Visualizing the quality of partially accruing data for use in decision making

Area of Science:

• Public Health Surveillance
• Health Informatics
• Data Quality Management

Background:

• Secondary use of clinical health data for public health surveillance faces challenges due to data quality issues.
• Timeliness, accuracy, and completeness are critical for effective real-time surveillance data.
• Aggregated data are often used for privacy, complicating the assessment of data quality.

Purpose of the Study:

• To disseminate data quality tools developed for the Distribute project, a near real-time influenza-like-illness (ILI) surveillance system.
• To provide insights into data quality problems inherent in aggregated, secondary clinical health data.
• To address the challenges in utilizing such data for public health surveillance.

Main Methods:

• Development of data quality tools through an end-user-as-developer paradigm.
• Exploratory data analysis to understand structural aspects of data quality.
• Development of visualization methods focusing on accrual lag and its impact on data quality dimensions.

Main Results:

• Data quality in partially accruing datasets must be analyzed within the context of accrual lag.
• Accrual lag, the delay between event occurrence and data availability, significantly impacts data quality.
• Visualization methods effectively illustrate the trade-off between timeliness and completeness, and the effects of accrual lag on accuracy.

Conclusions:

• Accounting for accrual lag is essential for accurate interpretation of trends in partially accruing data.
• Failure to account for accrual lag can lead to misleading or biased conclusions in public health surveillance.
• The developed tools offer general applicability to systems accumulating aggregate data over time.