Evaluation of the National Minimum Dataset for a diagnosis of dementia

To identify health-related outcomes from large administrative datasets there must fi rst be an evaluation of data concordance between the comprehensive clinical information and data collections. The International Resident Assessment Instrument (interRAI) is a validated source for capturing large repositories of valuable patient-specifi c information particularly for individuals with neurological conditions. Our study examined the degree of consensus between the National Minimum Dataset (NMDS) and the interRAI in capturing a diagnosis of Alzheimer's and other dementias. De-identifi ed NMDS International Statistical Classifi cation of Diseases and Related Health Problems, Tenth Revision, Australian Modifi cation (ICD-10-AM) coded data from 1 September 2012 to 30 June 2014 was matched with the interRAI Home Care (HC) assessment records. A diagnosis of Alzheimer's and other dementias was compared for each individual present in both the clinical and administrative records in the 90 days preceding and subsequent to the date of diagnosis in the interRAI-HC. Diagnostic cogency was measured through sensitivity, specifi city, positive predictive value (PPV), weighted kappa analyses and McNemar's test. In the two large study samples (NMDS: n=62,584 and interRAI: n=22,656) the NMDS demonstrated moderate and signifi cant agreement in capturing a diagnosis of Alzheimer's and other dementias when compared to the interRAI within three months. InterRAI assessments captured more diagnosis compared to the NMDS. There was 64.45% (95% CI: 62.30-66.61) sensitivity, 97.58% (95% CI: 97.38-97.78) specifi city and the PPV was 77.53%. Weighted kappa coeffi - cient (kappa=0.67, 95% CI: 0.65-0.69) and the McNemar's test was signifi cant at P<0.000. Routinely collected administrative datasets such as the NMDS can be a valuable source for research to impact evaluation and inform care planning, resource allocation decisions and in predicting adverse health outcomes.

AM