IJCS | Volume 31, Nº4, July / August 2018

416 Guerra et al. Screening: depression in heart failure Int J Cardiovasc Sci. 2018;31(4)414-421 Original Article severely depressed patients, scores between 18 and 24 points characterize patients moderately depressed, and scores between 7 and 17 points characterize patients with mild depression. 17,19 The items of the HAM-D scale focus on somatic symptoms (28%), cognitive symptoms (28%), motor symptoms (12%), anxiety (16%), mood (8%), and social symptoms (8%). 19 This scale has been validated by several studies comparing the scores in groups of patients with diseases of different severity, including HF. 10 Beck Depression Inventory-II The BDI-II 12 has been validated in psychiatric inpatients and outpatients by comparison with the HAM-D, in which the BDI-II has been shown to be more effective. 12 The BDI-II is a self-rating instrument that screens for the presence of depressive symptoms using a scoring scale comprising 21 questions with four answer choices ranging from zero to three, ordered according to severity. To fill out the inventory, the individual selects the option that best fits the way he feels at the moment, with the scores ranging from zero (absence of symptoms) to three (highest intensity of the symptom). 21 The overall BDI-II assessment is done by adding the numbers marked beside each question. A sum of 0-9 is considered normal, while 10-15 indicates mild depression, 16-23 indicates moderate depression, and 24 or more points indicate severe depression. The screening properties of the BDI-II have been validated in Brazil. This scale is considered the gold- standard method for depression screening and has demonstrated good psychometric and operational characteristics. 3,10,12,16,21 Patient Health Questionnaire-9 (PHQ-9) The PHQ-9 evaluates the presence of depressive symptoms using a Likert-like scale comprising nine questions categorized into four response options ranging from “not at all” (zero points) to “nearly every day” (3 points) and a total score ranging fromzero to 27 points. Thus, the higher the score, the worse the severity of the depressive signs. The PHQ-9 is a quick self-rating instrument that screens individuals at higher risk of presenting a major depressive episode. The screening properties of this questionnaire were validated in the general Brazilian population in 2013, 22 and the questionnaire has demonstrated good psychometric and operational characteristics, with a sensitivity of 77% to 98% and specificity of 75% to 80%, having also been validated for the population of adults and elderly individuals. 1,23,24 Procedures for data analysis The descriptive analysis characterized the studied population according to frequencies and proportions. For quantitative variables, we used the statistics of minimum, maximum, mean, median, values, standard deviation, and coefficient of variation (CV). The variability of the distribution of a quantitative variable was considered low if CV < 0.20, moderate if 0.20 ≤ CV < 0.40, and high if CV ≥ 0.40. In order to investigate the significance of the association between two qualitative variables, we used the chi-square test, and when this proved inconclusive, we applied Fisher’s exact test for 2 x 2 tables. 25 In the inferential analysis, complementary proportions were compared using the binomial test. The hypothesis of normality of a quantitative variable distribution was verified by the Kolmogorov-Smirnov (KS) and Shapiro- Wilk (SW) tests. The distribution was considered normal if both tests did not reject the null hypothesis of normality. In order to compare two independent groups when the variable had a normal distribution, we used unpaired Student’s t test; otherwise, the comparison between the groups was performed with the Mann- Whitney non-parametric test. 25 The agreement of the three instruments in diagnosing depression in the HF population was analyzed by Fleiss’ kappa (k F ) and Krippendorff’s alpha coefficients (C k ). The consistency of the three instruments in diagnosing depression in the HF population was assessed by Cronbach’s alpha coefficient (C c ). Agreement and consistent were evaluated according to the Landis & Koch classification (1977). We calculated the accuracy, sensitivity, specificity, and rates of false-positive and false-negative results of the HAM-D and PHQ-9 instruments considering the BDI-II as the gold-standard instrument in the diagnosis of depression (the definition of these concepts can be found in Medronho et al., 2009). The statistical analysis was performed using the programStatistical Package for the Social Sciences (SPSS), version 20.0, and all analyseswere carried out considering a maximum significance level of 5.0%.