Abstract
Changes in an individual’s contextual factors following HIV diagnosis may influence long-term outcomes. We evaluated how changes to contextual factors between HIV diagnosis and 9-month follow-up predict 5-year mortality among HIV-infected individuals in Durban, South Africa enrolled in the Sizanani Trial (NCT01188941). We used random survival forests to identify 9-month variables and changes from baseline predictive of time to mortality. We incorporated these into a Cox proportional hazards model including age, sex, and starting ART by 9 months a priori, 9-month social support and competing needs, and changes in mental health between baseline and 9 months. Among 1,154 participants with South African ID numbers, 900 (78%) had baseline and 9-month data available of whom 109 (12%) died after 9-month follow up. Those who reported less social support at 9 months had a 16% higher risk of mortality. Participants who went without basic needs or healthcare at 9 months had a 2.6 times higher hazard of death compared to participants who did not. Low social support and competing needs at 9-month follow-up substantially increase long-term mortality risk. Reassessing contextual factors during follow-up and targeting interventions to increase social support and affordability of care may reduce long-term mortality for HIV-infected individuals in South Africa.
Keywords: HIV Infection, Predictors of Mortality, Mortality, South Africa
Introduction
South Africa has the greatest burden of HIV in the world. Despite major scale-up of ART in the region (UNAIDS, 2017) and current treatment affording those on ART an opportunity to achieve similar life expectancies to those who are HIV-uninfected (Mills et al., 2011), long-term mortality among people living with HIV (PLWH) remains high (Cornell et al., 2017). We previously found that more self-perceived barriers to care at baseline increased 1-year and 5-year mortality for PLWH in South Africa (Bassett et al., 2017, 2019). Understanding how changing contextual factors over time might influence outcomes for PLWH in South Africa has not been investigated, but is critical to determining if interventions aimed at eliminating barriers post-HIV diagnosis might decrease long-term mortality for PLWH.
We and others have assessed measures of contextual factors such as mental health, social support, and ability to meet basic and healthcare needs at single time points in HIV care in Sub-Saharan Africa (Bassett et al., 2017; Layer et al., 2014; Mukoswa et al., 2017; Tomori et al., 2014). These studies have determined that poor mental health, low levels of social support, and presence of competing needs at HIV diagnosis affect mortality risk. However, these factors can change over time, particularly following a new HIV diagnosis, when stigma may increase barriers to access and decrease psychosocial support (Bogart et al., 2013; Katz et al., 2015). There is a dearth of research regarding how changes to emotional health, perceived support, and financial constraints over time affects long-term mortality risk. Future interventions might include monitoring changes in these contextual factors over time, rather than only at a single timepoint, to predict mortality.
We used data from the Sizanani Trial (NCT01188941) and the South African National Population Register (Cornell et al., 2014; Johnson et al., 2015) to assess how changes in contextual factors reported at both baseline and 9 months for PLWH could predict mortality at 5 years. We hypothesized that social support, mental health, and ability to meet financial needs at 9 months, as well as changes from baseline, would be associated with increasing 5-year mortality rates.
Methods
Study Setting/Design
The Sizanani Trial (NCT01188941) was a randomized controlled trial that assessed the efficacy of health system navigators and SMS-based reminders for improving linkage to care for HIV and HIV/TB coinfection. We enrolled adults (≥18y) before HIV testing at 4 outpatient sites in Durban, South Africa from August 2010-January 2013. The trial details are presented elsewhere (Bassett et al., 2013, 2016, 2017). Linkage to HIV care, TB treatment completion, and 1-year mortality were not substantially different between study arms. Thus, we performed a secondary analysis of PLWH from both study arms combined in this analysis.
Participants
Participants were eligible for enrollment if they were English or Zulu-speaking adults (≥18y) presenting to outpatient sites for HIV testing with unknown HIV status. Participants provided informed consent and completed a baseline questionnaire prior to HIV testing, allowing for unbiased assessment of emotional health, social support, and competing needs before knowledge of HIV status.
McCord Hospital Medical Research Ethics Committee, St. Mary’s Hospital Research Ethics Committee, University of Kwazulu-Natal Biomedical Research Committee, and Partners Institutional Review Board (Protocol 2011-P-01195, Boston, MA) all approved the study.
Data elements
We collected demographic information at baseline and data elements that could change over time at baseline and at 9-month follow-up using identical survey questions. These included mental health, social support, employment status, and self-perceived areas of competing needs.
