Patterns of Comorbidity and Sociodemographic and Psychosocial Correlates Among People Living With HIV in South Carolina, USA

X Yang X Li; S Qiao


HIV Medicine. 2020;21(4):205-216. 

In This Article


Background Characteristics

Participants were asked to provide information on sociodemographics, such as age, race, ethnicity, education level attained, marital status, sexual preference, employment status, annual household income, incarceration history, HIV infection duration and the most recent viral load.

Comorbidity Patterns

The data on comorbidities were collected via self-report. The patients were asked the following question: 'Have you ever been diagnosed with the following conditions?' The participants had to tick boxes to report the diagnosis of each condition (yes/no) in a predefined list, which included six STIs and three chronic conditions. The STI conditions included being seropositive for hepatitis viruses (regardless of type), human papillomavirus (HPV)/genital warts, herpes simplex virus (HSV), syphilis, gonorrhoea and chlamydia. These six types of STI are the most common types of comorbidity identified by the US CDC.[43] The noninfectious chronic comorbidities included in the survey were diabetes, cancers, and cardiovascular disease. These chronic comorbid conditions were selected because of their high prevalence among HIV-infected patients.[44,45] Given the constraints of the length of the questionnaire, we did not specify every type of certain diseases (e.g. types of cardiovascular disease and cancers) and did not include other potentially important comorbidities (e.g. hypertension and hyperlipidaemia) either. Considering the possible low health literacy of the participants in this study setting, AIDS-defining cancers were not distinguished from non-AIDS-defining cancers. For the purpose of data analysis in this study, participants were grouped by whether they reported STI comorbidities (yes/no), noninfectious chronic comorbidities (yes/no), or any comorbidity (yes/no).

Psychosocial Variables

Depression. Depressive symptoms were measured using the nine-item Patient Health Questionnaire Scale (PHQ-9).[46] The response categories of each item were 0 ('Not at all'), 1 ('Several days'), 2 ('Over half the days'), or 3 ('Nearly every day'), resulting in a possible total sum score of 0–27. This sum score has been commonly used in previous studies.[47] The scale has been found to have high sensitivity and specificity in meta-analytic reviews of previous studies[48] and has been used in PLWH in the USA.[47] The Cronbach alpha was 0.93 in the current study. A higher total sum score indicates a higher level of depressive symptoms.

Anxiety. The seven-item Generalized Anxiety Disorder Scale (GAD-7) was used to measure anxiety.[49] The GAD-7 has similar response options as the PHQ-9. This scale has good reliability and validity in the general population[49] and the US men who have sex with men (MSM) population.[50] The Cronbach alpha was 0.94 in the current study. A sum score was created, with a higher score representing a higher level of anxiety.

Coping. The modified COPE Inventory Scale was employed to measure coping.[51] Five domains (four items each) of the original scale were abstracted and used in the current analysis: 'positive reinterpretation and growth' (sample item: 'I try to grow as a person as a result of the experience'), 'active coping' (sample item: 'I take direct action to get around the problem'), 'denial' (sample item: 'I refuse to believe that it has happened'), 'turning to religion' (sample item: 'I put my trust in God') and 'alcohol-drug disengagement' (sample item: 'I drink alcohol or take drugs, in order to think about it less'). Response categories for each item were on a four-point scale: 0 ('Not at all'), 1 ('A little bit'), 2 ('Somewhat') and 3 ('Quite a bit'). Following a similar approach to that used in a previous study,[52] item scores of these domains were summed, with greater sum scores indicating a greater frequency of using a particular coping strategy. The Cronbach alphas for the five domains were 0.73, 0.63, 0.81, 0.87 and 0.92, respectively.

Resilience. The modified Connor–Davidson Resilience Scale (CD-RISC) was employed to measure resilience.[53] Examples of the items in CD-RISC include 'Bounce back after illness or injury' and 'Under pressure I stay focused'. This scale has 10 items that were assessed on a five-point scale: 0 ('Not at all'), 1 ('A little bit'), 2 ('Moderately'), 3 ('Quite a bit') and 4 ('Extremely'). A summary score was created, with a higher total score reflecting greater resilience. This scale score has been commonly used in previous studies.[54] The scale exhibited a good internal reliability (Cronbach alpha = 0.94).

Statistical Analysis

The χ 2 test or t-test was used to determine the significance of differences in categorical background characteristics or mean score differences in continuous psychosocial variables between participants with and without STI comorbidity, noninfectious chronic comorbidity and any type of comorbid condition. The associations between psychosocial variables and different types of comorbidity were examined in several steps using a logistic regression model. First, each psychosocial variable was standardized using the Z-score before being entered into the analysis. Secondly, a univariate logistic regression model was used to examine the association between each psychosocial variable and HIV comorbidity type. The univariate odds ratio (OR) and respective 95% confidence interval (95% CI) were derived from each analysis. Thirdly, associations between individual psychosocial variables and HIV comorbidity type were examined adjusting for key sociodemographic variables. Fourthly, a multivariate logistic regression analysis was conducted for each comorbidity type with all the psychosocial variables and significant sociodemographic variables being included in the model. Adjusted odds ratios (aORs) and respective 95% CIs were derived from each analysis in steps 3 and 4. To address the multiple testing problem, adjusted P-values using the Bonferroni correction method were determined in steps 3 and 4. SAS software version 9.4 (SAS Institute, Inc., Cary, NC) was used to conduct all analyses.