AIDS and Behavior

, Volume 18, Issue 2, pp 368–380

Community-Based Accompaniment and Psychosocial Health Outcomes in HIV-Infected Adults in Rwanda: A Prospective Study

Authors

    • Department of Global Health and Social MedicineHarvard Medical School
  • Michael L. Rich
    • Department of Global Health and Social MedicineHarvard Medical School
    • Partners in Health/Inshuti Mu Buzima
    • Division of Global Health EquityBrigham and Women’s Hospital
  • Felix Kaigamba
    • Rwanda Ministry of Health
  • Adrienne R. Socci
    • Partners in Health/Inshuti Mu Buzima
  • Massudi Hakizamungu
    • Partners in Health/Inshuti Mu Buzima
  • Emmanuel Bagiruwigize
    • Rwanda Ministry of Health
  • Agnes Binagwaho
    • Department of Global Health and Social MedicineHarvard Medical School
    • Rwanda Ministry of Health
  • Molly F. Franke
    • Department of Global Health and Social MedicineHarvard Medical School
    • Partners in Health/Inshuti Mu Buzima
Original Paper

DOI: 10.1007/s10461-013-0431-2

Cite this article as:
Thomson, D.R., Rich, M.L., Kaigamba, F. et al. AIDS Behav (2014) 18: 368. doi:10.1007/s10461-013-0431-2

Abstract

We examined whether the addition of community-based accompaniment to Rwanda’s national model for antiretroviral treatment (ART) was associated with greater improvements in patients’ psychosocial health outcomes during the first year of therapy. We enrolled 610 HIV-infected adults with CD4 cell counts under 350 cells/μL initiating ART in one of two programs. Both programs provided ART and required patients to identify a treatment buddy per national protocols. Patients in one program additionally received nutritional and socioeconomic supplements, and daily home-visits by a community health worker (“accompagnateur”) who provided social support and directly-observed ingestion of medication. The addition of community-based accompaniment was associated with an additional 44.3 % reduction in prevalence of depression, more than twice the gains in perceived physical and mental health quality of life, and increased perceived social support in the first year of treatment. Community-based accompaniment may represent an important intervention in HIV-infected populations with prevalent mental health morbidity.

Keywords

Community health workerHIVARTMental healthMixed-effect

Resumen

Este estudio evaluó si el agregado del acompañamiento comunitario al modelo nacional de tratamiento antirretroviral (TARV) utilizado en Ruanda, se asocia a mejores resultados en la salud psicosocial de los pacientes durante el primer año de tratamiento. Se enrolaron 610 adultos infectados con VIH, con recuento de células CD4 inferior a 350 por microlitro y que iniciaron el TARV en uno de dos programas. Ambos programas proporcionaron tratamiento antirretroviral y, según protocolo nacional, se les solicitó a los pacientes la identificación de un compañero para el tratamiento. En uno de los programas, los pacientes recibieron además: ayuda nutricional y socioeconómica y visitas diarias de agentes sanitarios de la comunidad (acompañadores) que proporcionaron apoyo social y que vigilaron directamente la toma de los medicamentos. El acompañamiento comunitario se asoció a una reducción adicional de 44,3 % en la prevalencia de depresión, a un aumento de más del doble en la calidad de vida física y mental percibida y a un aumento de la percepción del apoyo social durante el primer año de tratamiento. El acompañamiento comunitario puede representar una importante intervención en las poblaciones infectadas por el VIH y con una predominante morbilidad en la salud mental.

Introduction

Depression, lack of perceived social support and low health-related quality of life are prevalent among individuals living with human immunodeficiency virus (HIV) [14]. Among studies of mental health in HIV patients living in resource-poor settings worldwide, people living with HIV have higher prevalence of depressive symptoms than sero-negative individuals, and mental illness (including adjustment disorder, depressive mood, and major depression) worsens with advanced stages of HIV and AIDS. The same studies show that patients who experience mood disturbance are more likely to report poor mental health quality of life, and that worsening physical symptoms of HIV are associated with poorer perception of physical health quality of life and anxiety [5].