Self-perceived barriers to care
We assessed self-perceived barriers to healthcare in the 6 months prior with 12 questions (Craw et al., 2008). We grouped barriers into 5 domains: 1) concerns about service delivery (waiting time to see a provider, treatment by clinic staff), 2) financial concerns (ability to afford medication or transportation), 3) perception of personal health (not being sick enough/being too sick), 4) logistical concerns (unable to get out of work, responsibilities to care for others), 5) structural concerns (impaired clinic access [clinic hours, transportation], lack of knowledge about where to find care, language barrier). We created a total number of barriers variable by adding up all barriers in all 5 categories for each participant. We created a total number of domains variable by adding up the total number of domains under which a participant reported a barrier.
Mental health
We adapted the 5-item Mental Health Inventory and calculated a mental health composite (MHC) score (Hays et al., 1993) to assess mental health status. This scale was converted to a scaled score from 0 to 100, where a higher number indicated better mental health. MHC score ≤ 52 qualified as a positive depression screen.
Social support
We used a 13-question social support questionnaire consolidated into 4 subscales (emotional/informational, tangible, positive interaction, and affectionate) to calculate the Social Support Index (SSI) from the Medical Outcomes Study (Sherbourne & Stewart, 1991). This was converted to a 0 to 100 scale; a higher number indicated better social support. SSI score below sample median indicated lack of social support.
Competing needs
To determine if patients had competing needs we asked if, in the past 6 months, they had ever gone without healthcare because they needed money for basic needs (food, clothing or housing), or if they had gone without basic needs because they needed money for healthcare (Cunningham et al., 1999; RAND Corporation, n.d.).
9-month follow-up
We assessed if participants had started ART by 9 months. Participants were determined to have started ART if they were ART-eligible by contemporaneous guidelines and confirmed to have been on ART for at least 3 months based on pharmacy/medical record review or self-report. We also asked participants if they had received medication to treat or prevent HIV-related opportunistic infections.
Outcome ascertainment
Our primary outcome was time to death after study enrollment, ascertained by checking vital status in the National Population Register, which incorporates >90% of deaths in South Africa (Cornell et al., 2014). We used South African ID numbers (SAIDs) obtained at enrollment to match participants to the Register (median follow-up time 5.8y (IQR 5.2–6.5y).
Statistical analysis
We used random survival forest to identify the most important variables predicting mortality after 9-month follow-up (9-month values and changes from baseline). There were 36 potential predictors: 20 variables collected at 9-month follow-up and 16 variables calculated as a change from baseline to 9-month, plus age, sex, and ART use at 9 months, which we specified a priori as included in the model. Two measures of variable importance were used in variable selection: permutation importance (VIMP) and minimal depth. To identify variables for inclusion in a subsequent model, we eliminated variables that were not among the 20 most important by either measure and repeated the same selection procedure on this smaller set of variables.
Both results at 9 months and change from baseline to 9 months were significant predictors. We therefore tested combinations of variables (e.g. value at 9 months of one variable and change from baseline to 9 months of another variable) in separate analyses using both random survival forests and the Cox proportional hazards model. We used the out-of-bag (OOB) error rate and integrated area under the curve (AUC) as the measure of accuracy for the random survival forest and Cox models separately (Heagerty & Zheng, 2005).
We describe the association of each variable with time of death using hazard ratios (HR), 95% confidence intervals, and P-values. Two-tailed P-values < 0.05 were defined as statistically significant. Statistical analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC) and R version 3.4.2. Random survival forests were implemented using the “randomForestSRC” R-package version 3.4.2.50.
Results
Cohort characteristics
Among 1,154 participants with valid SAIDs, 1,020 were alive at 9-months; of those, 900 (88%) had all baseline and 9-month data available. Median (IQR) age was 34 (28–42) years, 49% were female, and 109 (12%) participants died after 9-month follow up. Participants with data available at both time points and those excluded because of missing data had a similar gender distribution; however, included participants were older than those excluded (median 35 vs 32 years, p=0.015). At baseline, 906 (89%) had some high school education or greater and 41% worked outside the home ≥40 hours/week (Table 1). At baseline, median CD4 count was 215cells/μL (IQR: 91–364cells/μL). By 9-month follow up, 310 (55% of eligible) participants had started ART. Fewer participants, 242 (27%) had received medication to treat or prevent HIV-related OIs.
Table 1.
