nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
Introduction
Since the emergence of antiretroviral treatment (ART) worldwide, infection by the human immunodeficiency virus (HIV) has no longer been regarded as fatal. Until 2018, a prevalence of .18% to .32% was estimated in the general population ages 15 to 49, with new cases appearing every year. HIV causes immune dysfunction, which can become serious, progressive, and disabling when there are delays in case detection or in the absence of treatment (Censida, 2018a; Dávila, Castellanos, & García, 2015).
In addition to immunosuppression and opportunistic diseases, infection-associated comorbidities have been described, including mental disorders (Censida, 2018b). Depression has been reported as one of the main mental health problems affecting people living with HIV (PLWH), as demonstrated by a reported prevalence between 12% and 40%, obtained using various methodologies (Chichetto et al., 2020; Ciesla & Roberts, 2001; Nanni, Caruso, Mitchell, Meggiolaro, & Grassi, 2015; Vos et al., 2015; White et al., 2015; Wolff, Alvarado, & Wolff, 2010).
Depression is defined as a mood disorder characterized by an individual’s loss of the capacity to take an interest in and enjoy everyday things, expressed as sadness, loss of interest or pleasure, feelings of guilt or lack of self-esteem, sleep, or appetite disorders, feeling tired, and lack of concentration (American Psychiatric Association, 2015).
It has been reported that psychosocial factors such as stigma, social isolation, negative emotional burden, and self-stigma could be associated with certain mental health problems reported in PLWH. However, the tropism of the virus and the effects of antiretroviral treatment (ART) have also been found to be related to the presence of depression symptoms (Govender & Schlebusch, 2012; Hudelson & Cluver, 2015; Mendoza, Ramírez, & Morales, 2013).
Screening and evaluating depression symptoms in this population is important because of the increase in these symptoms, which in turn are associated with the growth of physical health problems and a decrease in functionality related to the symptoms and chronicity of the infection itself (Berger-Greenstein et al., 2007; Gonzalez, Batchelder, Psaros, & Safren, 2011; Hammond et al., 2016; Kalichman, Rompa, & Cage, 2000). Depression symptoms are also linked to lower adherence to ART, reducing the subjective feeling of well-being and worsening the perception of overall health status. More objectively, failure to adhere to treatment can lead to resistance to ART drugs, the only treatment method currently available for HIV infection (Chichetto et al., 2020; Engler, Lènàrt, Lessard, Toupin, & Lebouché, 2018; Nanni et al., 2015; Turan et al., 2019; Gebreyesus, Belay, Berhe, & Haile, 2020; Uthman, Magidson, Safren, & Nachega, 2014).
Depression symptoms and advanced stages of HIV infection (C3) may present similar symptoms: anergy, alterations in the sleep-wake cycle, appetite, psychomotor skills, libido, and cognitive function (Pala et al., 2016; Sowa, Bengtson, Gaynes, & Pence, 2016).
Some researchers (Pala et al., 2016; Sowa et al., 2016; Tsai, 2014) have questioned the use of BDI-IA in PLWH because of the high correlations between the somatic component of depression and symptomatic variables of HIV infection, such as fatigue, wasting syndrome, the number of opportunistic diseases and the CD4 cell count. These findings indicate that symptomatic carriers should be evaluated with instruments that do not include questions on somatic factors, while depression scores require different interpretations from those for the general population.
Although there are currently several validated instruments for measuring depression in Mexico, none have demonstrated their psychometric properties and utility in PLWH. We therefore obtained the reliability, construct validity, sensitivity, and specificity of a short version of the BDI-IA for PLWH patients with depressive episode, adjustment disorder, and no depression symptoms.
We objective was obtain the internal consistency, factorial structure, sensitivity, and specificity of a short version of the BDI-IA in PLWH that enables one to evaluate and distinguish cases with a depressive episode, adjustment disorder, or no depression symptoms.
Method
Study design
This is a cross-sectional, prospective, correlational study conducted between January 2016 and December 2017.
Place
Leading public health institution in the diagnosis and treatment of HIV-infected patients living in Mexico City who do not receive care from any other health system.
