The individual,interactive,and syndemic effect of substance use,depression,education,and ethnicity on retention in HIV care

Abstract

In this study, we sought to assess the individual, syndemic, and interactive associations between individual-level factors and retention in care. The sample was derived from the Miami Adult Studies on human immunodeficiency virus (HIV)/ acquired immune deficiency syndrome (AIDS) cohort from 2009 to 2014. The variables were entered into a multiple logistic regression with retention as the outcome. Backward regression, adjusting for all main effects, was conducted to determine which two-way interactions were associated with retention. Multivariable logistic regression was used to test which number of factors were associated with retention. Non-Hispanic Black race/ethnicity was associated with improved retention (odds ratio [OR] = 2.44, 95% confidence interval [CI]: 1.06–5.75, p ≤ 0.05) when compared to Non-Hispanic White persons. Black-Hispanic and Other racial/ethnic identities were associated with increased retention (OR = 4.84, 95%CI: 1.16–25.79, p ≤ 0.05 and OR = 7.24, 95%CI: 1.54–54.05, p ≤ 0.05, respectively) when compared to Non-Hispanic White persons. The interaction between depressive symptoms and Alcohol Use Disorder Identification Test (AUDIT, a test that assesses alcohol use disorder) score was significantly and negatively associated with retention in HIV care (OR = 0.14, 95%CI: 0.01–1.11, p ≤ 0.10). The interaction between age and male gender was also negatively associated with retention (OR = 0.95, 95%CI: 0.88–1.01, p ≤ 0.10), and the interaction between male gender and depression was positively associated with retention (OR = 7.17, 95%CI: 0.84–98.49, p ≤ 0.10). In conclusion, multiple races/ethnicities, specifically Non-Hispanic Black, Black-Hispanic, and Other racial/ethnic identification, were associated with increased odds of retention. Multiple interactions, specifically depressive symptoms * alcohol use disorder and male gender * age, were negatively associated with retention. The male gender * depression interaction was positively associated with retention in HIV care.

Keywords

Epidemiology human immunodeficiency virus antiretroviral therapy

Introduction

Reduced engagement in human immunodeficiency virus (HIV) medical care has been associated with poor clinical outcomes including reduced retention, virologic failure, and mortality.¹ Multiple individual-level factors have been associated with reduced progression along the HIV care continuum.¹ Recent literature indicates that to mitigate losses of patient engagement at different stages of the continuum, it is highly important to focus on individual-level factors at each stage of the continuum.² According to the behavioral model for vulnerable populations, individual-level factors are stratified into the following categories: traditional factors, vulnerability factors, and enabling factors.³ Traditional factors include demographic information such as age, gender, education, and race/ethnicity. Vulnerability factors include factors such as mental illness and substance abuse.³ Enabling factors are those that can impede or enhance access to HIV care and include income, insurance, housing, and social support.³ Specific, individual-level factors that have been suggested to impede progress to retention include less than high school education, depressive symptoms, cocaine use, marijuana use, and alcohol use.^4–7

Retention is a pivotal component of the HIV care continuum. According to recent literature, retention is defined as two or more HIV-related physician visits or diagnostic tests (CD4 cell count or viral load tests) in the calendar year at least 3 months apart.^8,9 Recent literature suggests that individual-level traditional, enabling, and vulnerability factors are each associated with retention in HIV care.^5,10,11

The effect of multiple factors on retention is generally assessed individually; however, these factors occur in concert with one another. One type of association that may explain the multifaceted manner in which individual-level characteristics are associated with retention is syndemic or additive association. Syndemic theory, established by Merill Singer, involves a set of mutually enhancing health problems, when working together in a context of harmful social and physical conditions increase vulnerability, and significantly affect the overall disease status of a population.¹² Syndemic theory postulates that a constellation of health problems accumulates over a person’s lifespan and can amplify the negative impact of one or more health problems.¹³ Determining syndemic or additive associations involves determining the sum of multiple factors on a particular outcome.¹² These individual-level factors may also be effect modifiers in assessing retention. An interactive association among factors involves the simultaneous, non-additive influence of two or more variables on an outcome.¹⁴ Using the syndemic and interactive frameworks is an innovative way of examining the association of individual-level factors on retention in care. In this study, we sought to assess the individual, syndemic, and interactive associations between individual-level factors and retention in care.

Materials and methods

Setting and population

The sample was derived from the Miami Adult Studies on HIV/acquired immune deficiency syndrome (AIDS) (MASH) cohort. Further details related to the methodology of the original project can be found in a paper entitled, ‘Cocaine Use and Liver Disease are Associated with All-Cause Mortality in the Miami Adult Studies in HIV (MASH) Cohort’.¹⁵ This is a secondary analysis of a prospective cohort study involving 407 individuals living with HIV and residing in the south Florida region. Participants were a subset of the more than 800 participants from the MASH cohort who have been followed from 2009 to 2014. All independent variables were assessed at screening, and participants were subsequently followed for a 1-year period after screening to determine retention status. All participants provided written consent to participate in the original study (MASH Cohort Study) consistent with the Florida International University Institutional Review Board IRB: 13–0038.