Cohort characteristics for HIV-infected participants at baseline and 9 months
| Baseline n=1020 | 9 months n=1020 | Δ Baseline to 9 months | ||
|---|---|---|---|---|
| Age, yrs | ||||
| Median (IQR) | 34 (28–42) | |||
| Sex, n (%) | ||||
| Male | 520 (51) | |||
| Female | 500 (49) | |||
| Missing | -- | |||
| Marital status, n (%) | ||||
| Never married | 801 (79) | 700 (77) | Same | 711 (79) |
| Currently married | 162 (16) | 176 (19) | Change | 192 (21) |
| Divorce/separated/ Widowed | 52 (5) | 31 (3) | ||
| Missing | 5 | 113 | ||
| Education, n (%) | ||||
| Some high school or Greater | 906 (89) | |||
| Primary school or less | 108 (11) | |||
| Missing | 6 | |||
| Mode of transport, n (%) | ||||
| Public transport (bus, taxi) | 484 (48) | |||
| Private transport | 354 (35) | |||
| Other | 177 (17) | |||
| Missing | 5 | |||
| Distance from clinic, n (%) | ||||
| Less than 5 km | 156 (15) | |||
| At least 5 km | 859 (85) | |||
| Missing | 5 | |||
| Work hours outside home, n (%) | ||||
| None | 394 (39) | 429 (47) | Less work outside home | 319 (35) |
| Less than 40 hours | 208 (20) | 242 (27) | No change | 421 (46) |
| 40 hours or more | 418 (41) | 238 (26) | More work outside home | 169 (19) |
| Missing | -- | 111 | ||
| Prior HIV testing, n (%) | ||||
| Yes | 249 (25) | |||
| No | 766 (75) | |||
| Missing | 5 | |||
| Health care use in prior year, n (%) | ||||
| None | 175 (17) | 10 (1) | Less healthcare use | 213 (24) |
| 1–2 times | 273 (27) | 257 (28) | No change | 299 (33) |
| 3–5 times | 365 (36) | 425 (47) | More healthcare use | 391 (43) |
| >5 times | 202 (20) | 216 (24) | ||
| Missing | 5 | 112 | ||
| Visit to traditional healer in prior year, n (%) | ||||
| Yes | 354 (35) | 70 (8) | Stop visiting | 290 (32) |
| No | 661 (65) | 839 (92) | No change | 571 (63) |
| Missing | 5 | 111 | Start visiting | 43 (5) |
| Social support score | ||||
| Median (IQR) | 64 (50, 81) | 54 (48, 62) | Change | −6 (−25, 6) |
| Missing | 5 | 115 | ||
| Mental health score | ||||
| Median (IQR) | 64 (56, 76) | 56 (44, 92) | Change | −4 (−16, 12) |
| Missing | 5 | 112 | ||
| Reported barriers to healthcare, n (%) | ||||
| Yes | 366 (36) | 243 (27) | Yes to no | 210 (23) |
| No | 649 (64) | 665 (73) | No change | 563 (62) |
| No to yes | 130 (14) | |||
| Number of barriers for participants reporting barriers | ||||
| Median (IQR) | 4 (2, 5) | 2 (1, 3) | Change | −2 (−4, 1) |
| Number of barrier domains for participants reporting barriers | ||||
| Median (IQR) | 3 (2, 4) | 2 (1, 3) | Change | −1 (−3, 1) |
| Gone without healthcare for basic needs, n (%) | ||||
| Yes | 195 (19) | 54 (6) | Yes to no | 141 (16) |
| No | 825 (81) | 855 (94) | No change | 732 (80) |
| Missing | -- | 111 | No to yes | 36 (4) |
| Gone without basic needs for healthcare, n (%) | ||||
| Yes | 157 (15) | 54 (6) | Yes to no | 115 (13) |
| No | 863 (85) | 855 (94) | No change | 754 (83) |
| Missing | -- | 111 | No to yes | 40 (4) |
Changes in characteristics measures at baseline and 9 months
Between baseline and 9 months, 192 (21%) participants changed their marital status, with more people married at 9 months (176, 19%) than at baseline (162, 16%). Thirty-five percent of participants reported fewer work hours at 9 months compared to baseline, whereas only 19% reported more work hours at 9 months. Reported frequency of healthcare use in the prior year increased for 43% of participants, whereas 24% reported less frequent use. More participants (290, 32%) reported they stopped visiting a traditional healer than that they had started visiting one (43, 5%).
Median (IQR) social support score decreased from 64 (50, 81) at baseline to 54 (48, 62) at 9-months; the median change was −6 (−25, 6). Mental health also decreased, with a median score of 64 (56, 76) at baseline to 56 (44, 92) at 9-months, with median change of −4 (−16, 12).
