Sociodemographic factors affecting viral load suppression among people living with HIV in South Carolina

Abstract

Adherence to antiretroviral therapy (ART) enables people living with HIV (PLWH) to reach and maintain viral suppression. As viral suppression significantly reduces risk for secondary transmission, this study aimed to examine sociodemographic factors associated with viral suppression among PLWH in South Carolina (SC). We analyzed cross-sectional data collected from 342 PLWH receiving HIV care from a large clinic in SC and provided complete information on most recent viral load, ART adherence, and sociodemographic factors. Bivariate analysis examined associations between key variables, and logistic regression was used to calculate the odds of viral suppression among select sociodemographic groups and adherence levels. Results indicated that approximately 82% of participants reported achieving viral suppression. PLWH who were older, male, and employed full-time had higher odds of being virally suppressed compared to those who were younger, female, and unemployed. PLWH with medium (adjusted Odds Ratio [aOR]: 3.79; 95% CI: 1.15–12.48) and high (aOR: 3.51; 95% CI: 1.21–10.24) levels of adherence were more likely to report viral suppression than those with low adherence. Targeted interventions are warranted for groups at-risk of low ART adherence, and healthcare providers should also be aware of contextual factors that serve as barriers to adherence for PLWH.

Introduction

With the advent of highly active antiretroviral therapy (ART), human immunodeficiency virus (HIV) has been largely transformed from an acute to a chronic health condition. Successful management of HIV can be achieved through an effective HIV care continuum – including initial testing, rapid linkage-to-care, engagement in-care, and ART adherence (Deeks, Lewin, & Havlir, 2013). ART adherence helps in achieving viral suppression, which, in turn, improves quality of life and reduces opportunistic infections, comorbidities, and HIV-related mortality (Cohen et al., 2011; Gill et al., 2002). The HIV care continuum is critical in reducing HIV incidence, as evidence suggests that people living with HIV (PLWH) with an undetectable viral load (VL) have effectively no risk of transmitting HIV to uninfected persons (Gardner, McLees, Steiner, Del Rio, & Burman, 2011). As a result, “Treatment as Prevention” (TasP) has become the dominant HIV prevention strategy in the United States (US) and across the globe (Dieffenbach, 2012), and has been found to be effective in preventing HIV transmission (Bavinton et al., 2018; Cohen, McCauley, & Gamble, 2012). This evidence spurred the new movement titled “Undetectable = Untransmittable (U = U)” by Prevention Access Campaign in 2016 (PAC, 2016) and endorsed by UNAIDS in 2018 (UNAIDS, 2018).

In 2016, the Centers for Disease Control and Prevention (CDC) estimated that 86% of all PLWH in the US were diagnosed, and of these, 64% were linked to HIV care, 49% were retained in-care, and 53% achieved viral suppression (CDC, 2019c). This low level of viral suppression is highly concerning as individuals with a detectable VL may transmit the virus to others (CDC, 2019d). A recent study showed that 91.5% of new HIV infections were transmitted by PLWH who were either undiagnosed or not engaged in medical care; while, less than 6% of new cases were transmitted from those receiving care and prescribed with ART (Skarbinski et al., 2015).

Compared to other US regions, in 2014, nearly half (44%) of all new HIV diagnoses were in the South, which contains only around one-third (37%) of the total population (CDC, 2016). The South also accounts for 52% of new AIDS cases in the US (CDC, 2018b). Furthermore, half (51%) of the PLWH who are unaware of their serostatus reside in the South (Reif, Safley, McAllaster, Wilson, & Whetten, 2017).

South Carolina (SC), a predominantly rural southern state, has a relatively high incidence of HIV and AIDS cases. African Americans constitute 70% (47% men, 23% women) of the total PLWH in SC (SCDHEC, 2017). In a study conducted in SC, almost a quarter of newly-diagnosed PLWH were delayed in being linked to care (i.e., 30–365 days to first CD4 or VL test after diagnosis) or were not in care at all (Gill, Babatunde, & Weissman, 2017). Cumulatively only 68% of PLWH in SC received any care during 2016, while only 53% were consistently engaged in care, and 57% were virally suppressed at their most recent test (SCDHEC, 2017). These numbers highlight the importance of tracking the progress of Southern states like SC in improving the HIV care continuum, including increasing viral suppression rates. The current study examines associations between key sociodemographic factors and adherence to ART and viral suppression among a diverse sample of PLWH in SC.