Measurements
The Beck Depression Inventory (BDI-IA) is a self-administered questionnaire designed to measure depression symptoms during the past week (Valdés et al., 2017). The 21-item inventory has an answer scale with four points for each item, with total scores ranging from 0 to 63 points (Beck & Steer, 1984; Beck, Steer, & Carbin, 1988).
There are hundreds of published studies on the BDI-IA confirming its psychometric properties in different samples that support its cultural equivalence (Muhktar & Oei, 2008; Steer, Clark, Beck, & Ranieri, 1999). It has numerous advantages including the fact that since it is self-applicable, once it has been adapted and validated for a population, it can be used in everyday clinical practice (Torres-Castillo, Hernández-Malpica, & Ortega-Soto, 1991). It can be administered by health personnel not involved in mental health, provided mechanisms are in place to refer patients in the event they test positive.
The BDI-IA groups these symptoms into various factors: cognitive-affective and somatic, which explain 37.41% and 6.28% of the variance respectively. Regarding reliability, internal consistency coefficients from .76 to .95 have been found, as well as (r) test-retest coefficients from .48 to .86 (Beltrán, Freyre, & Hernández-Guzmán, 2012; González-Celis Rangel, 2009; Rodríguez et al., 2012; Tsai, 2014). The BDI-IA version, standardized by Jurado et al. (1998), is still widely used in Mexico. In the validation, they reported an internal consistency of .87 for the total inventory, mean correlations with the Zung Self-Rating Depression Scale of .65, and in the factor analysis they reported three factors: negative attitudes towards oneself, performance deterioration, and somatic alteration.
We used the Mexican version of the BDI-IA (Jurado et al., 1998) and then we obtained a short version based on the results of the factor analysis using principal component analysis.
The mental health diagnosis made by the professional (a psychologist or a psychiatrist) was identified according to the criteria of the American Psychiatric Association (APA, 2015) and based on the codes in the International Classification of Diseases, Tenth Revision (ICD-10) (World Health Organization, 2000): no mental disorder symptoms, adjustment disorder symptoms, and major depressive disorder symptoms. In the case of adjustment disorder, the criteria included the emergence of emotional or behavioral symptoms in response to an identifiable stressor that occurs within three months of the presence of the stressor, clinically expressed as a) greater discomfort than expected in response to the stressor and b) significant impairment of social or work (or academic) activity. The criteria for major depressive disorder included five or more of the following symptoms for at least two weeks: low mood most of the time, anhedonia, hyporexia or hyperorexia, insomnia or hypersomnia, cognitive impairment, psychomotor impairment, anergy, ideas of disability or guilt, thoughts of death and suicidal ideation (APA, 2015).
Procedure
The study was conducted at a cutting-edge unit of the Mexico City Health Secretariat that provides outpatient care for PLWH over 18 years old. After the medical evaluation, all patients diagnosed with HIV were routinely assessed by the Mental Health service and the Antiretroviral Adherence Clinic of the unit. This assessment is randomly conducted by psychology personnel to detect any relevant psychopathology that could hamper or jeopardize the onset of antiretroviral therapy. If necessary, a psychiatric interconsultation and evaluation are undertaken. Subsequently, ART is begun by consensus in the Internal Medicine/Infectology department, based on the recommendations issued during this evaluation.
The study was conducted in two stages. The first, piloting the questionnaire, took place during the first two weeks. Patients who came to the clinic in those two weeks, from Monday to Thursday from 10 a.m. to 5 p.m., were invited to participate by filling out the questionnaire on their own. During this first phase, subjects were asked whether they understood the questions and the answer options to identify any misleading words in the application forms (Carretero-Dios & Pérez, 2005; Guillemin, 1995).
During the second phase (application), all patients who attended from the third week of January 2016 to the third week of December 2017 were invited to participate at the same time from Monday to Thursday from 10 a.m. to 5 p.m. Those who agreed to participate filled out the informed consent form guaranteeing data confidentiality and the differential handling of the cases. They then proceeded to answer the cover sheet of the questionnaire that included variables such as age, gender, years of schooling, length of time since diagnosis, last value of viral load and CD4 cells, and the BDI-IA version. Based on the calculation of the sample size of Mexican prevalence studies (Bautista-Arredondo, Colchero, Romero, Conde-Glez, & Sosa-Rubí, 2013), and considering the estimated prevalence for CDMX and the probability that a man who has sex with men could be HIV positive, a minimum of 90 observations per month was considered. The study was conducted until the minimum sample was achieved.
