Analysis of erroneous data entries in paper based and electronic data collection

Area of Science:

• Biomedical research methodology
• Health informatics
• Data management in clinical trials

Background:

• Paper-based data collection (PBDC) has traditionally been used in biomedical research, but faces challenges with data accuracy and completeness.
• Electronic data collection (EDC) offers a digital alternative, potentially improving data quality and efficiency.
• Resource-poor settings often rely on PBDC, highlighting the need to evaluate EDC's feasibility and benefits in these environments.

Purpose of the Study:

• To compare the accuracy and completeness of data collected via EDC versus PBDC in a biomedical survey conducted in Nepal.
• To identify specific types of data entry errors and their prevalence in both data collection systems.
• To assess the suitability of EDC as an alternative to PBDC, particularly in resource-limited research settings.

Main Methods:

• A comparative study was conducted in Nepal, collecting data from 358 participants using both EDC and PBDC systems simultaneously.
• Collected data were rigorously checked for completeness, realistic value ranges, internal logic, and date consistency.
• Variables were categorized, and discrepancies between EDC and PBDC datasets were quantified and analyzed overall and by category.

Main Results:

• Discrepancies were found in 12.6% of all entries between the two systems.
• Data omissions constituted 64% of all discrepancies, with categorical variables being most affected.
• Paper-based data collection (PBDC) exhibited twice the rate of data omissions compared to electronic data collection (EDC) (p < 0.001).

Conclusions:

• Electronic data collection (EDC) is a more effective method for reducing data omissions than paper-based data collection (PBDC), especially for categorical data.
• EDC systems, when properly designed, can significantly improve data quality and reliability in biomedical research.
• The findings support the adoption of EDC, even in resource-poor settings, to enhance the integrity of research data.