Fourteen percent (130) of participants reported barriers to care at 9 months after not experiencing barriers at baseline, but more participants reported the opposite; 23% (210) of participants reported experiencing barriers at baseline but then not experiencing any at 9 months. For those reporting barriers at either timepoint, the median number of barriers decreased by 2 barriers over the 9 months.
Participants’ reports of going without healthcare for basic needs or going without basic needs for healthcare both decreased between baseline and 9-months. At baseline 19% (195) reported having gone without healthcare for basic needs, whereas 6% (54) reported this competing need at 9 months. While 15% (157) reported having gone without basic needs to have money for healthcare at baseline, only 6% (54) reported having to make this choice at 9 months.
Results for individual predictors of mortality
Reporting any competing needs at 9-months led to a higher mortality risk than reporting those competing needs at baseline, compared to reporting no competing needs. At baseline, hazard ratios ranged from 1.29 (95% CI: 0.86–1.93, P = 0.22) for reporting forgoing healthcare for basic needs to 1.38 (95% CI: 0.93–2.05, P = 0.11) for those reporting either forgoing healthcare or basic needs for the other (Table 2). At 9 months, hazard ratios were much higher, ranging from 2.97 (95% CI: 1.84–4.79, P < 0.01) for reporting either forgoing healthcare or basic needs for the other to 3.54 (95% CI: 2.15–5.82; P < 0.01) for reporting forgoing basic needs for healthcare. When compared to those reporting no competing needs at both baseline and 9 months (4 groups), those who did not report competing needs at baseline but did report them at 9-months (HR 3.40, 95% CI: 1.93–6.00, P < 0.01), or who reported them at both time points (HR 2.86, 95% CI: 1.24–6.61, P = 0.01) showed significant increased risk for long-term mortality.
Table 2.
Individual predictors of mortality risk for HIV-infected individuals
| Factor | Hazard ratio (95% CI) | p-value |
|---|---|---|
|
Core Variables
| ||
| Older (per 1 year) | 1.03 (1.02–1.05) | <0.01 |
| Male | 1.47 (1.03–2.09) | 0.03 |
| Did not start ART by 9 months | 1.37 (0.92–2.03) | 0.12 |
| Baseline a | ||
| Less social support (per 10 points) | 1.04 (0.96–1.14) | 0.35 |
| Lower mean mental health (per 10 points) | 1.00 (0.89–1.12) | 0.96 |
| Went without healthcare for basic needs | 1.29 (0.86–1.93) | 0.22 |
| Went without basic needs for healthcare | 1.30 (0.84–2.00) | 0.24 |
| Went without healthcare or basic needs | 1.38 (0.93–2.05) | 0.11 |
|
| ||
|
9 months
a
| ||
| No OI treatment | 1.72 (1.04–2.84) | 0.03 |
| Less social support (per 10 points) | 1.22 (1.08–1.39) | <0.01 |
| Lower mental health (per 10 points) | 1.12 (1.02–1.23) | 0.02 |
| Went without healthcare for basic needs | 3.22 (1.94–5.36) | <0.01 |
| Went without basic needs for healthcare | 3.54 (2.15–5.82) | <0.01 |
| Went without healthcare or basic needs | 2.97 (1.84–4.79) | <0.01 |
|
| ||
|
Δ Baseline to 9 months
a
| ||
| Increase social supportb | 0.80 (0.46–1.39) | 0.43 |
| Decrease social supportb | 1.00 (0.68–1.48) | 0.99 |
| Increase mental healthb | 0.96 (0.56–1.66) | 0.89 |
| Decrease mental healthb | 1.78 (1.18–2.70) | 0.01 |
| Competing needs: no to yes | 3.40 (1.93–6.00) | <0.01 |
| Competing needs: yes to no | 1.41 (0.91–2.17) | 0.12 |
| Competing needs: yes to yes | 2.86 (1.24–6.61) | 0.01 |
Adjusted for age, gender, starting ART by 9 months.
Change of >10 points from baseline
Baseline social support or mental health measurements did not significantly affect mortality, but 9-month measurements increased mortality risk. At 9-month follow-up, a lower social support score increased risk of death by 22% per 10-point decrease (HR 1.22, 95% CI: 1.08–1.39, P < 0.01) and a lower mental health score increased risk of death by 12% per 10-point decrease (HR 1.12, 95% CI: 1.02–1.23, P = 0.02). For individuals whose mental health score decreased by >10 points between baseline and 9-months, the risk of long-term mortality increased by 78%, versus those with a minimal change (HR 1.78, 95% CI: 1.18–2.70, P = 0.01).