Statistical analysis
SPSS Statistics 25 was used for the analysis. For the description of the sample, we used central tendency and variability measures for both the demographic variables and the viral load count and CD4 cells at the time of evaluation, as well as the Student’s t test to calculate the differences in sociodemographic variables (p < .05). For the confirmatory factor analysis (CFA) with structural equations we used SPSS 22.
The data were analyzed according to the guidelines provided by Carretero-Dios and Pérez (2005), Nunnally and Bernstein, (1994) and Reyes-Lagunes and García y Barragán (2008) for psychometric assessment.
Item-by-item frequency analysis to determine whether all the answer options were attractive.
Discrimination analysis through Student’s t test with extreme quartiles. Items whose t was not significant (p > .05) were eliminated.
Item-by-item cross tables to observe the direction.
Internal Consistency Analysis.
Items that passed the internal consistency criteria were used to make an intercorrelation to determine the type of rotation to be carried out in the factor analysis.
As moderate intercorrelations were obtained, an exploratory factor analysis with orthogonal rotation was conducted. The principal components were obtained, and the Cattell breaking point was revised to determine the number of real factors. To consider an item within a factor, we considered a factorial load of ≥ .40.
Cronbach’s alpha coefficient was obtained from the total number of questions that passed the factor analysis criteria, as well as for each of the factors.
A CFA was performed using the Maximum Likelihood method. The fit of the data to the model was evaluated based on current recommendations (Byrne, 2010), using the X2/df ratio, the mean square root of the approximation error (RMSEA), with its associated confidence interval of 90%, the comparative fit index (CFI), and the Tucker-Lewis Index (TLI). Two unifactorial models and one with three factorial factors were evaluated to determine the degree of adjustment of each one to the structure of the questionnaire.
A short version of the IDB-IA was obtained with the 12 items that explained the highest percentage of the variance with the highest reliability coefficient and achieved acceptable parameters in the CFA.
The cut-off points were obtained from the distribution of the sample in the short, 12-question version (the factor with the highest percentage of explained variance). As for the mental health diagnosis, the sensitivity and specificity of those cut-off points were obtained, with the respective Receiver Operating Characteristic (ROC) curve.
Ethical considerations
Patients were invited to participate by researchers while they were waiting for this evaluation by the Mental Health Department, in keeping with the requirements of the Declaration of Helsinki 1975 for research in humans. The protocol was authorized by the Ethics Committe of the Condesa Specialized Clinic (Clínica Especializada Condesa).
Only subjects who agreed to participate answered the questionnaire, without receiving remuneration for their participation. When the questionnaires were collected, subjects were informed that, in addition to the fact that the confidentiality of their data would be guaranteed, if they needed to talk about any of the issues in the questionnaire or had any particular concern, they could make an appointment in the mental health area of the unit (World Medical Association, 2013).
Results
Out of a total of 3,380 applicants who agreed to participate, 2,050 were diagnosed with adjustment disorder symptoms, major depressive disorder symptoms, or no symptoms of mental disorder. Twenty-eight applications (1%) with more than three unanswered items were eliminated. This left 2,022 subjects (88% men) who had completed the questionnaire, with an average age of 31.9 ± 9.3 years, 12.0 ± 5.6 years of schooling and 4.5 ± 4.3 years since their confirmed diagnosis. Table 1 shows the sociodemographic characteristics, the mean years of diagnosis, and the score on the questionnaire. Statistically significant differences were only observed in years of schooling, with men reporting more years of schooling than women.
