Objective: The Australian National Death Index (NDI) provides a comprehensive and accessible source of mortality information for epidemiological research. Use of the index requires a probabilistic matching process that inevitably results in some inaccuracy. In this paper, accuracy is assessed.
Methods: Results of a matching process against the NDI performed by the Australian Institute of Health and Welfare in Canberra were compared with information provided by the Medical Device Outcomes study cohort and their families (n = 2,990). Indices of accuracy for the NDI were calculated.
Results: For this particular study, the NDI has sensitivity 88.8% (84.9-92.8) and specificity 98.2% (97.4-98.7).
Conclusions and implications: The relatively low sensitivity is of some concern to those using the NDI for health outcomes research. The importance of such a national database is evident; however, to improve accuracy the introduction of a national unique patient identifier is necessary.