IJCS | Volume 32, Nº5, September/October 2019

441 Table 1 - Psychometric values – factorial loading and communalities of confirmatory and exploratory analysis of the Brazilian version of the STOP-D Item Confirmatory Exploratory Factorial loading Communality Factorial loading Communality Feeling sad, down, or uninterested in life? 0.664 0.441 0.751 0.564 Feeling anxious or nervous? 0.564 0.318 0.720 0.518 Feeling stressed? 0.772 0.596 0.782 0.612 Feeling angry? 0.710 0.503 0.668 0.435 Not having the social support (family members and friends) you feel you need? 0.483 0.234 0.544 0.296 Eigenvalue 2.092 2.425 Explained variation (%) 41.84 48.503 Internal consistency coefficient ( α ) 0.780 0.726 α : Cronbach’s alpha: in the exploratory analysis, 69 heart disease patients were evaluated. Gontijo et al. Distress evaluation (stop-d) brazilian version Int J Cardiovasc Sci. 2019;32(5):438-446 Original Article allows rapid corrections for future applications. AHADS ≥ 15 was used as cut-off. The level of significance was set at 5% (0.05). Results A total of 114 patients aged from 18 to 84 years (mean of 55.85 ± 14.55 years) were studied. Most participants were women (n = 85, 59.0%). Ninety-four (65.3%) lived in the city the hospital was located. Regarding the marital status, only 55 (38.20%) were single. With respect to educational attainment, most participants had some elementary education (n = 66, 45.8%) and 31 (21.5%) had completed high school. The source of income for most patients was the pension (n = 61, 42.4%), with a monthly income of up to one minimumwage (n = 62; 41.3%). Most patients reported to have religious beliefs (n= 129, 89.7%) and religious practice was reported by 95 patients (66.0%). Most patients were seen at the outpatient department (n = 122, 84.7%); the majority had cardiac arrhythmia (n = 51; 35.4%), followed by heart failure (n = 34, 23.6%). Eighty-eight (n = 61.1%) were older than 60 years, and 72 (50.0%) had comorbidities. Properties of the Brazilian version of the STOP-D For validation of an instrument, a minimum of five patients is required per variable. Therefore, for validation of the Brazilian version of the STOP-D, only 25 patients would be needed, but we opted for a larger sample to perform concurrent validation. 28 Construct validity was established by exploratory and confirmatory factor analysis (Table 1). Factorability analysis of the correlationmatrix did not reveal any factor that would make the factorial solution unfeasible. The KMO identifies sampling adequacy, with values ranging from 0.728 to 0.729. TEB showed significant differences (p < 0.001) between correlation matrix and identity matrix, corroborating the evidence of factorability of the matrix. As predicted by the theory, the instrument showed a single-factor nature, confirmed by the criteria of eigenvalues, in which only the first component of the scale had a value greater than 1. Reliability index of the instrument corroborated the choice of a unique factor, ranging from 0.782 to 0;726 (Table 1). The answers of all patients (n = 144) were used for the analysis of sensitivity and specificity of the STOP-D, as well as the cut-off for detection of distress by the instrument; the results were measured using the ROC curve. The adoption of a cut-off of 15 resulted in a specificity greater than 92.9% and specificity greater than 32.2% (Figure 1). In the present study, we obtained an area under the ROC curve of 0.85%, representing 85% of accuracy. Therefore, the Brazilian version of the STOP-D showed a diagnostic ability of 85%.