Methods

Data source and study design

In 2018, we partnered with a large immunology center (IC) in Columbia, SC to complete a cross-sectional survey with a subset of their current PLWH. The University of South Carolina Institutional Review Board approved the study. Data were collected between May and September of 2018, with all surveys completed on site at the Ryan White Program (RWP)-funded clinic. All PLWH attending the IC for follow-up HIV care visits were informed about the study. Those interested in participating were directed to a research team member, who then further explained the study and obtained informed consent from participants. Study inclusion criteria included: (1) resident of SC, (2) ≥18 years of age, (3) PLWH, and (4) willing to participate in a 35–40 min survey. Each participant received a $20 gift card in appreciation of their time and effort. The anonymous survey was administered in designated clinic areas and, when requested, in a private room. More than 80% of invited individuals agreed to participate in the survey. A total of 402 PLWH initiated the survey. A small number of individuals (n = 5) did not complete the survey due to time constraints (e.g., had to get labs, had another appointment), resulting in a total of 397 completing the survey. Due to missing data on key study variables, the final analysis for the current study included 342 PLWH.

Measures

Viral suppression

Viral suppression was measured through participant self-report. Specifically, participants were asked to report the results of their most recent VL test result. Although different organizations set undetectable VL at different levels (e.g., CDC set it at <200 copies/ml (CDC, 2019b), while San Francisco AIDS Foundation sets it at <50 copies/ml (Land, 2018), it really depends on the available test’s lower limit of detecting HIV RNA copies/ml of blood. Therefore, we used 100 copies/ml as the undetectable VL cut-off level in this study. Possible responses included undetectable, detectable but less than 100 viral copies/ml, detectable but 100 to 500 viral copies/ml, 500 to 5000 viral copies/ml, greater than 5000 viral copies/ml, and “I don’t know”. For the current study, a VL was considered undetectable if <100 copies/ml – a stricter threshold than the current threshold that is used by the CDC (i.e., (i.e., <200 copies/ml) (CDC, 2019b). Thus, if the participant reported an undetectable VL or a VL of<100 viral copies/ml, we defined the participant as virally suppressed.

ART adherence

Self-reported ART adherence was measured by four items adopted from the Adult AIDS Clinical Trials Group adherence instrument (Chesney et al., 2000). The items were modified to consider ART adherence over last month. Three items asked PLWH to report whether they had missed their medication “yesterday”, the “day before yesterday”, and “over the past weekend (Saturday/Sunday)”. Another item asked PLWH to estimate how frequently they missed a dose of ART over a typical month, with options including never miss, miss less than one day, miss one to three days, miss three to five days, and miss at least six days. Cronbach’s alpha of the ART adherence scale was 0.63, indicating acceptable internal reliability. A single indicator of ART adherence was generated by a principal component analysis which assessed the factor structure of the four adherence items. The eigenvalue was 2.46 for the first component and 0.62 for the second component, which strongly suggested a single factor structure. Accordingly, based on the ART adherence factor score, we categorized patients into three tertiles containing a third of the population in each. The “low adherence” tertile had a cutoff value of 0.07, the “medium adherence” tertile had a cutoff value of 0.51, and the “high adherence” tertile had a cutoff value of 0.95.

Sociodemographic variables

Participants reported on a number of sociodemographic variables that have been previously associated with viral suppression (Mugavero, Amico, Horn, & Thompson, 2013). Participants were asked to report their age (18–29, 30–49, ≥50 years), gender (male, female), sexual orientation (“lesbian, gay or bisexual”, and heterosexual), race (African American, White, other), marital status (married/living together, never married, other), educational attainment (grades 1–11, grade 12 or GED, some college, Bachelor’s degree or higher), employment status (unable to work/unemployed, employed part-time, employed full-time, other), annual household income (<$10,000, $10,000–$24,999, $25,000–$49,999, ≥$50,000), and whether they had been previously incarcerated (no, yes).

Statistical analysis

First, we generated descriptive statistics for sociodemographic variables (Table 1). Next, we used bivariate analysis (chi-square test for categorical variables) to examine associations between our key variables of interest (Table 2). Finally, to examine the association between sociodemographic factors, ART adherence, and viral suppression, logistic regression was used to calculate the adjusted odds ratios (aOR) and 95% confidence intervals (95%CI) of reporting viral suppression (Table 3). In the adjusted model we included all sociodemographic variables and ART adherence because these variables were found to be significantly correlated with viral suppression in prior research (Blank et al., 2015; Chakraborty et al., 2015; Mugavero et al., 2013). We chose not to exclude any variables found to be nonsignificant in bivariate analysis because this type of variable selection often rejects potentially important variables (Sun, Shook, & Kay, 1996). Due to casewise deletion (deleting observations with covariates with missing values) the resultant sample size was 193 for the multivariable analysis. All analyses were performed with Stata 14.2 (StataCorp, 2015).