Eligibility criteria

Participants were eligible to participate if they were (1) ≥18 years of age, (2) diagnosed with HIV by a standardized testing method, (3) participated in the study from 2009 to 2014, (4) provided informed consent, (5) and completed the screening questionnaire.

Measures

The primary outcome of this study was retention in care, defined as two or more HIV-related physician visits or diagnostic tests (CD4 cell count or viral load tests) during the calendar year at least 3 months apart.^8,9 Those who met this criterion were considered ‘retained’, those who did not, were codified as ‘not retained’.

Age, race/ethnicity, and gender were determined using the validated ‘Current Age, Gender, and Race Questionnaire’ survey from the Tier 1 Phen X Toolkit.¹⁶ Education was determined using the validated ‘Current Educational Attainment Questionnaire’ from the Tier 1 Phen X Toolkit.¹⁶

Age was determined in the MASH survey through self-report and calculated from the year of birth. Age was categorized in the current study as follows: ‘≤25’, ‘26–34’, ‘35–44’, and ‘≥45’, for the purpose of assessing frequencies among age groups.¹⁷ However, age was assessed as a continuous variable in all logistic regressions.¹⁸

Race/ethnicity was assessed via self-report and was categorized as follows: ‘White (Non-Hispanic)’, ‘Black (Non-Hispanic)’, ‘White (Hispanic)’, ‘Black (Hispanic)’, ‘Asian’, and ‘Other’.¹⁹ The ‘Other’ racial/ethnic categories include persons of mixed races/ethnicities and those in categories that are not listed in the survey.

Gender was defined as a categorical variable including: ‘Males’, ‘Females’, ‘Transgender’, and ‘Declined’.²⁰

Education was assessed via self-report. In the current study, educational attainment was assessed using the following categories: ‘Less than high school education’, ‘High school graduate or equivalent’, and ‘More than high school education’ for the purpose of assessing the frequencies of educational attainment of the population.²¹ However, in the logistic regressions education was assessed as a continuous variable.²²

In the MASH cohort study, the validated Center for Epidemiologic Studies-Depression Scale (CES-D) was used to assess depression at baseline.²³ A score of 16 or higher indicated symptoms suggestive of clinical depression diagnosis, while scores of 15 and below indicated that persons did not have symptoms suggestive of clinical depression diagnosis.²⁴ This variable was coded dichotomously: ‘Depressed’ and ‘Not Depressed’.²⁴

Cocaine use and marijuana use were assessed using the Multi-Drug 5 Panel Urine Dip Test Kit (WDOA-754). This test allowed researchers to assess drug use in the past 3 days. This testing method has been approved by the FDA.²⁵ Cocaine and marijuana use within the past 3 days were each categorized dichotomously (positive or negative).

Alcohol use disorder was determined from the Alcohol Use Disorder Identification Test (AUDIT) at screening; this form has been validated by Saunders et al.²⁶ This is a ten-item scale in which each question is worth one point. A score of 8 or more indicates symptoms suggestive of alcohol use disorder. Alcohol use disorder was categorized as a dichotomous variable, either as ‘having alcohol use disorder’ or ‘not having alcohol use disorder’.²⁶

Analysis

First, frequency distributions were examined for the variables (age, race/ethnicity, education, substance use status, depression status, and retention status). The bivariate association between these variables and retention was determined using a Chi square test of independence. These variables were subsequently entered into a multiple logistic regression with retention as the outcome to determine which individual-level factors were significantly associated with retention. Individual-level factors were determined to be significant at the p ≤ 0.05 threshold.²⁷ After determining the individual-level factors associated with retention, all two-way interactions among these factors were included in the logistic regression. Subsequently, backward regression, adjusting for all main effects was conducted, to determine the two-way interactions that were associated with retention. All interactions with p ≤ 0.10 were included in the final model.²⁸ The threshold of p ≤ 0.10 was used to determine exclusion of interactions as the sample size was relatively small and 0.05 would be highly exclusive to interactions of interest.²⁹ To explore the relationship between significant interactive variables and retention, stratified logistic regressions were performed among variables with significant interaction terms. Therefore, if a variable was included in an interaction term, two logistic regressions were conducted; one for those who tested positively for a variable and the other for those who tested negatively for a variable. Individual-level factors in the stratified analysis were determined to be significant at the p≤ 0.05 threshold. To determine the syndemic or additive association between independent factors and retention, researchers developed a count score of the number of individual-level factors (from 0 to 5). All variables were categorized dichotomously, whereby a score of 5 indicates that all factors are present, while a score of 0 indicates that none of the factors is present.³⁰ All persons within the sample were assigned a count score based on the number of factors that are present; these factors were indicated to be associated with reduced retention by the literature (i.e. positive for cocaine use, positive for marijuana use, less than a high school education, symptoms suggestive of depression, and positive for alcohol use disorder).^4–7 For the purpose of the syndemic analysis, all independent variables not already dichotomous were collapsed into dichotomous variables.³⁰ Education was reduced to two categories (‘graduated high school’ and ‘did not graduate high school’).³⁰ Mantel-Haenszel Chi square method was used to test for the association between the number of individual-level factors and retention within the sample. A multivariable logistic regression model was used to test which number of individual-level factors were associated with retention.³⁰ Odds ratios (ORs) were reported at 95% confidence intervals (CI) to estimate the effect of each variable and their interactions at each stage in the continuum. All statistical analyses were performed using R Version 3.4.1.

