Quality Improvement of Data Management Process Among Hospitals in Narathiwat: Case Studies on Indicators of Rational Use of Antibiotics

Abstract

Objectives: To identify the causes of inconsistencies in the indicators of antibiotic rational reported from the information system of the Ministry of Public Health (Health Data Center: HDC) and those from the ad hoc reporting system developed in the area; and to develop a systematic approach to data management to be mutually accepted in order to resolve data inconsistencies. Methods: This research was an action research divided into 3 phases. Phase 1 involved the identification of causes of the inconsistencies by analyzing secondary data, review of relevant documents, and in-depth interviews with those in charge of the indicators and system administrators. Phase 2 was the development of sustainable data management approaches through group meetings with those in charge of the indicators and system administrators. Phase 3 was the evaluation of the developed approaches by its implementation on voluntary hospitals. Results: Most of the inconsistencies in the indicator values in the study occurred in the process of importing raw data, both drug codes and disease codes, leading to discrepancies in values after processing in the HDC system. The ad hoc reporting system had additional raw data cleaning procedure by those in charge of the indicators. The procedure to filter out some of the raw data before processing the indicators became additional cause of the inconsistency due to human error. In addition, the programming code used to process indicators was not updated when the definitions of indicators were changed. The HDC system had no such procedures for filtering raw data by human, and its programming code was updated according to changed definitions of indicators. The main approach of data management to resolve such problems in the study was to have a single reporting system, the HDC system, by developing procedures to improve the quality of data input. Implementation of the developed data management practices led to a higher volume and quality of raw data entering to the HDC. Drug coding errors were reduced to 0, while ICD-10 coding errors were reduced. Time required for reviewing medical records was reduced from 7 days to 4-5 days. Delays in data transmission was reduced from 21-30 days to 5-7 days. Conclusion: The main cause of the inconsistency of indicator report in the study was in the process of importing raw data into the hospital database. This led to the development of a parallel reporting system that required the data cleaning to filter out some of the raw data before processing. The process was prone to human error and was a burdensome and redundant work. The development of a systematic approach to manage the quality of upstream raw data was well accepted by all parties involved leading to one reporting system that was an effective solution to the problem.