IJCS | Volume 32, Nº5, September/October 2019

440 Gontijo et al. Distress evaluation (stop-d) brazilian version Int J Cardiovasc Sci. 2019;32(5):438-446 Original Article We obtained the permission of the first author of the STOP-D, Dr. Quincy Young, to have it translated and adapted to the Brazilian culture. This process was conducted by four independent translators, specialists in the field of health psychology. After two translation steps, the Brazilian version of the STOP-D was administered in four patients with different educational levels at the outpatient department and wards. These patients answered the questionnaire and evaluated the instrument. Finally, the back-translation was performed by two independent translators, who translated the instrument from Portuguese to English. Both translated versions were sent to Dr Quincy Young, for her analysis. The steps of evaluation of the instrument by the target public and of back-translation are essential for adequate adaptation of the instrument, as they guarantee both conceptual and idiomatic equivalence. 21,22 The Brazilian version of the STOP-D is composed of five items that evaluate markers of distress – depression, anxiety, stress, anger, and quality of social support – each one rated on a 10-point (0 to 9) ordinal scale. This instrument was developed to be used by an interprofessional staff, and it is brief and free to use. • Hospital Anxiety and Depression Scale (HADS): 14-item scale that evaluates anxiety and depression using a Likert-type scale. The instrument can also be used as a single-factor tool to measure distress, using a score ≥ 15 as the cut-off. 23 The HADS has a mean application time of four minutes, good specificity and sensitivity, and is free to use. 24 Data analysis Participants were invited to answer the questionnaire at the waiting room while waiting to be seen by the physician, or on the bed in case of inpatients. The instrument was answered with the help of one of the investigators, who is a qualified researcher in psychology, who also explained the aims of the study. All participants signed an informed consent form. Statistical analysis Descriptive and inferential statistics were performed using the Statistical Package for the Social Sciences (SPSS) software, version 22. The prerequisites for performing multivariate analysis (atypical data, missing data, normal distribution of the variables, multicollinearity, linearity, homoscedasticity and singularity). Decisions on inclusion and exclusion of data based were also made on recommendations by Tabachnick & Fidell. 25 The Kolmogorov-Smirnov test was used to verify the normality of the Brazilian version of the STOP-D scores and the HADS. The present study was divided in two phases: First phase: Evaluation of psychometric characteristics of STOP-D for validity evidence by exploratory factor analysis. The method consists in describing the correlation structure between variables based on the number of non-observable variables (latent variables). 26,27 The initial sample was randomly divided into two subgroups, of approximately the same number, using a random number generator. The number of both subgroups was sufficient to perform factor analysis, since a proportion of ten participants to each item of the scale was maintained. 25,26 Formation of these subgroups made it possible to perform both an exploratory factor analysis (n = 69 patients) to evaluate the psychometric characteristics of the instrument and a confirmatory factor analysis (n = 75), to assess the stability of the factorial structure of the Brazilian version of the STOP-D. 26 Factor analysis can be divided into four stages: (1) factorability analysis of the correlation matrix, performed by adequacy of the variance caused by the sample (KMO -Kaiser-Meyer-Olkin) 25 and correlation between variables by the Bartlett’s test of sphericity tests; (2) determination of the number of factors to be extracted, performed based on factor retention criteria – factors that overcome the variance, (i.e., eigenvalues > 1.00) were maintained. 27 The number of factors was confirmed by analysis of internal consistency using the Cronbach’s alpha (> 0.70); 25 (3) extraction of factors, conducted by factor rotation, in which the values of the highest factor loadings are put in evidence. However, in a single-factor analysis, it is not necessary to perform rotations, but rather use the criteria for maintenance of the variables. 27 For the present study, factor loadings greater than 0.40 were considered for analysis; and (4) interpretation of the factor – in this phase, the factorial structure obtained was compared with the theoretical model proposed. Although the exploratory analysis allows a higher degree of leniency, we adopted the same criteria to both exploratory and confirmatory analyses. Second phase: (n = 144) analysis of the STOP-D score accuracy by the ROC (Receiver Operating Characteristics) curve, using the HADS as reference. The STOP-D score calculation was made by summation of the scores, which