Results

A subsample of 407 MASH participants was eligible for this analysis. Table 1 summarizes retention frequencies of participants in the sample as well as their demographic characteristics. Overall, 267 (65.6%) of the persons living with HIV were retained in HIV care, while 140 (34.4%) were not retained. The majority of participants were Non-Hispanic Black persons (n = 278, 68.5%). The sample was predominately male (n = 263, 65.2%). The majority of persons were ≥35 years of age (n = 371, 91.2%). In Table 1, we also present the outcome of the bivariate association between the variables and retention. There were no significant associations.

Table 1.

Frequency of demographic characteristics and drug use among a subsample of the MASH cohort in Miami-Dade County from 2009 to 2014.

Variable	Retained (n = 267, 65.60%)	Not retained (n =140, 34.40%)	Missing	X², p-value
Traditional
	n (row %)	n (row %)
Race/Ethnicity
White	14 (50.00%)	14 (50.00%)
White (Hispanic)	42 (59.15%)	29 (40.85%)
Black	187 (67.27%)	91 (32.73%)
Black (Hispanic)	12 (80.00%)	3 (20.00%)
Other	12 (85.71%)	2 (14.29%)
Missing			1 (0.25%)	8.57, 0.07
Gender
Male	172 (65.40%)	91 (34.60%)
Female	91 (65.00%)	49 (35.00%)
Missing			4 (0.98%)	0.00, 0.99
Age (years)
≤25	5 (83.33%)	1 (16.67%)
26–34	19 (67.86%)	9 (32.14%)
35–44	76 (66.09%)	39 (33.91%)
≥45	165 (64.45%)	91 (35.55%)
Missing			2 (0.49%)	1.05, 0.79
Education
<HS	92 (68.15%)	43(33.85%)
HS	121 (66.12%)	62 (33.88%)
>HS	49 (60.49%)	32 (39.51%)
Missing			8 (1.97%)	1.35, 0.51
Vulnerability
AUDIT score
<8	145 (63.88%)	82 (36.12%)
≥8	119 (67.61%)	57 (32.39%)
Missing			4 (0.98%)	0.46, 0.50
Cocaine use
Positive	67 (68.37%)	31 (31.63%)
Negative	194 (64.67%)	106 (35.33%)
Missing			9 (2.21%)	0.30, 0.58
Marijuana use
Positive	51 (58.62%)	36 (41.38%)
Negative	210 (67.31%)	102 (32.69%)
Missing			8 (1.97%)	1.05, 0.21
Depression
Positive (CES-D score ≥16)	251 (66.40%)	127 (33.60%)
Negative (CES-D score <16)	16 (55.17%)	13 (44.83%)

Note: Chi squared estimates do not include NA observations.

AUDIT: Alcohol Use Disorder Identification Test; CES-D: Center for Epidemiologic Studies-Depression scale; HS: high school; MASH: Miami Adult Studies on human immunodeficiency virus/acquired immune deficiency syndrome.

Table 2 presents model 1 and model 2. In model 1, we summarize results of the logistic regression that assessed the association between all variables (excluding the interaction terms) and retention in HIV care. Model 1 indicates that race/ethnicity had a significant association with retention. The results suggested that Non-Hispanic Black race/ethnicity was associated with improved retention (OR = 2.44, 95%CI: 1.06–5.75, p = 0.04). Model 1 also indicates that both ‘Black-Hispanic’ and ‘Other’ racial/ethnic identities were associated with increased retention among participants (OR = 4.84, 95%CI: 1.16–25.79, p = 0.04 and OR = 7.24, 95%CI: 1.54–54.05, p = 0.02, respectively). White-Hispanic race/ethnicity was not significantly associated with retention in HIV care (OR = 1.75, 95%CI: 0.68–4.57, p = 0.24). All races/ethnicities were compared to the White Non-Hispanic reference group. Marijuana use, cocaine use, and other demographic factors (education, age, and male gender) were not associated with retention.

Table 2.

Adjusted odds ratios for retention among the MASH cohort participants in Miami-Dade, FL, from 2009 to 2014, N = 407.