Table 1.

Social demographic and HIV-related characteristics of the PLWH, N = 342.

Characteristics	n (%)
Viral suppression
No	63 (18.4)
Yes	279 (81.6)
Age
18–29 Years	71 (18.4)
30–49 Years	163 (42.1)
≥50 Years	153 (39.5)
Sex
Female	135 (34.3)
Male	258 (65.7)
Sexual orientation
Lesbian, gay or bisexual	203 (53.3)
Heterosexual	178 (46.7)
Race
Other	21 (5.2)
White	67 (16.8)
Black	312 (78.0)
Marital status
Married/Living together	69 (21.0)
Never married	193 (58.6)
Other	67 (20.4)
Education
Grades 1–11	51 (15.4)
Grade 12 or GED	104 (31.3)
Some college	114 (34.3)
Bachelor or higher	63 (19.0)
Employment
Full-time	133 (40.2)
Part-time/Other	81 (24.5)
Unable to work/unemployed	117 (35.3)
Annual household income
<$10,000	135 (36.0)
$10,000-$24,999	106 (28.3)
$25,000-$49,999	90 (24.0)
≥$50,000	44 (11.7)
Incarceration history
No	235 (68.7)
Yes	107 (31.3)
Adherence to ART
Low	105 (30.8)
Medium	77 (22.6)
High	159 (46.6)
Total	342 (100.0)†

^†Due to missing data, some variables do not have 342 observations.

Table 2.

Bivariate analysis of viral suppression by demographic and HIV-related characteristics of PLWH, n = 342.

Characteristics	Viral suppression		p-Value
	No n (%)	Yesn (%)
Age			0.001
18–29 Years	20 (32.8)	41 (67.2)
30–49 Years	27 (19.2)	114 (80.8)
≥50 Years	14 (10.7)	117 (89.3)
Sex			0.147
Female	24 (22.6)	82 (77.4)
Male	37 (16.1)	193 (83.9)
Sexual orientation			0.485
Lesbian, gay and bisexual	31 (16.8)	154 (83.2)
Heterosexual	29 (19.7)	118 (80.3)
Race			0.290
Other	3 (15.8)	16 (84.2)
White	7 (11.7)	53 (88.3)
Black	53 (20.2)	209 (79.8)
Marital status			0.001
Married/Living together	9 (15.5)	49 (84.5)
Never married	44 (27.1)	118 (72.8)
Other	3 (5.4)	53 (94.6)
Education			0.005
Grades 1–11	8 (23.5)	26 (76.5)
Grade 12 or GED	24 (30.4)	55 (69.6)
Some college	20 (19.2)	84 (80.8)
Bachelor’s or higher	4 (6.5)	58 (93.5)
Employment			0.070
Full-time	17 (14.7)	85.3 (99)
Part-time/Other	13 (18.8)	81.2 (56)
Unable to work/unemployed	25 (27.5)	66 (72.5)
Annual household income			0.036
<$10,000	25 (23.2)	83 (76.8)
$10,000-$24,999	20 (21.3)	74 (78.7)
$25,000-$49,999	12 (14.5)	71 (85.5)
≥$50,000	2 (4.7)	41 (93.3)
Incarceration history			0.326
No	34 (16.4)	174 (83.6)
Yes	18 (21.2)	67 (78.8)
Adherence to ART			0.116
Low	22 (24.2)	69 (75.8)
Medium	13 (18.3)	58 (81.7)
High	20 (13.6)	127 (86.4)

Table 3.

Multivariable model for key sociodemographic factors associated with viral suppression among PLWH in South Carolina (n = 193).

