Ensuring data consistency in national residents records is crucial for accurate citizen identification, public administration, and research. In this study, we address the problem of optimizing blood type consistency in family records within the National Residents Database of Vietnam. Given inconsistencies in declared blood types, we propose an Integer Programming (IP) approach to minimize the number of corrections required while ensuring logical consistency. We compare our method with a simple Branch-and-Bound (B&B) algorithm and demonstrate its superior efficiency. Experimental results on 90 instances of varying family sizes and error rates show that the IP approach consistently outperforms B&B, maintaining stable computation times even for large problem sizes. Our findings highlight the practical applicability of integer programming in real-world data correction and validation tasks, particularly in large-scale government databases.