Deterministic record linkage in AIDS mortality databases has presented problems caused by errors in files. Thus, the aims of this study were to assess the performance of deterministic record linkage in the Aids mortality databases of the Mortality Information Improvement Program of the city of Sao Paulo (PRO-AIM) and of the SEADE Foundation between the years 2000 and 2004 and to estimate the coverage of these databases. A software merge process was used to link the records. The first stage linked automatically and, in the second stage, each database was checked for possible matches. Deaths were estimated by summing the matches and non-matches for each database coverage calculation. The first stage of record linkage identified 91.6% matches. The second stage added 457 matches. Deaths were estimated to be 5,855 cases, PRO-AIM covering 97.1% and SEADE 96%. The use of deterministic record linkage covered a great part of the cases. PRO-AIM databases provided the best case coverage, with the largest amount of complete information and better geographic location.
Mortality Information System; AIDS; Mortality Registers