Multivariable predictors of mortality
Each additional year of age increased mortality by 4% (HR 1.04, 95% CI: 1.02–1.05, P < 0.01) (Table 3). Being male also increased the risk of mortality (HR 1.37, 95% CI 0.94–2.01, P = 0.11). Although not statistically significant, participants who had not started ART by 9 months showed greater risk for long-term mortality (HR 1.46, 95% CI: 0.96–2.22, P = 0.08). Likewise, if participants reported not receiving medication for treatment or prevention of OIs at 9 months, they were 67% more likely to die in the subsequent 5 years (HR 1.67, 95% CI: 0.98–2.83, P = 0.06), albeit, non-significantly.
Table 3.
Cox model of factors related to mortality after 9 months
| Factor | Adj hazard ratio (95% CI) | p-value |
|---|---|---|
| Older | 1.04 (1.02–1.05) | <0.01 |
| Male | 1.37 (0.94–2.01) | 0.11 |
| Did not start ART by 9 months | 1.46 (0.96–2.22) | 0.08 |
| Did not receive OI treatment | 1.67 (0.98–2.83) | 0.06 |
| Social support at 9 monthsa | 1.16 (1.01–1.32) | 0.03 |
| Lower mental health score at 9 months compared to baselineb | 1.06 (0.96–1.18) | 0.26 |
| Went without basic needs or healthcare at 9 months | 2.62 (1.55–4.42) | <0.01 |
For every 10-point decrease in social support on 100-point scale
For every 10-point decrease in mental health score at 9 months compared to baseline
The level of social support at 9 months was important; for every ten-point decrease in social support score at 9-months, the hazard of death increased by 16% (HR 1.16, 95% CI: 1.01–1.32, P = 0.03). A ten-point decrease in mental health score between baseline and 9 months only marginally increased mortality risk (HR 1.06, 95% CI: 0.96–1.18, P = 0.26).
Competing needs, choosing between paying for healthcare or basic needs, showed the largest impact on long-term mortality risk. Participants who went without basic needs or healthcare at 9 months had a 2.62 times higher hazard of death compared to participants who did not report competing needs at 9 months (HR 2.62, 95% CI: 1.55–4.42, P <0.01).
Discussion
Among 900 PLWH who had complete baseline and 9-month data and were enrolled at outpatient sites in Durban, South Africa, less social support and competing needs at 9-month follow-up significantly increased mortality at 5 years. Risk of death at 5 years increased by 46% for those who had not started ART and by 67% for those who had not received OI prophylaxis by 9-months. There was a significant effect of mortality with age; each additional year of life increased risk of 5-year mortality by 4%. However, lows social support and decreases in mental health at 9 months and evidence of competing needs increased 5-year mortality risk beyond these known contributors to mortality. Social support was a significant contributor to mortality risk, with every 10-point decrease in social support at 9 months contributing to a 16% increase in 5-year mortality risk. Moreover, evidence of competing needs at 9-months showed a large impact on mortality, increasing 5-year risk of death almost 3-fold.
Levels of social support and social capital are inversely related to morbidity and mortality among PLWH in sub-Saharan Africa (Croome et al., 2017; Mukoswa et al., 2017; Ncama et al., 2008). Furthermore, PLWH report lower levels of social support than HIV-uninfected counterparts (Drain et al., 2015); social support is likely to change over time, especially following HIV diagnosis, given the potential effects of HIV-associated stigma (Bonnington et al., 2017; (GNP+) GNPoPLwH, 2015). We found that participants who reported lower social support at 9 months following HIV diagnosis had significantly higher mortality risk. Therefore, monitoring social support over time may be important for mitigating risk. In this study, we used a 13-question Social Support questionnaire, which could be repurposed for use in clinical settings (Sherbourne & Stewart, 1991). In other settings, social integration, as a type of social support, has been shown to be most predictive of mortality (Holt-Lunstad et al., 2010, 2017). Measuring social integration over time and developing methods to augment it could be helpful to health service providers caring for PLWH.
Our univariate analysis suggested that decreasing mental health over time can negatively impact survival rates. We and others have found a correlation between depressive symptoms and decreased likelihood of obtaining a CD4 count or taking ART in sub-Saharan Africa (Kinyanda et al., 2018; Ramirez-Avila et al., 2012). There is an urgent need for increased integration between mental healthcare delivery and HIV services (Dos Santos & Wolvaardt, 2016). Furthermore, improving mental health alongside social support further increases ART adherence and decreases mortality (Huynh et al., 2013). Thus, identifying people in need of psychosocial interventions through integration of mental healthcare and HIV services may be important for decreasing mortality risk. Mental health screening could be accomplished using a short 5-item Mental Health Inventory questionnaire (Hays et al., 1993), both at entry into HIV services and on an ongoing basis to mitigate long-term mortality risk.