Variable	Model 1		Model 2
	AOR	95%CI	AOR	95%CI
Vulnerability
Depression
CES-D score <15	1.00		1.00
CES-D score ≥16	0.53	0.84–4.31	1.20	0.15–6.88
Marijuana use
Negative	1.00		1.00
Positive	0.73	0.43–1.27	0.69	0.40–11.21
Alcohol abuse
AUDIT score <8	1.00		1.00
AUDIT score ≥8	1.17	0.74–1.85	7.39	1.22–68.03
Cocaine use
Negative	1.00		1.00
Positive	1.15	0.68–1.95	1.18	0.69–2.05
Traditional
Education (years)	0.91	0.82–1.01	0.90	0.81–1.00
Age (years)	0.98	0.96–1.01	1.02	0.97–1.08
Race/Ethnicity
White (Non-Hispanic)	1.00		1.00
White (Hispanic)	1.75	0.68–4.57	1.56	0.61–4.14
Black (Non-Hispanic)	2.44**	1.06–5.75	2.25	0.98–5.31
Black (Hispanic)	4.84**	1.16–25.79	5.37**	1.23–29.96
Other	7.24**	1.54–54.05	6.17**	1.34–47.94
Gender
Female	1.00		1.00
Male	0.88	0.55–1.39	1.72	0.34–76.70
Two-way interactions
Depression × AUDIT			0.14*	0.01–1.11
Depression × Gender			7.17*	0.84–98.49
Age × Gender			0.95*	0.88–1.01

**p ≤ 0.05, *p ≤ 0.10.

AOR: adjusted odds ratio; AUDIT: Alcohol Use Disorder Identification Test; CES-D: Center for Epidemiologic Studies-Depression scale; MASH: Miami Adult Studies on human immunodeficiency virus/acquired immune deficiency syndrome.

Model 2 summarizes final results of the backward regression which included all single variables and two-way interactions that had a p-value ≤0.10. The interaction between depressive symptoms and the AUDIT score was significantly and negatively associated with retention in HIV care (OR = 0.14, 95%CI: 0.01–1.11, p = 0.08). The interaction between age and male gender was also negatively associated with retention (OR = 0.95, 95%CI: 0.88–1.01, p = 0.10), and the interaction between male gender and depression was positively associated with retention (OR = 7.17, 95%CI: 0.84–98.49, p = 0.10).

Table 3 presents the outcomes for all stratified analysis (models 3–6). The stratified analysis indicated that among those who tested positive for depression, non-Hispanic Black race/ethnicity was associated with retention (OR = 2.36, 95%CI: 1.02–5.52, p = 0.05) (Model 3). Black-Hispanic race/ethnicity and Other race/ethnicity were also associated with retention when compared to their White Non-Hispanic counterparts (OR = 5.99, 95%CI: 1.25–44.70, p ≤ 0.04 and OR = 6.82, 95%CI: 1.45–50.40, p ≤ 0.03) (model 3). In model 4, we assessed the association between variables and retention among those who scored less than 16 on the CES-D scale and found no significant association (model 4). Gender was removed from this particular regression as there were only seven women in this subsample, none of whom tested positive for alcohol abuse in the non-retained group, which resulted in zero value for one of the cells. There were only 29 non-depressed persons overall, which limited the power of findings among this group. For most multiple logistic regressions to function, no cells should be zero and at least 80% should be >5; this was not the case when cross tabulating female gender with alcohol dependence.³¹ For the reasons mentioned above, the inclusion of gender in the logistic regression that assessed variables’ association with retention among the non-depressed would have been inappropriate. Among males, depression was determined to be positively associated with retention (OR = 2.77, 95%CI: 1.06–7.46, p = 0.04) (model 5).

Table 3.

Adjusted odds ratios for retention stratified by CES-D score and gender from 2009 to 2014.

