The Consortium to Investigate Vascular Impairment of Cognition: Methods and first findings

Background: The Consortium to Investigate Vascular Impairment of Cognition (CIVIC) is a Canadian, multi-centre, clinic-based prospective cohort study of patients with Vascular Cognitive Impairment (VCI). We report its organization and the impact of diagnostic criteria on the study of VCI. Methods: Nine memory disability clinics enrolled patients and recorded their usual investigations and care. A case report form included all vascular dementia (VaD) individual criteria for each of four sets (National Institute of Neurological Disorders and Stroke (NINDS-AIREN), Alzheimer's Disease Diagnostic Treatment Centers (ADDTC), the ICD-10 Classification of Mental and Behavioural Disorders (ICD-10), and the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)) of consensus-based diagnostic criteria and for the Hachinski Ischemia Score (HIS). Investigators, having completed the case report form, were asked to make a clinical judgement about the cognitive diagnosis based on the best available information, including neuroimaging. Results: Of 1,347 patients (mean age 72 years; 56% women), 846 (63%) were diagnosed with dementia and 324 (24%) were diagnosed with VCI. The proportion of patients diagnosed with VaD by the diagnostic criteria was: 23.9% (n=322) by DSM-IV, 10.2% (n=137) by HIS, 4.3% (n=58) by ICD-10, 3.8% (n=51) by ADTCC, and 3.6% (n=48) by NINDS-AIREN. Judged against a clinical diagnosis of VaD, the sensitivity/specificity of each was: DSM-IV (0.77/0.80); HIS (0.41/0.92); ICD-10 (0.29/0.98); ADTCC (0.24/0.98); NINDS-AIREN (0.42/0.995). Compared with a clinical diagnosis of VCI, sensitivities were lower for the diagnostic criteria, reflecting the exclusion of patients who did not have dementia. Conclusions: Consensus-based criteria for VaD omit patients who do not meet dementia criteria that are modeled on Alzheimer's disease. Even for patients who do, the proportion identified with VaD varies widely. Criteria based on empirical analyses need to be developed and validated.