Poor mental health can be a risk factor for HIV-risky behavior [6], a product of HIV diagnosis [7], and a risk factor for poor adherence to HIV treatment [8]. Although lifesaving antiretroviral therapy (ART) may have a positive impact on psychosocial outcomes [911], mental health morbidity that persists after ART initiation may lead to decreased adherence and hasten progression of HIV to AIDS [1218]. The adverse effects of depression, low social support, and low quality of life extend beyond HIV-related outcomes and may also lead to negative life events including job loss, marital problems, and lack of social agency [2, 19]. Given the simultaneous and compounding impacts of HIV-infection and poor mental health on a patient’s physical health and general well-being, it is important to evaluate and integrate effective mental health and psychosocial interventions into the delivery of HIV care.

To be eligible for national HIV care in Rwanda, patients are required to self-identify a treatment buddy, typically a family member or friend, to attend a 2- to 3-day education session and provide voluntary social support for ART [20]. The extent and type of support provided by treatment buddies is highly variable and dependent on the established relationship between the treatment buddy and patient. Treatment buddies have been shown in other resource-limited settings to improve adherence [2124] and immunological outcomes [25] via improved social support and social capital. Patients who rely exclusively on a self-identified treatment buddy, however, often still experience fear of disclosure, inability to pay for supplemental food and transportation, fear of ART side effects, and difficulty making sense of conflicting health information [26, 27].

To fill these health support and service gaps, and identify and address any other barriers to care and adherence, Partners In Health, in collaboration with the Rwanda Ministry of Health (MOH), implemented a community-based accompaniment program for HIV care and ART delivery in rural, resource-limited Rwanda in 2005 [28]. The strategy consisted of (a) daily home-visits by a trained community health worker (CHW), called an “accompagnateur”, who provided social support and directly-observed ART; (b) nutritional and transportation support; and (c) other socioeconomic support as needed. We previously reported that community-based accompaniment was associated with higher treatment retention with viral load suppression, and a lower rate of death and loss to follow up in this cohort [29].

CHWs were a cornerstone of this intervention and were trained and paid to deliver comprehensive support, including emotional and adherence support, while building trust and providing ongoing HIV education and advocacy to the patients to whom they were matched. Previously implemented for HIV and multidrug-resistant tuberculosis care in Peru, Haiti, and the United States, this model has been shown to be associated with improved adherence, ability to resume social roles, increased self-efficacy, decreased hopelessness, positive changes in non-HIV-related behavior (e.g. substance use, interactions with others, maturity, and physical appearance), reduced internalized and externalized stigma, and improved ability to disclose HIV-status [3032].

We hypothesized that the additional support and services provided by the accompaniment program in Rwanda would lead to greater improvements in participant’s’ psychosocial health compared to clinical-care alone. We conducted a prospective cohort study to examine whether the addition of community-based accompaniment to the Rwanda national model for ART delivery was associated with greater improvements in social support, perceived mental and physical health quality of life, and depression during the first year of ART in a cohort of HIV-infected adults in rural Rwanda.

Methods

Ethics Statement

The Partners Human Research Committee, Boston, USA and the Rwanda National Ethics Committee, Rwanda approved this study.

Study Setting

HIV prevalence in rural Rwanda is estimated to be 2.6 % [33]. The Rwanda MOH, in collaboration with Partners In Health, implemented community-based accompaniment in addition to the Rwanda national model for ART delivery in southeastern Rwanda in 2005. We selected comparison ART sites in Musanze district in northern Rwanda, which was considered to have a well-functioning clinic-based program that was representative of Rwanda’s national model for ART delivery, but lacked a non-governmental partner that provided additional program and patient support. In both study regions, ART was provided first through a district hospital and then decentralized to smaller satellite clinics. Each district hospital is located 2 h by car from the capital city of Kigali. The Global Fund to Fight AIDS, TB, and Malaria provided funding to all of the health centers in the study.

Study Population

To reach our target sample size of 300 participants per region, we enrolled individuals from 5 health centers in Kayonza and Kirehe districts where the community-based accompaniment model was added to the national model, and 4 health centers in Musanze district that implemented the clinic-based national model alone. We consecutively enrolled consenting HIV-infected adults age 21 years and older with a CD4 cell count less than 350 cell/μL who initiated ART at one of these sites between June 2007 and August 2008. Participants were eligible if they were residents of one of the two catchment areas, planned to reside in that area for at least 1 year, and had not previously initiated lifelong ART.

