Abstract
Reaching a diagnosis may be difficult in the initial stages of dementia, especially in low educated individuals, when informant reports may be useful. OBJECTIVES: To compare the sensitivity and specificity of the IQCODE against cognitive tests applied in clinical practice and to evaluate the possible cut-off points in Brazil. METHODS: Individuals without dementia (CDR=0; N=5), with Mild Cognitive Impairment (MCI) (CDR=0.5; N=15) and demented (CDR≥1; N=29) were evaluated using the Short IQCODE, a 16-item questionnaire applied to an informant, and on standard cognitive and functional scales. Diagnosis was reached by a consensus team with expertise in dementia, according to DSM-IV criteria, which was blind to the IQCODE results. RESULTS: IQCODE scores were positively correlated to the CDR (r=0.65, p<0.001) and negatively correlated with years of schooling (r= -0.33, p=0.021). IQCODE scores were positively correlated with CDR controlled by age and education (r=0.61, p<0.001). Linear regression showed that age was associated with the IQCODE (p=0.016) whereas education was not associated (p=0.078). IQCODE means according to the CDR classification were: CDR 0-3.37; CDR 0.5-3.75; CDR 1-4.32; CDR 2-4.61; CDR 3-5.00. The area under the ROC curve for dementia vs. controls was 0.869 (p<0.001), MCI vs. controls, 0.821 (p<0.001); and according to the groups classified by the CDR was: CDR 0.5 vs. CDR 1=0.649 (p=0.089), CDR 1 vs. CDR 2=0.779 (p=0.009), and CDR 2 vs. CDR 3=0.979 (p=0.023). CONCLUSIONS: These preliminary findings suggest that the short IQCODE can be used for the screening of MCI and dementia in Brazil.