The amount of data generated by health institutions is abundant and obtaining knowledge and insightsfrom such data is a major challenge in the process of digital transformation in the area of healthcare, due to the factthat such data are complex, high-dimensional and heterogeneous. This work-in-progress has as main goal tointegrate two Brazilian health database in order to improve the data quality regarding tuberculosis death information.For that, we applied a phonetic encoding technique (Soundex) and a pattern matching recognition (Jaro), andcompare the results.