Table 1 Sociodemographic characteristics of sample studied
| Variable |
Total
N = 2022 |
Women
N = 242 |
Men
N = 1780 |
t *p < .01 |
|---|---|---|---|---|
| Age, years | 31.9 (9.3) | 28.2 (10.9) | 33.5 (15.1) | 1.349 |
| Schooling, years | 12.3 (5.6) | 9.3 (3.0) | 12.8 (9.1) | 2.050* |
| Diagnosis, years | 4.5 (4.3) | 4.2 (4.0) | 4.8 (4.8) | .124 |
| BDI-IA Score | 11.0 (10.2) | 13.0 (11.3) | 10.7 (10.0) | 1.183 |
Notes: BDI-IA: Beck Depression Inventory-IA; Mean (standard deviation).
Regarding the discrimination between items, since an average of t = 7.77 (p < .05) was obtained between the groups at either extreme, no item was eliminated because of this criterion. An average Cronbach’s α of .91 was obtained for the 21 questions. No items were removed because of this criterion.
For factor analysis, an orthogonal rotation (intercorrelations r = .40 p < .01) was used. The Kaiser-Meyer-Olkin measure of sufficiency for the sample was .929. As can be seen from Table 2, items were grouped into three factors.
Table 2 Rotated component matrix with factor loads (> .40)
| Component | |||
|---|---|---|---|
| Item | 1 | 2 | 3 |
| Sense of failure | .716 | ||
| Displeasure with oneself | .695 | ||
| Guilt | .686 | ||
| Punishment | .675 | ||
| Pessimism | .660 | ||
| Suicide ideation | .629 | ||
| Sadness | .604 | ||
| Anhedonia | .599 | ||
| Self-criticism | .573 | ||
| Indecisiveness | .516 | ||
| Crying | .411 | ||
| Tiredness | .701 | ||
| Difficulty working | .653 | ||
| Somatic work | .647 | ||
| Hiporexia | .623 | ||
| Insomnia | .549 | ||
| Weight loss | .525 | ||
| Loss of libido | .454 | ||
| Social isolation | .591 | ||
| Irritability | .585 | ||
| Change in body image | .547 | ||
The first factor (arranged from the highest to the lowest factorial load) grouped together the items of negative self-perception or general depression factor (39.7% of explained variance). The second comprised the questions referring to the somatic factor (explaining 6.01% of the variance), while the third included the item on social dysfunction (5.49% of explained variance).
As we can see in Table 3, the “negative self-perception” factor obtained a Cronbach’s α of .89, the “somatic” factor .80 and the “social dysfunction” factor .66.
Table 3 Psychometric Properties of BDI-IA in PLWH
| Factor (# items) |
Average score/
range: 0-3 |
Cronbach’s α |
Explained
variance (%) |
Item- total
correlation |
|---|---|---|---|---|
| Negative perception of self (10) | .44 | .89 | 39.7% | .52-.75 |
| Somatic (8) | .57 | .80 | 6.01% | .38-.66 |
| Social Dysfunction (3) | .65 | .66 | 5.49% | .40-.69 |
| Total BDI | .91 | 51.2% |
According to the CFA indices of the two models evaluated, the one- and three-factor models fit the data adequately. The indices can be observed in Table 4, with a Hoelter adjustment index of 465 for n = 2022.
Table 4 Factorial structures analyzed and adjustment indices obtained for the BDI-IA in PLWH
| Factorial structure | X2/df | TLI | CFI | RMSEA |
|---|---|---|---|---|
| Unifactorial: All items are grouped into a single factor | 15.83 | .919 | .923 | .070 (.068-.072) |
| Three factors: Items are grouped into three different but correlated factors | 11.54 | .929 | .938 | .053 (.051-.055) |
Items in the first factor were analyzed as if they comprised the whole questionnaire (12 versus 21 items) and the mean scores were obtained in patients diagnosed with symptoms of major depressive disorder (n = 291), adjustment disorder (n = 809), and without symptoms. (n = 922). It was found that with a cut-off point of 11 no statistically significant differences were observed between groups above and below 11 in infection variables such as viral load, CD4, or years since diagnosis, (t = 1.70 to 3.25, p = .09), although they were observed among patients with symptoms of major depressive disorder and without symptoms of mental disorder, yielding an area under the curve of 86.5% (AUC) = .865; 95% CI [.83, .90], p ≤ .001 (Figure 1).
