Validity and reliability of stillbirth data using linked self-reported and administrative datasets.

Area of Science:

• Epidemiology
• Reproductive Health
• Data Science

Background:

• The Australian Longitudinal Study of Women's Health (ALSWH) observed a high stillbirth rate.
• Previous studies lacked validation for self-reported stillbirth data.

Purpose of the Study:

• To assess the validity and reliability of self-reported stillbirth data.
• To compare self-reported data with administrative health records.

Main Methods:

• Linked self-reported data from ALSWH (1973-1978 cohort) with New South Wales administrative datasets (1996-2009).
• Defined true stillbirth cases by consistent recording across two or more sources.
• Calculated sensitivity, specificity, and agreement statistics.

Main Results:

• Administrative datasets demonstrated higher accuracy and agreement than self-reported data.
• Self-reported data had 100% sensitivity but 30% specificity for stillbirths.
• Data linkage identified and corrected approximately half of misreported stillbirths in ALSWH.

Conclusions:

• Data linkage is valuable for validating self-reported health data and resolving administrative data inconsistencies.
• Understanding the strengths and limitations of both data types enhances epidemiological research methods and findings interpretation.