Participants in the Sizanani Trial reported high rates of competing needs, such as going without healthcare for basic needs or vice versa (Drain et al., 2013). We found that reporting competing needs 9-months after HIV diagnosis had the largest impact on mortality risk among all variables examined, increasing risk of death by almost 3-fold. This finding expands upon previous studies showing competing needs negatively affect ART and appointment adherence and viral load status (Croome et al., 2017; Palar et al., 2018). While the proportion of participants reporting competing needs decreased between baseline and 9-month follow-up, despite fewer people working, participants may have received social grants due to disability post-HIV diagnosis. Nevertheless, monitoring competing needs over time and facilitating additional interventions that could reduce competing needs for PLWH in South Africa is of paramount importance to improving survival.
This research must be considered in the context of its limitations. Only 52% of participants provided valid SAIDs for death registry cross-matching. We previously noted that participants with valid SAID’s did not differ by gender, but were slightly older (30y versus 27y) (Bassett et al., 2019). Moreover, 12% of participants alive at 9 months did not provide complete 9-month follow-up data and thus our comparisons were limited to those who provided both baseline and 9-month data. It is possible that participants’ answers at 9 months were affected by social desirability bias, although this would be expected to lessen the observed effect of social support and competing needs on mortality. We did not collect data on HIV status disclosure and thus cannot relate disclosure to perceived social support. We also did not measure income or poverty directly, potential mortality risk factors, though competing needs may be a proxy. It may not be feasible to use the methods employed in this study to determine risk factors for higher mortality in a clinical setting. Future steps should include the development of predictive instruments that can be implemented within HIV services to identify people who are at-risk for psychosocial reasons.
HIV-associated mortality in South Africa remains high. Understanding how changes in contextual factors over time might influence outcomes for PLWH in South Africa is important for decreasing long-term mortality. We found that less social support as well as competing needs at 9-month follow-up significantly increased long-term mortality risk. Reassessing contextual factors during follow-up and targeting interventions to increase social support, mental health, and affordability of seeking care may reduce long-term mortality for PLWH in South Africa.
Acknowledgements
We would like to acknowledge the assistance of Ogochukwu Ufio in the preparation of this manuscript.
Funding
This work was supported by the US National Institute of Mental Health [grant number R01 MH090326 and R01 MH108427]; and Weissman Family MGH Research Scholar Award. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the US National Institutes of Health or the MGH Executive Committee on Research.
Footnotes
Disclosure statement
All authors declare that they have no conflicts of interest.
Data availability statement
The data that support the findings of this study are available from the corresponding author, [IVB], upon reasonable request.
References
- Bassett IV, Coleman SM, Giddy J, Bogart LM, Chaisson CE, Ross D, Flash MJE, Govender T, Walensky RP, Freedberg KA, & Losina E (2017). Barriers to Care and 1-Year Mortality Among Newly Diagnosed HIV-Infected People in Durban, South Africa. Journal of Acquired Immune Deficiency Syndromes (1999), 74(4), 432–438. 10.1097/QAI.0000000000001277 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bassett IV, Coleman SM, Giddy J, Bogart LM, Chaisson CE, Ross D, Jacobsen MM, Robine M, Govender T, Freedberg KA, Katz JN, Walensky RP, & Losina E (2016). Sizanani: A Randomized Trial of Health System Navigators to Improve Linkage to HIV and TB Care in South Africa. J Acquir Immune Defic Syndr, 73(2), 154–160. 10.1097/qai.0000000000001025 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bassett IV, Giddy J, Chaisson CE, Ross D, Bogart LM, Coleman SM, Govender T, Robine M, Erlwanger A, Freedberg KA, Katz JN, Walensky RP, & Losina E (2013). A randomized trial to optimize HIV/TB care in South Africa: Design of the Sizanani trial. BMC Infect Dis, 13, 390. 