The COVID-19 pandemic necessitated the rapid establishment of the COVID-19 mortality epidemiological surveillance database (SURV) in Belgium due to a significant delay in availability of the cause of death database (COD). Understanding differences and limitations in both databases is crucial for contextualising COVID-19 mortality statistics. This study assesses SURV’s data quality, raises awareness of differences and limitations in both databases, and proposes recommendations for future pandemic mortality surveillance.

SURV and COD were linked probabilistically to explore overall coverages and discrepancies. Factors such as region and place of death, case classification, epidemic wave, age group, sex, and number of conditions, were analysed using logistic regression models.

SURV identified 90% (n=19,801) of COVID-19-related deaths from COD (n=22,015). Coverage was higher in hospitals (98%, n=11,130 in SURV, n=11,335 in COD) and long-term care facilities (90%, n=8,602 in SURV, n=9,580 in COD) compared to deaths at home (5%, n=52 in SURV, n=1,057 in COD). However, 83.9% of SURV records listed COVID-19 as the underlying cause of death in COD, and 75.4% of COVID-19 deaths in COD were identified in SURV. Reduced COVID-19 activity and diagnostic uncertainty resulted in lower agreement between databases. Variations in data quality were observed across epidemic waves, regions, and healthcare facilities.

In addition to reaching real-time objectives, SURV exhibited good data quality with limited discrepancies, but underreported COVID-19 deaths at home. Presuming neither database can be unequivocally considered as gold standard for COVID-19 mortality statistics, they provide valuable insights for policy formulation. Improving real-time mortality data collection is crucial, emphasising the need for effective collaboration among stakeholders