Stratified by depression
Variable	Model 3 (CES-D score ≥16, n = 378)		Model 4 (CES-D score <16, n = 29)
	AOR	95%CI	AOR	95%CI
Vulnerability
Marijuana use
Negative	1.00		1.00
Positive	0.69	0.39–1.22	–	–
Alcohol abuse
AUDIT score <8	1.00		1.00
AUDIT score ≥8	1.00	0.64–1.65	15.23	0.63–6.64 × 10²
Cocaine use
Negative	1.00		1.00
Positive	1.11	0.64–1.95	12.81	1.31–6.01 × 10²
Traditional
Education (years)	0.92	0.83–1.01	–	–
Age (years)	0.98	0.95–1.01	1.14	0.95–1.55
Race/Ethnicity
White (Non-Hispanic)	1.00		1.00
White (Hispanic)	1.67	0.64–4.39	–	–
Black (Non-Hispanic)	2.36**	1.02–5.52	0.02	0.00–1.00
Black (Hispanic)	5.99**	1.25–44.70	–	–
Other	6.82**	1.45–50.40	–	–
Gender
Female	1.00		1.00
Male	0.98	0.95–1.01	N/A	N/A
Stratified by gender
	Model 5 (Male, n=263)		Model 6 (Female, n=140)
	AOR	95%CI	AOR	95%CI
Vulnerability
Depression
CES-D score <15	1.00		1.00
CES-D score ≥16	2.77**	1.06–7.46	0.83	0.11–4.31
Marijuana use
Negative	1.00		1.00
Positive	0.72	0.36–1.46	0.66	0.27–1.67
Alcohol abuse
AUDIT score <8	1.00		1.00
AUDIT score ≥8	1.2	0.69–1.54x10³	1.22	0.53–2.89
Cocaine use
Negative	1.00		1.00
Positive	1.15	0.61–2.24	1.22	0.45–3.49
Traditional
Education (years)	0.90	0.79–1.00	0.93	0.76–1.13
Age (years)	0.97	0.93–1.00	1.02	0.97–1.08
Race/Ethnicity
White (Non-Hispanic)
White (Hispanic)	1.65	0.47–5.87	1.40	0.30–7.10
Black (Non-Hispanic)	0.51	0.63–6.30	3.00	0.81–11.82
Black (Hispanic)	5.47	0.89–47.94	3.97	0.35–99.48
Other	4.48	0.75–37.71	1.92 × 10⁷	0–N/A

Note: N/A indicates a variable has been removed due to abnormal distribution and – indicates removal by stepwise regression.

**p ≤ 0.05.

AOR: adjusted odds ratio; AUDIT: Alcohol Use Disorder Identification Test; CES-D: Center for Epidemiologic Studies-Depression scale.

In the syndemic analysis, we determined that 35.14% (n = 143) of the participants had two individual-level factors (Table 4). Approximately 31.2% (n = 127) of participants had three factors (Table 4). The ‘four factors’ category contained 12.03% (n = 49) of the participants in the sample (Table 4). The ‘five factors’ category included 1.97% (n = 8) of the sample (Table 4). The ‘zero factor category’ contained 0.98% (n = 4) of the sample (Table 4). Of the 407 participants, 19 were not included in this section of the analysis due to missing values for one of the factors.

Table 4.

Frequency distribution of factors among those retained and not retained in the MASH cohort in Miami-Dade county from 2009 to 2014.

Number of factors	Frequency (%) of category among retained	Frequency (%) of category among non-retained	Total frequency (%) overall	Chi squared, p value
Zero	1 (0.24)	3 (0.74)	4 (0.98)
One	41 (10.07)	16 (3.93)	57 (14.00)	1.96, 0.16
Two	92 (22.61)	51 (12.53)	143 (35.14)	1.17, 0.28
Three	83 (20.39)	44 (10.81)	127 (31.20)	1.27, 0.26
Four	33 (8.11)	16 (3.93)	49 (12.04)	1.34, 0.25
Five	5 (1.23)	3 (0.74)	8 (1.97)	0.37, 0.23
Missing	12 (2.95)	7 (1.72)	19 (4.67)

MASH: Miami Adult Studies on human immunodeficiency virus/acquired immune deficiency syndrome.

The bivariate association between the number of factors and retention yielded no significant results, nor did the remainder of the syndemic analyses (Tables 4 and 5).

Table 5.

Adjusted odds ratio for the syndemic association with retention (model 7, n = 388).

Variable	Exp (Est.)	Exp (CI)
Number of factors	0.99	0.79–1.24
Age	0.98	0.96–1.01
White Hispanic	1.7	0.68–4.39
Black	2.51**	1.09–5.87
Black Hispanic	4.48	1.08–23.57
Other race/ethnicity	7.32**	1.57–54.05
Male gender	0.89	0.55–1.37

**p ≤ 0.05.

Discussion

The only individual variable that was associated with retention was the traditional variable race/ethnicity. While multiple races/ethnicities were associated with improved odds of retention, persons in the Black Non-Hispanic racial/ethnic group had the lowest magnitude of improved odds of retention. The data support previous literature which suggests that Black Non-Hispanic race/ethnicity is associated with a reduced magnitude of improved odds of retention when compared to other racial/ethnic groups.¹⁰ The study also provides evidence that Black Hispanic persons and persons of the Other race/ethnicity group have a higher odds of retention in HIV care than their Non-Hispanic White counterparts. Current literature supports the finding that Hispanic persons have higher odds of being retained than do Non-Hispanic Blacks or Non-Hispanic Whites.⁵ A clinical implication of this finding would be to provide retention counseling to Non-Hispanic Black persons who have been recently diagnosed. There is scarce literature assessing the association of Black Hispanic race/ethnicity with retention, most literature typically assesses the relationship between general Hispanic ethnicity and retention. More research is necessary to determine the specific factors that contribute to this racial/ethnic group’s improved retention. Literature indicates that persons of ‘Multiple’ races/ethnicities have a greater likelihood of achieving retention than do Non-Hispanic Whites living with HIV, which may explain the ‘Other’ racial/ethnic category’s higher association with retention.³² There were no individual vulnerability factors associated with retention among this group.

