County Birth-Record Completeness in Public U.S. Natality Data: A Suppression-Aware County Index, 2016–2024

Birth certificates are the foundation of U.S. natality statistics, yet the share of records with an item left unknown varies across reporting areas, and public data identify counties only when they have at least 100,000 residents. We found no reproducible public measure of item completeness at the county level. We present the County Birth-Record Completeness Index (CBRCI), built from CDC WONDER natality data for 12 birth-certificate items in the 570 counties WONDER identifies, pooled over 2016-2024 and in three 3-year windows. WONDER suppresses counts of 1-9, and suppression concentrates in the best-reporting counties, so CBRCI treats every suppressed "unknown" count as interval-censored rather than substituting a value. It combines interval bounds, truncated beta-binomial multiple imputation and an interval-censored beta-binomial likelihood, decides whether a single composite is justified with pre-specified reliability and rank-stability criteria, and enforces a fail-closed disclosure audit on every released file. Pooled unknown rates ranged from 0.06% (delivery method) to 3.2% (maternal weight gain); 12.9% of county-item cells remained suppressed after pooling; and 44-74% of county variance lay between states. A composite was not supported once state differences were removed, so the index is released as three domain scores. Within states, no association with provider supply, shortage designation, uninsurance or routine checkups survived false-discovery control across 72 tests. In simulation, the interval-censored likelihood was unbiased with 0.94-0.96 coverage, whereas dropping suppressed cells biased estimates by up to 0.19 on the log-odds scale. State estimates agreed closely with NCHS published percentages (Spearman 0.86-0.97 for 10 of 12 items).