Discussion and conclusion
This study tested the reliability and validity of the Mexican version of the BDI-IA in Mexican PLWH, and a short version was obtained based on the factor that explained the highest percentage of variance according to and supported by the CFA.
We consider that the results obtained for reliability and validity using the CFA support a short version for the screening of depression symptoms specifically for patients with HIV.
Replicating the methodology of the BDI-IA validation study for the Mexican population and analyzing the data in relation to the HIV infection variables, such as years of diagnosis, viral load, and CD4 cells was a good decision. Although we failed to find any differences in the comparison of groups based on the median age, we did report them in the years of schooling variable, in which men had more schooling. Although educational attainment has been defined as a variable that may increase the risk of acquiring HIV infection, we do not consider it to be a variable that has made it difficult to understand the questions in the case of women, who had a mean age comparable to age for having completed secondary school (Bautista-Arredondo et al., 2013).
The internal consistency coefficients of the questionnaire are comparable with the original version and the version adapted for the Mexican population, which supports the reliability of this version of the questionnaire in this HIV population. With respect to other published versions, the internal consistency coefficients of the BDI-IA were like those of the original version and slightly higher than those of the Mexican version. We obtained construct validity, thereby replicating the psychometric analysis procedure of the original version, and updated it with the current criteria, using CFA (DiPrete, Pence, Grelotti, & Gaynes, 2018). This means that we can have a tool in Mexico for screening for depression symptoms in PLWH since several authors have highlighted the need for methods that have been validated in this population, which could be used on a massive scale and incorporated into decision-making (Eshun-Wilson et al., 2018; Perazzo et al., 2020).
In this study, we obtained data on cut-off points according to the diagnoses of adjustment disorder and major depressive disorder. We consider that this finding could be beneficial for use in outpatient settings for PLWH care (DiPrete et al., 2018). A cut-off point of > 11 is suggested to be able to more specifically evaluate the depression symptoms of a PLWH who is going to begin ART, and above all, to confirm a diagnosis, provide treatment for depression by being clear about which symptoms existed prior to treatment, and be able to provide multidisciplinary management for these patients.
This is because we excluded participants with HIV symptoms compatible with an advanced stage of infection since, as we mentioned, several of the criticisms leveled at the use of both the BDI-IA and BDI-II state that the presence of symptoms of the infection could be associated with a ceiling effect in the Beck questionnaire scores, especially the somatic factor. This was one of the reasons for using the first version of the questionnaire. Since it has been studied more than the second version, there are more data and suggested methods for distinguishing somatic depression symptoms from those due to the progression of HIV. We analyzed the scores above and below the median viral load, CD4 cells and years since diagnosis, because a mixture of higher viral load and fewer defenses is associated with a more advanced degree of infection. Moreover, after crossing it with an expert’s diagnosis, we can provide a better explanation associated with HIV symptoms or the somatic symptoms of depression (Chichetto et al., 2020; Ciesla & Roberts, 2001; Nanni et al., 2015; Vos et al., 2015; White et al., 2015; Wolff, Alvarado, & Wolff, 2010).
Nevertheless, this study has certain methodological limitations, such as using a sample of participants only from Mexico City (CDMX), obtained at a specific facility, and including very few women, which could mean that the comparisons by gender are skewed and may be more representative of CDMX men living with HIV rather than PLWH in general. This suggests we should continue with this line of research in groups of women living with HIV and in groups of PLWH in other parts of Mexico. In addition to the effect of infection on mood, there are psychosocial factors such as stigma, discrimination, and social support that may have a negative impact on mood, which are variables we did not consider for the purpose of this study (Hammond et al., 2016; Chichetto et al., 2020).
Not having a random sample may also have skewed the data, since only “regular” patients were considered. In other words, these patients do not have as many psychosocial challenges as those who are unable to attend on a regular basis due to psychosocial barriers, in whom we would expect to see a greater presence of affective symptoms.
A short version is therefore proposed since it has proved reliable, valid, sensitive, and specific for use with patients with HIV and considers a cut-off point of ≥ 11 to more exhaustively evaluate symptoms that may be compatible with major depressive disorder.