Characteristics	Unadjusted odds ratio	95% CI	p-Value	Adjusted odds ratio	95% CI	p-Value
Age
18–29 Years	Ref.			Ref.
30–49 Years	2.06*	1.04–4.06	0.037	3.29*	1.01–10.34	0.048
≥50 Years	4.08***	1.89–8.81	0.000	6.22*	1.29–27.39	0.022
Sex
Female	Ref.			Ref.
Male	1.53	0.86–2.71	0.149	4.35*	1.04–18.22	0.043
Sexual orientation
Lesbian, gay, or bisexual	Ref.			Ref.
Heterosexual	0.82	0.47–1.43	0.485	0.68	0.18–2.55	0.566
Race
Other	Ref.			Ref.
White	1.42	0.33–6.13	0.639	1.55	0.14–17.03	0.720
Black	0.74	0.21–2.63	0.641	0.77	0.10–5.71	0.799
Marital status
Other	Ref.			Ref.
Married/Living together	0.31	0.79–1.20	0.091	0.11*	0.01–0.74	0.024
Never married	0.15**	0.05–0.51	0.002	0.09*	0.02–0.56	0.010
Education
Bachelor or higher	Ref.			Ref.
Grades 1–11	0.22*	0.06–0.81	0.023	0.25	0.04–1.47	0.125
Grade 12 or GED	0.16**	0.05–0.48	0.001	0.26	0.06–1.06	0.060
Some college	0.29*	0.09–0.89	0.031	0.69	0.18–2.68	0.592
Employment
Unable to work/ unemployed	Ref.			Ref.
Full-time	2.21*	1.11–4.40	0.025	4.03*	1.27–12.83	0.018
Part-time/Other	1.63	0.76–3.49	0.206	2.75	0.78–9.73	0.117
Annual household income
<$10,000	Ref.			Ref.
$10,000-$24,999	1.11	0.57–2.17	0.750	0.21*	0.06–0.73	0.014
$25,000-$49,999	1.78	0.84–3.80	0.135	0.11**	0.03–0.52	0.005
≥$50,000	6.17*	1.39–27.35	0.016	0.78	0.06–10.61	0.851
Incarceration history
No	Ref.			Ref.
Yes	0.73	0.38–1.38	0.327	0.39	0.13–1.19	0.096
Adherence to ART
Low	Ref.			Ref.
Medium	1.42	0.66–3.07	0.369	3.79*	1.15–12.48	0.028
High	2.02*	1.03–3.97	0.040	3.51*	1.21–10.24	0.021

^***p-Value < 0.0001;

^**p-Value < 0.001;

^*p-Value < 0.01.

Results

In Table 1 we present the demographic characteristics of the study participants. Approximately four in ten participants (42.1%) were aged 30–49, and more than one-third (39.5%) of participants were over 50 years old. The majority of the sample (65.7%) was male, with 34.3% identifying as female. More than half (53.3%) reported being lesbian, gay, or bisexual, and more than three-quarters (78.0%) were African American. More than half (58.6%) of the sample were never married, one-third (34.4%) had some college education, two-fifths (40.2%) were employed full-time, and more than one-third (36.0%) reported an annual household income of less than $10,000. Almost one-third (31.3%) reported that they had previously been incarcerated. Nearly half (46.6%) of participants reported a high level of ART adherence and more than four-fifths (81.6%) reported an undetectable VL.

The results of the bivariate analyses are shown in Table 2. There were statistically significant differences in viral suppression by age, marital status, education, and annual household income.

Table 3 shows results of the multivariable analysis that was conducted to assess associations between sociodemographic factors, ART adherence, and viral suppression. In the adjusted model, PLWH aged 30–49 years (adjusted odds ratio [aOR] = 3.29, 95% CI = 1.01–10.34) and ≥50 years (aOR = 6.22, 95% CI = 1.29–27.39) had higher odds of being virally suppressed PLWH aged 18–29 years. Men (aOR = 4.35, 95% CI = 1.04–18.22) than females had higher chance of being virally suppressed. PLWH wo were employed full-time (aOR = 4.03, 95%CI = 1.27–12.83) than unemployed/unable to work had higher odds of being virally suppressed. PLWH with medium levels of adherence (aOR = 3.79, 95%CI = 1.15–12.48) and high levels of adherence (aOR = 3.55, 95%CI = 1.21–10.24) than those with low levels. PLWH who were married/living together (aOR = 0.11, 95% CI = 0.01–0.74) and never married (aOR = 0.09, 95% CI = 0.02–0.56) had lower odds of viral suppression than those with other relationship status, PLWH with a household annual income of $10,000–$24,499 (aOR = 0.21, 95%CI = 0.06–0.73) and $25,000–$49,999 (aOR = 0.11, 95%CI = 0.03–0.52) had lower odds of viral suppression than those with an annual household income of <$10,000.

