Agreement between self-report of coronary heart disease by participants in the 45 and Up Study and hospital administration data: A record linkage study

OBJECTIVE: The objective of this study was to assess the agreement, diagnostic properties and predictive ability of self-reported coronary heart disease in the 45 and Up Study using record linkage.

METHODS: Baseline Australian 45 and Up Study data (2006-2009) were linked to hospital administrative data from the New South Wales Admitted Patient Data Collection and New South Wales Emergency Department Data Collection (2005-2009). Participants were aged ≥45 years with and without self-reported coronary heart disease (n=49,844). Coronary heart disease diagnosis was determined from responses to four questions. Agreement of the individual and combined questions from the 45 and Up Study with relevant diagnostic and procedure codes from the Admitted Patient Data Collection and Emergency Department Data Collection were explored using agreement, sensitivity, specificity and predictive-ability statistics.

RESULTS: Sensitivities for the individual and combined questions varied between 16.7 and 93.9% and specificities between 74.8 and 98.1%. Kappa values ranged from 0.07 to 0.52 for the individual and combined questions. Overall, the combination of all four questions more accurately identified participants with coronary heart disease than the individual questions.

CONCLUSIONS: Self-report of coronary heart disease diagnosis using a combination of four questions was able to identify participants with coronary heart disease with moderate agreement and moderate to high sensitivity and specificity compared to available hospital administrative data collections.

IMPLICATIONS FOR PUBLIC HEALTH: Identifying participants with coronary heart disease from self-report is useful for studies investigating health outcomes in this population.