10.1186/1471-2334-13-390 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bassett IV, Xu A, Giddy J, Bogart LM, Boulle A, Millham L, Losina E, & Parker RA (2019). Assessing rates and contextual predictors of 5-year mortality among HIV-infected and HIV-uninfected individuals following HIV testing in Durban, South Africa. BMC Infectious Diseases, 19(1), 751. 10.1186/s12879-019-4373-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bogart LM, Chetty S, Giddy J, Sypek A, Sticklor L, Walensky RP, Losina E, Katz JN, & Bassett IV (2013). Barriers to Care among People Living with HIV in South Africa: Contrasts between Patient and Healthcare Provider Perspectives. AIDS Care, 25(7), 843–853. 10.1080/09540121.2012.729808 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bonnington O, Wamoyi J, Ddaaki W, Bukenya D, Ondenge K, Skovdal M, Renju J, Moshabela M, & Wringe A (2017). Changing forms of HIV-related stigma along the HIV care and treatment continuum in sub-Saharan Africa: A temporal analysis. Sexually Transmitted Infections, 93(Suppl 3). 10.1136/sextrans-2016-052975 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cornell M, Johnson LF, Wood R, Tanser F, Fox MP, Prozesky H, Schomaker M, Egger M, Davies M-A, & Boulle A (2017). Twelve-year mortality in adults initiating antiretroviral therapy in South Africa. Journal of the International AIDS Society, 20(1). 10.7448/IAS.20.1.21902 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cornell M, Lessells R, Fox MP, Garone DB, Giddy J, Fenner L, Myer L, Boulle A, & Ie DEASAC (2014). Mortality among adults transferred and lost to follow-up from antiretroviral therapy programmes in South Africa: A multicenter cohort study. J Acquir Immune Defic Syndr, 67(2), e67–75. 10.1097/QAI.0000000000000269 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Craw JA, Gardner LI, Marks G, Rapp RC, Bosshart J, Duffus WA, Rossman A, Coughlin SL, Gruber D, Safford LA, Overton J, & Schmitt K (2008). Brief strengths-based case management promotes entry into HIV medical care: Results of the antiretroviral treatment access study-II. J Acquir Immune Defic Syndr, 47(5), 597–606. 10.1097/QAI.0b013e3181684c51 [DOI] [PubMed] [Google Scholar]
- Croome N, Ahluwalia M, Hughes LD, & Abas M (2017). Patient-reported barriers and facilitators to antiretroviral adherence in sub-Saharan Africa. AIDS (London, England), 31(7), 995–1007. 10.1097/QAD.0000000000001416 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cunningham WE, Andersen RM, Katz MH, Stein MD, Turner BJ, Crystal S, Zierler S, Kuromiya K, Morton SC, St Clair P, Bozzette SA, & Shapiro MF (1999). The impact of competing subsistence needs and barriers on access to medical care for persons with human immunodeficiency virus receiving care in the United States. Medical Care, 37(12), 1270–1281. 10.1097/00005650-199912000-00010 [DOI] [PubMed] [Google Scholar]
- Dos Santos M, & Wolvaardt G (2016). Integrated intervention for mental health co-morbidity in HIV-positive individuals: A public health assessment. African Journal of AIDS Research: AJAR, 15(4), 325–331. 10.2989/16085906.2016.1229683 [DOI] [PubMed] [Google Scholar]
- Drain PK, Losina E, Coleman SM, Bogart L, Giddy J, Ross D, Katz JN, & Bassett IV (2015). Social support and mental health among adults prior to HIV counseling and testing in Durban, South Africa. AIDS Care, 27(10), 1231–1240. 10.1080/09540121.2015.1046417 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Drain PK, Losina E, Parker G, Giddy J, Ross D, Katz JN, Coleman SM, Bogart LM, Freedberg KA, Walensky RP, & Bassett IV (2013). Risk Factors for Late-Stage HIV Disease Presentation at Initial HIV Diagnosis in Durban, South Africa. PLoS ONE, 8(1), e0055305. 10.1371/journal.pone.0055305 [DOI] [PMC free article] [PubMed] [Google Scholar]
- (GNP+) GNPoPLwH. (2015). The people living with HIV stigma index. https://www.stigmaindex.org/
- Hays RD, Sherbourne CD, & Mazel RM (1993). The RAND 36-Item Health Survey 1.0. Health Economics, 2(3), 217–227. 10.1002/hec.4730020305 [DOI] [PubMed] [Google Scholar]
- Heagerty PJ, & Zheng Y (2005). Survival model predictive accuracy and ROC curves. Biometrics, 61(1), 92–105. 10.1111/j.0006-341X.2005.030814.x [DOI] [PubMed] [Google Scholar]
