BACKGROUND: The Birth Information Network (BIN) was established by Vermont legislation in 2003 and began collecting data in 2006 to conduct statewide, population-level surveillance of 44 structural and chromosomal birth defects, metabolic and endocrine conditions, congenital hearing loss, and very low birth weight. The BIN uses multiple data sources to identify potential cases and then conducts follow-up to confirm or rule out those cases. The program monitors trends, promotes prevention, and links families to resources. An evaluation of the birth cohorts 2006-2014 was conducted to assess the overall effectiveness of the BIN and provide recommendations for improvements.
METHODS: Case information in the BIN was evaluated using the 2001 Centers for Disease Control and Prevention’s Updated Guidelines for Evaluating Public Health Surveillance Systems. Simplicity, flexibility, and acceptability were assessed by interviewing key stakeholders and by reviewing all conversation records from parents of children meeting criteria to be accepted into the BIN. Data quality performance indicators were categorized into three levels of performance: rudimentary, essential, or optimal, using criteria from the National Birth Defects Prevention Network’s (NBDPN) Data Elements Quality Assurance Standards and Recommendations. Diagnosis confirmation status was used to calculate Predictive Positive Value (PPV). Representativeness was assessed by geographical hospital region, specific condition, and maternal demographics.
RESULTS: The BIN is not simple; it aggregates data from as many as 11 diverse primary sources, as well as medical records. The BIN is flexible, and changes to individual cases can be made easily; however, structural changes, such as altering variables to adapt to modified case definitions over time can be difficult. The BIN is widely accepted by the community, with 0.8% of parents/guardians requesting to opt-out of BIN. Per NBDPN standards, overall data quality is within expectations; Vermont achieved an essential level for all completeness and timeliness measures, and an optimal level for three of four accuracy measures. For the 2006-2014 birth cohorts 5497 provisional cases were ascertained, of which 3047 met case definitions, resulting in a PPV of 55.43% (95% CI: 54.12%-56.74%).
CONCLUSIONS: The BIN is widely accepted by the community, and exceeds NBDPN standards in most accuracy categories. However, developing a formal database quality assurance process will bring all accuracy measures to an optimal level. Expanding autofill fields in data tables will increase completeness and minimize human error. Ensuring goals in case completion times are met and increasing the PPV, without missing potential cases, will improve overall timeliness measures.