Discussion

Participants in this study reported high overall rates of viral suppression, which has important implications in for reducing the HIV epidemic in the Southern US. These high rates show that PLWH who are linked-to-care can achieve viral suppression and strengthen the HIV care continuum. Participants in this study reported a rate of viral suppression of 81.6%, which is higher than the current statewide estimate (53%) (SCDHEC, 2017) and is also higher than current national estimates of viral suppression (49%) (CDC, 2018a). These differences could be explained by having recruited participants from a comprehensive HIV care center, as it is likely that these PLWH had better quality of HIV care. However, there is still a need for improvement in testing, linkage, retention in care, and viral suppression to attain the targets set by the National HIV/AIDS Strategy (The White House, 2016). Understanding differences in viral suppression rates among various sub-populations in the Southern US may help to determine potential target populations for future tailored interventions.

Our findings showed that older PLWH were more likely to achieve viral suppression compared to younger PLWH. Similar results were found in other studies (Chakraborty et al., 2015; Crepaz, Dong, Wang, Hernandez, & Hall, 2018). One study conducted in SC found that youth (aged 13–19) were slower to achieve reductions in VL (Chakraborty et al., 2015). Similarly, a nationwide study found that African American PLWH (aged 13–24 years) had the lowest prevalence of sustained viral suppression (Crepaz et al., 2018). Young people (13–24 years) constitute about a quarter of the new infections and an estimated 60% of these youth PLWH are unaware of their serostatus (CDC, 2013).

Women living with HIV in the current study had lower odds of achieving viral suppression compared to men. This is consistent with multiple studies from the US (Chakraborty et al., 2015; Nance et al., 2018; Xia, Robbins, Lazar, Torian, & Braunstein, 2017). Other studies have also documented that women living with HIV are vulnerable to experiencing delayed care, greater stigma and other barriers, and poorer health outcomes (Aziz & Smith, 2011). These barriers may reduce medication adherence leading to lower rates of viral suppression. Moreover, existing literature on adherence in other higher-income countries suggests that women are less likely to adhere to ART than men and recommends specialized care for women to increase their adherence (Puskas et al., 2011). Future research should explore the needs of women across the HIV care continuum to better understand unique barriers at each stage.

In our study, PLWH with higher educational attainment were more likely to be virally suppressed, which may be due to have higher health literacy (Kalichman, Ramachandran, & Catz, 1999), or higher levels of knowledge and understanding about HIV (Kalichman et al., 2000; Kalichman & Rompa, 2000). As a result, they may be more likely to access care, adhere to ART, and thus attain viral suppression.

PLWH who were employed full-time were more likely to be virally suppressed. Similar socioeconomic disparities in viral suppression have been reported in the United Kingdom (Burch et al., 2016) and US (Hussen et al., 2018; Xia et al., 2017). Another study noted that unemployment was associated with twice the adjusted risk of virologic failure (Saracino et al., 2016). Socioeconomic status could impact virologic outcomes by delays in receiving a timely HIV diagnosis (Girardi, Sabin, & Monforte, 2007), low CD4 count at ART initiation (Le Moing et al., 2002), differences in experiences or quality of health care, and pharmacokinetics (e.g., nutrition deficiencies) (Weiser et al., 2009). Conversely, our results show that middle-income PLWH had lower odds of achieving viral suppression than low-income PLWH. More research is warranted to understand this surprising finding, but potential hypothesis included that the lower-income PLWH may be linked to more federal and state programs or receive more services funded through the Ryan White Program. In one recent study, uninsured patients supported through the program more likely to be virally suppressed (Bradley et al., 2015) than uninsured patients without assistance. These findings highlight the need for policies that ensure that all PLWH have access to comprehensive HIV care, ART, and social services (e.g., employment and income assistance).

Finally, in our study, high levels of ART adherence were positively associated with higher odds of viral suppression. This finding is consistent with current scientific evidence and clinical guidelines supporting immediate and lifelong adherence to ART (Howard et al., 2002; Sithole et al., 2018; Ti et al., 2014).