The data revealed interaction among multiple vulnerability and traditional factors. The interaction between CES-D depression score and AUDIT score, two vulnerability factors, was negatively associated with retention. This negative interaction suggests that the negative impact of depression on retention differs by alcohol dependence. While the literature suggests that depressive symptoms and alcohol abuse reduce retention in HIV care separately, this is the first paper to assess the effect of a two-way interaction between the two variables on retention.^4,33 Alcohol has been determined to exacerbate the effect of depression among persons with HIV, and increased depressive symptoms have been shown to reduce retention.^4,34 In addition, those who abuse alcohol are at increased risk of major depression, and depression is associated with reduced retention.^4,35 Recent research related to the direction of causality between alcohol use and depressive symptoms provides inconsistent results.^35,36 A clinical implication of this finding would be to provide depressed persons living with HIV with counseling regarding alcohol use. Another clinical implication may be to screen persons with HIV who have alcohol use disorder for depression during HIV-related visits.

The interaction between age and male gender, two traditional factors, is also negatively associated with retention. The literature supports the notion that both male gender and increased age reduce retention in care.^37,38 This suggests that the negative impact of male gender on retention may be influenced by increased age (or equivalently the effect of age on retention varies by male gender). A clinical implication of this finding would possibly be to provide retention-related counseling to males living with HIV, specifically among older men, during their initial visit. A notable observation is that the vast majority of this sample is over the age of 35 years (n = 371, 91.60%). These data do not provide findings that are applicable to youth and emerging adults; however, the majority of studies suggest that persons under the age of 35 are the least retained age group in the United States.¹⁰ The interaction between depression and male gender yielded a positive association with retention. While both male gender and depression have been indicated to reduce retention in care among persons living with HIV, some studies indicate that there may be higher odds of retention among males when compared to females.^37–39 The literature does suggest that intravenous drug use among exclusively female samples may alter retention and is a common route of infection for women.^40,41 A clinical implication of this finding would be to provide women living with HIV who are depressed with additional counseling and follow-up regarding HIV-related visits during their initial visit.

To better understand the interactive effects among these (traditional and vulnerability) factors and retention, we conducted a stratified analysis which indicated that among those who are depressed race/ethnicity, a traditional factor, was associated with retention. Non-Hispanic Blacks had an increased odds of retention when compared to Non-Hispanic White persons; however, the magnitude of increased odds of retention was larger among other racial/ethnic groups (Black-Hispanic persons and Other persons). This trend is consistent with the current literature.^5,32 The reduced magnitude of this association among Non-Hispanic Black race/ethnicity with retention may be related to stigma within the Black community related to HIV care.⁴² A clinical implication of this finding may be to increase efforts to improve retention among Non-Hispanic Black persons living with HIV. The increased retention among the Black Hispanic and other racial/ethnic groups may be the result of specialized care in the Miami-Dade location. Previous literature suggests that bilingual personnel in an HIV care facility increases clinical visits among Hispanic persons with HIV.⁴³ Given the large Hispanic population in Miami, bilingualism is prevalent in most healthcare institutions. Among males, depression was significantly associated with retention.

We stratified the analysis by gender, a traditional factor, and noticed a positive association between depression (a vulnerability factor) and retention among males. A positive association between depression and retention in HIV care is presented in the literature, however, a majority of literature does indicate that depressive symptoms are generally negatively associated with retention.^4,33 The relationship between depression and retention has not been assessed among an exclusively male group prior to this article. Among females, there were no significant associations.

In the syndemic analysis, most persons who progressed to retention had two to three factors that could affect their health outcomes. Also, in the syndemic analysis, the number of factors was negatively associated with retention. Although this association was not significant, these trends could inform clinical practices and improve care for those with HIV.

A large number of Black-Hispanic persons in this cohort makes the data unique as most state-level studies related to retention in HIV care seldom have large numbers of persons in this racial/ethnic group. Including this racial/ethnic group in future studies may provide insight into racial and ethnic intersectionality as it relates to retention in HIV care.

Despite the strengths of this study, there were limitations. One of the major limitations was a relatively small sample size for these types of analyses, likely contributing to the wide confidence intervals related to some of the point estimates. The small sample size limited the statistical power of the syndemic analysis, which may have contributed to no significant findings. Other limitations of the analysis included the absence of other variables traditionally associated with retention (poverty, IDU, CD4 cell count, social support, and transportation), the underrepresentation of those less than 35 years of age, and the absence of sexual orientation identification in the data. Also, variables that are known to vary with time (AUDIT, CES-D, and substance use) were assessed only cross-sectionally, although persons were observed prospectively to determine retention status. Virologic suppression data, a key clinical outcome, were also not reported. Finally, enabling factors that have been suggested to be associated with reduced retention (i.e. income and employment status) were not assessed in this study.

