholders valued identification of different types of errors. Clinicians valued those signifying questionable practice or potentially impacting care while secondary data users most valued identification of outliers. All participants identified subsequent follow-up action based on the rule results. All participants were willing to attend follow-up meetings to see results if identified follow-up actions. Results indicate organizational interest in and acceptance of healthcare data quality monitoring. While there is significant additional work to be done in this area, the exploration of the system design and validation here shows rules-based data quality assessment and monitoring in healthcare facilities to be possible and scalable.

Download

Share

COinS