- Holt-Lunstad J, Robles T, & Sbarra DA (2017). Advancing Social Connection as a Public Health Priority in the United States. The American Psychologist, 72(6), 517–530. 10.1037/amp0000103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Holt-Lunstad J, Smith TB, & Layton JB (2010). Social relationships and mortality risk: A meta-analytic review. PLoS Medicine, 7(7), e1000316. 10.1371/journal.pmed.1000316 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Huynh AK, Kinsler JJ, Cunningham WE, & Sayles JN (2013). The role of mental health in mediating the relationship between social support and optimal ART adherence. AIDS Care, 25(9), 1179–1184. 10.1080/09540121.2012.752787 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Johnson LF, Dorrington RE, Laubscher R, Hoffmann CJ, Wood R, Fox MP, Cornell M, Schomaker M, Prozesky H, Tanser F, Davies M-A, & Boulle A (2015). A comparison of death recording by health centres and civil registration in South Africans receiving antiretroviral treatment. Journal of the International AIDS Society, 18(1). 10.7448/IAS.18.1.20628 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Katz IT, Dietrich J, Tshabalala G, Essien T, Rough K, Wright AA, Bangsberg DR, Gray GE, & Ware NC (2015). Understanding treatment refusal among adults presenting for HIV-testing in Soweto, South Africa: A qualitative study. AIDS and Behavior, 19(4), 704–714. 10.1007/s10461-014-0920-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kinyanda E, Levin J, Nakasujja N, Birabwa H, Nakku J, Mpango R, Grosskurth H, Seedat S, Araya R, Shahmanesh M, & Patel V (2018). Major Depressive Disorder: Longitudinal Analysis of Impact on Clinical and Behavioral Outcomes in Uganda. Journal of Acquired Immune Deficiency Syndromes (1999), 78(2), 136–143. 10.1097/QAI.0000000000001647 [DOI] [PubMed] [Google Scholar]
- Layer EH, Kennedy CE, Beckham SW, Mbwambo JK, Likindikoki S, Davis WW, Kerrigan DL, Brahmbhatt H, & Tanzania LTC Collaborative Study Team. (2014). Multi-level factors affecting entry into and engagement in the HIV continuum of care in Iringa, Tanzania. PLoS One, 9(8), e104961. 10.1371/journal.pone.0104961 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mills EJ, Bakanda C, Birungi J, Chan K, Ford N, Cooper CL, Nachega JB, Dybul M, & Hogg RS (2011). Life expectancy of persons receiving combination antiretroviral therapy in low-income countries: A cohort analysis from Uganda. Ann Intern Med, 155(4), 209–216. 10.7326/0003-4819-155-4-201108160-00358 [DOI] [PubMed] [Google Scholar]
- Mukoswa GM, Charalambous S, & Nelson G (2017). The association between social capital and HIV treatment outcomes in South Africa. PloS One, 12(11), e0184140. 10.1371/journal.pone.0184140 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ncama BP, McInerney PA, Bhengu BR, Corless IB, Wantland DJ, Nicholas PK, McGibbon CA, & Davis SM (2008). Social support and medication adherence in HIV disease in KwaZulu-Natal, South Africa. International Journal of Nursing Studies, 45(12), 1757–1763. 10.1016/j.ijnurstu.2008.06.006 [DOI] [PubMed] [Google Scholar]
- Palar K, Wong MD, & Cunningham WE (2018). Competing subsistence needs are associated with retention in care and detectable viral load among people living with HIV. Journal of HIV/AIDS & Social Services, 17(3), 163–179. 10.1080/15381501.2017.1407732 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ramirez-Avila L, Regan S, Giddy J, Chetty S, Ross D, Katz JN, Freedberg KA, Walensky RP, Losina E, & Bassett IV (2012). Depressive symptoms and their impact on health-seeking behaviors in newly-diagnosed HIV-infected patients in Durban, South Africa. AIDS and Behavior, 16(8), 2226–2235. 10.1007/s10461-012-0160-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rand Corporation. (n.d.). HIV Cost and Services Utilization Study. Retrieved October 31, 2014, from https://www.rand.org/health-care/projects/hcsus.html
- Sherbourne CD, & Stewart AL (1991). The MOS social support survey. Social Science & Medicine (1982), 32(6), 705–714. 10.1016/0277-9536(91)90150-b [DOI] [PubMed] [Google Scholar]
- Tomori C, Kennedy CE, Brahmbhatt H, Wagman JA, Mbwambo JK, Likindikoki S, & Kerrigan DL (2014). Barriers and facilitators of retention in HIV care and treatment services in Iringa, Tanzania: The importance of socioeconomic and sociocultural factors. AIDS Care, 26(7), 907–913. 10.1080/09540121.2013.861574 [DOI] [PubMed] [Google Scholar]
- UNAIDS. (2017). UNAIDS DATA 2017. Joint United Nations Programme on HIV/AIDS. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author, [IVB], upon reasonable request.