Conclusion

Overall, we found multiple racial/ethnic groups, multiple levels of a traditional factor, are associated with improved retention and also multiple factors interact to alter retention in HIV care. This trend of ethnic disparities in retention persists among depressed persons living with HIV in this study. This is important because these findings suggest discrepancies among different race/ethnic groups related to retention, and clinical practices should be implemented to reduce these discrepancies. We also determined that interactions among vulnerability (depression and AUDIT score) factors are deleterious to the likelihood of retention. We also determined that traditional factors (male gender and age) can interact to reduce the likelihood of retention. This is important as barriers that may reduce retention in HIV are not unimodal and can inform clinical practices. The syndemic analysis revealed that most persons who were retained in care had two or three factors (traditional and vulnerability) that could reduce retention in care. This is also important as it can influence clinical practice to assess patients for more than one factor when attempting to identify factors that may negatively affect persons living with HIV.

Footnotes

Acknowledgements

We thank Dr Marianna Baum for being primarily responsible for data acquisition and for acquiring funding for the Miami Adult Studies on HIV/AIDS (MASH) cohort study from NIDA (5UO1DA040381). The data used for this article were derived from the MASH cohort study.

Authors’ contributions

KM wrote the manuscript and analyzed the data. MB oversaw the collection data and administration of the study. KF and GI proofread and assisted in the revision of the manuscript. TL revised and oversaw statistical analysis.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work was supported by the National Institute for Drug Abuse (5UO1DA040381). This work was also supported by Florida International University (Dissertation Year Fellowship) and by the Florida Education Fund (McKnight Fellowship).

References

Wawrzyniak

Rodríguez

Falcon

, et al. Association of individual and systemic barriers to optimal medical care in people living with HIV/AIDS in Miami-Dade County. J Acquir Immune Defic Syndr 2015; 69: S63–S72.

Kay

Batey

Mugavero

MJ.

The HIV treatment cascade and care continuum: Updates, goals, and recommendations for the future. AIDS Res Ther 2016; 13: 35.

Christopoulos

Das

Colfax

GN.

Linkage and retention in HIV care among men who have sex with men in the United States. Clin Infect Dis 2011; 52: S214–S222.

Zuniga

Yoo-Jeong

Dai

, et al. The role of depression in retention in care for persons living with HIV. AIDS Patient Care STDs 2016; 30: 34–38.

Althoff

Zelenev

Meyer

, et al. Correlates of retention in HIV care after release from jail: Results from a multi-site study. AIDS Behav 2013; 17: 156–170.

Kipp

Rebeiro

Shepherd

, et al. Daily marijuana use is associated with missed clinic appointments among HIV-infected persons engaged in HIV care. AIDS Behav 2017; 21: 1996–2004.

Cunningham

Sohler

Tobias

, et al. Health services utilization for people with HIV infection: Comparison of a population targeted for outreach with the U.S. population in care. Med Care 2006; 44: 1038–1047.

Mugavero

Davila

Nevin

, et al. From access to engagement: Measuring retention in outpatient HIV clinical care. AIDS Patient Care STDS 2010; 24: 607–613.

Lourenço

Colley

Nosyk

, et al. High levels of heterogeneity in the HIV cascade of care across different population subgroups in British Columbia, Canada. PLoS One 2014; 9: e115277.

10.

Horberg

Hurley

Klein

, et al. The HIV care cascade measured over time and by age, sex, and race in a large national integrated care system. AIDS Patient Care STDs 2015; 29: 582–590.

11.

Colasanti

Stahl

Farber

, et al. An exploratory study to assess individual and structural level barriers associated with poor retention and re-engagement in care among persons living with HIV/AIDS. J Acquir Immune Defic Syndr 2017; 74: S113–S120.

12.

Brennan

Kuhns

Johnson

, et al. Syndemic theory and HIV-related risk among young transgender women: The role of multiple, co-occurring health problems and social marginalization. Am J Public Health 2012; 102: 1751–1757.

13.

Dyer

Shoptaw

Guadamuz

, et al. Application of syndemic theory to black men who have sex with men in the Multicenter AIDS Cohort Study. J Urban Health 2012; 89: 697–708.

14.

Upton

Cook

A dictionary of statistics 3e. Oxford: Oxford University Press, 2014.

15.

Campa

Martinez

Sherman

, et al. Cocaine use and liver disease are associated with all-cause mortality in the Miami Adult Studies in HIV (MASH) Cohort. J Drug Abuse 2016; 2: 27.

16.

McCarty

Berg

Rottscheit

, et al. Validation of PhenX measures in the personalized medicine research project for use in gene/environment studies. BMC Med Genomics 2014; 7: 3.

17.

Raimondo

Camoni

Suligoi

, et al. HIV-positive individuals on antiretroviral therapy and with viral load suppressed in 12 infectious diseases clinics in Italy: Successes and disparities in the HIV continuum of care. AIDS Res Hum Retroviruses 2017; 33: 575–582.