Limitations

The study had several limitations. First, all patients were currently receiving care at comprehensive IC in SC. This sample included in the analysis (N = 342) represents ~2.6% of the total population of PLWH in SC who received HIV care in 2017 (N = 13,348) (SCDHEC, 2017). However, the sample is not representative of all PLWH in SC and may underrepresent rural patients. Moreover, rather than using electronic medical records, we relied on self-reported levels of viral suppression. Recall bias may impact results, although, previous studies have suggested that PLWH who are in care are able to provide accurate information with regard to their VL (Carter et al., 2017; Grabowski et al., 2018; Sewell et al., 2017). Future studies should seek to incorporate more objective measures (i.e., biomarkers) of viral suppression and adherence, and use other methods (e.g., adherence monitoring devices) to confirm self-reports. In addition, while self-reported ART adherence has shown a strong correlations with virologic response (DeMasi et al., 2001; Shi et al., 2010) and adherence monitoring devices (Arnsten et al., 2001), future studies may use the latter along with patient self-report. In a post-hoc analysis for missing data, we found that there were some significant differences in participants who were excluded from the analysis due to missing data, in terms of gender, sexual orientation, education, annual household income, and incarceration history. Thus, the associations between these variables and viral suppression might be biased in the current analysis. However, as discussed before, PLWH with these characteristics (i.e., females, low education, low income) had lower odds of viral suppression. Therefore, it is likely that the associations in the current study are underestimates of the “true association.”

Similarly, neither sexual orientation nor incarceration history was associated with the outcome; however, the missing data remains a limitation. In addition, the wide confidence intervals for some variables in the multivariable model may indicate sparse data bias (i.e., some combinations of variables may lack adequate case numbers resulting in a lack of power to detect statistically significant differences) (Greenland, Mansournia, & Altman, 2016).

Conclusions

Despite these limitations, this study identifies important factors associated with viral suppression of PLWH in SC and has important implications for HIV treatment and care. These findings highlight the need for the development of intervention strategies that target PLWH who are most at-risk for failing to reach and maintain viral suppression in the Southern US. Particularly, targeted approaches in SC may be needed to helpyoung and unemployed PLWH as well as women living with HIV to achieve viral suppression. As adherence is critical to viral suppression, identifying facilitators and barriers to adherence that are encountered by PLWH in the Southern US is critical to improving individual outcomes and reducing the incidence of secondary transmission across the region. Such findings can also be helpful in developing adherence interventions and advocating for policy changes that support effective patient-provider partnerships (CDC, 2019a).

Appendix

Table A1.

Table: post-hoc analysis for missing data.

Characteristics	Included in analysis		p-Value
	N = 60 No n (%)	N = 342 Yes n (%)
Age			0.974
18–29 Years	10 (14.1)	61 (85.9)
30–49 Years	22 (13.5)	141 (86.5)
≥50 Years	22 (14.4)	131 (85.6)
Sex			0.004
Female	29 (21.5)	106 (78.5)
Male	28 (10.9)	230 (89.1)
Sexual orientation			0.013
Lesbian, gay and bisexual	31 (16.8)	154 (83.2)
Heterosexual	29 (19.7)	118 (80.3)
Race			0.397
Other	2 (9.5)	19 (90.5)
White	7 (10.5)	60 (89.5)
Black	50 (16.0)	262 (84.0)
Ethnicity			0.399
Hispanic	3 (11.7)	13 (88.3)
Non-hispanic	37 (18.7)	279 (81.3)
Marital status			0.997
Married/Living together	11 (15.9)	58 (84.1)
Never married	31 (16.1)	162 (83.9)
Other	11 (16.4)	56 (83.6)
Education			0.000
Grades 1–11	17 (33.3)	34 (66.7)
Grade 12 or GED	25 (24.0)	79 (76.0)
Some college	10 (8.8)	104 (91.2)
Bachelor’s or higher	1 (1.6)	62 (98.4)
Employment			0.119
Full-time	17 (14.7)	116 (87.2)
Part-time/Other	12 (18.8)	69 (85.2)
Unable to work/unemployed	26 (27.5)	91 (77.8)
Annual household income			0.004
<$10,000	27 (20.0)	108 (80.0)
$10,000-$24,999	12 (11.3)	94 (88.7)
$25,000-$49,999	7 (7.8)	83 (92.2)
≥$50,000	1 (2.3)	43 (97.7)
Incarceration history			0.026
No	27 (11.5)	208 (88.5)
Yes	22 (20.6)	85 (79.4)
Adherence to ART			0.248
Low	14 (13.3)	91 (67.2)
Medium	6 (7.8)	71 (92.2)
High	12 (7.6)	147 (92.4)