18.

Cysique

Maruff

Bain

, et al. HIV and age do not substantially interact in HIV-associated neurocognitive impairment. J Neuropsychiatry Clin Neurosci 2011; 23: 83–89.

19.

Friedman

Stall

Plankey

, et al. Effects of syndemics on HIV viral load and medication adherence in the multicenter AIDS cohort study. AIDS 2015; 29: 1087.

20.

Clements-Nolle

Marx

Guzman

, et al. HIV prevalence, risk behaviors, health care use, and mental health status of transgender persons: Implications for public health intervention. Am J Public Health 2001; 91: 915–921.

21.

Kalichman

Weinhardt

Benotsch

, et al. Internet access and Internet use for health information among people living with HIV–AIDS. Patient Educ Couns 2002; 46: 109–116.

22.

Watson

Roberts

Saunders

MR.

Factors associated with anxiety and depression among African American and White Women. ISRN Psychiatry 2012; 2012.

23.

Devins

Orme

Costello

, et al. Measuring depressive symptoms in illness populations: Psychometric properties of the Center for Epidemiologic Studies Depression (CES-D) scale. Psychology & Health 1988; 2: 139–156.

24.

Van Dam

Earleywine

Validation of the Center for Epidemiologic Studies Depression Scale—Revised (CESD-R): Pragmatic depression assessment in the general population. Psychiatry Res 2010; 186: 128–132.

25.

United States Food & Drug Administration. 510 (k) premarket Notification. Available at: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K993796 (2019, accessed 28 November 2019).

26.

Saunders

Aasland

Babor

, et al. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption-II. Addiction 1993; 88: 791–804.

27.

Swann

Economic strengthening for retention in HIV care and adherence to antiretroviral therapy: A review of the evidence. AIDS Care 2018; 30: 99–125.

28.

van den Putte

Yzer

Brunsting

Social influences on smoking cessation: A comparison of the effect of six social influence variables. Prev Med 2005; 41: 186–193.

29.

Parboteeah

Hoegl

Cullen

JB.

Managers' gender role attitudes: A country institutional profile approach. J Int Bus Stud 2008; 39: 795–813.

30.

Mizuno

Purcell

Knowlton

, et al. Syndemic vulnerability, sexual and injection risk behaviors, and HIV continuum of care outcomes in HIV-positive injection drug users. AIDS Behav 2015; 19: 684–693.

31.

Garson

Multiple regression. 2014 ed. Statistical Associates Blue Book Series 6 [Kindle Edition]. 2014.

32.

Doshi

Milberg

Isenberg

, et al. High rates of retention and viral suppression in the US HIV Safety Net System: HIV Care Continuum in the Ryan White HIV/AIDS Program, 2011. Clin Infect Dis 2015; 60: 117–125.

33.

Monroe

Lau

Mugavero

, et al. Heavy alcohol use is associated with worse retention in HIV care. J Acquir Immune Defic Syndr 2016; 73: 419–425.

34.

Sullivan

Saitz

Cheng

, et al. The impact of alcohol use on depressive symptoms in human immunodeficiency virus-infected patients. Addiction 2008; 103: 1461–1467.

35.

Fergusson

Boden

Horwood

LJ.

Tests of causal links between alcohol abuse or dependence and major depression. Arch Gen Psychiatry 2009; 66: 260–266.

36.

Marmorstein

NR.

Longitudinal associations between alcohol problems and depressive symptoms: Early adolescence through early adulthood. Alcohol Clin Exp Res 2009; 33: 49–59.

37.

Fleishman

Yehia

Moore

, et al. Establishment, retention, and loss to follow-up in outpatient HIV care. J Acquir Immune Defic Syndr 2012; 60: 249–259.

38.

Torian

Wiewel

EW.

Continuity of HIV-related medical care, New York City, 2005–2009: Do patients who initiate care stay in care?

AIDS Patient Care STDs 2011; 25: 79–88.

39.

Kinsler

Sheng

, et al. Using laboratory surveillance data to estimate engagement in care among persons living with HIV in Los Angeles County, 2009. AIDS Patient Care STDS 2012; 26: 471–478.

40.

Stone

VE.

HIV/AIDS in women and racial/ethnic minorities in the US. Curr Infect Dis Rep 2012; 14: 53–60.

41.

Blank

Fletcher

Verdecias

, et al. Factors associated with retention and viral suppression among a cohort of HIV women of color. AIDS Patient Care STDS 2015; 29: S27–S35.

42.

Giordano

Hartman

Gifford

, et al. Predictors of retention in HIV care among a national cohort of US veterans. HIV Clin Trials 2009; 10: 299–305.

43.

Enriquez

Farnan

Cheng

, et al. Impact of a bilingual/bicultural care team on HIV-related health outcomes. J Assoc Nurses AIDS Care 2008; 19: 295–301.