Rwanda National Model for ART Delivery

ART was provided free-of-charge to all individuals who met MOH eligibility criteria. Cotrimoxazole was routinely prescribed to individuals with CD4 cell counts less than 350 cells/μL or with WHO HIV disease stage 2 or above. CD4 cell counts were conducted prior to ART initiation and repeated every 6 months. According to the Rwanda Ministry of Health guidelines for the provision of HIV care, all patients were urged to disclose their HIV status to at least one person and required to identify a treatment buddy in order to facilitate treatment adherence [34].

Community-Based Accompaniment for ART Delivery

In addition to receiving clinic-based care, individuals receiving community-based accompaniment were visited daily in their homes by a community health worker who provided social support, monitored for drug side effects and adverse events, identified potential barriers to adherence (e.g., lack of familial support, perceived stigma, other illness), and directly-observed ingestion of medications. A monthly food ration based on a family of four was provided for the first 10 months of ART, and transportation support was provided for routine clinic visits. Individuals were accompanied to clinic visits by CHWs for the first 4 monthly visits, and then as needed. Clinic-based social workers screened individuals and arranged for additional support as needed including transportation to clinic for routine follow-up, payment of school or health insurance fees, microfinance activities, employment assistance, or home repairs.

Data Collection

Clinic nurses and doctors prospectively collected clinical data as part of routine care at the time of enrollment and during the one-year follow-up period, using standardized forms. Additionally, the study team conducted interviews in the local language (Kinyarwanda) to obtain psychosocial data at the time of ART initiation (i.e., baseline) and after 6 and 12 months of treatment.

Outcomes and Measures

The primary outcomes of interest were (a) perceived functional social support, (b) perceived mental health quality of life, (c) perceived physical health quality of life, and (d) depression. To measure social support, we used the 8-item Duke UNC Functional Social Support Questionnaire (DUFSSQ). The DUFSSQ asks respondents to classify supportive situations such as “You get help when you are sick in bed” by the amount of support received relative to their ideal, and their responses are scaled 1–5 with higher scores indicating greater social support (range of possible scores: 8–40) [35].

The mental health and physical health subscales of the Medical Outcomes Survey HIV Scale were used to assess self-perceived mental and physical health quality of life [36]. The mental health subscale is comprised of five items that reflect major mental health dimensions including anxiety, depression, loss of behavioral/emotional control, and psychological wellbeing. The six items of the physical health subscale reflect physical function, pain, and role functioning. Subscales are scored through summation of the response choice item values, with higher scores indicating better health. To facilitate comparability, subscale scores are transformed to a 0–100 scale.

Depression was measured with the Hopkins Symptoms Checklist-15 (HSCL-15) which asks respondents to classify 15 symptoms (such as low energy, loneliness, and poor appetite) during the past week. Responses are scored on a severity scale from 1 to 4, with a higher score indicating more depressive symptoms [37, 38]. Individuals with a mean score greater than 1.75 were classified as depressed.

All instruments were translated from English to Kinyarwanda and then back-translated to English to ensure accuracy. Each scale demonstrated standardized Cronbach’s alpha coefficients of 0.79 or greater in the entire study population, and 0.75 or greater in subgroups of gender, literacy, and mode of ART delivery, indicating good reliability [39]. Results from a principal components analysis supported 1 factor structures of each scale, and we found expected correlations among scales indicating good construct validity in this population [39].

Data Analysis

We sought to answer the question, “Does the group of patients receiving community-based accompaniment differ from the group that received that national model alone in terms of change in psychosocial health outcomes during the first year of ART?” However, because we observed that the patients in the two programs had different baseline levels of perceived quality of life, social support, and depression, and because these baseline values may be important predictors of subsequent outcomes, we also adjusted for baseline measurements of psychosocial health outcomes. By adjusting for baseline difference, we addressed the question of whether psychosocial outcomes would be expected to differ across exposure groups (accompaniment vs. no accompaniment) given that the two groups had the same baseline distributions of psychosocial health morbidity. Because adjusting for baseline values can introduce spurious statistical association under some circumstances, we compared our results to a second analysis in which we did not adjust for baseline [40, 41].

We compared baseline prevalence of depression between the treatment groups using a Chi square test, and we compared baseline mental health quality of life score, physical health quality of life score, and social support score using t-tests. We used the following correlation coefficients to evaluate bivariate relationships among potential confounders and outcome variables, and to test for multicolinearity among potential confounders: Pearson’s (continuous variables), point biserial (binary and continuous/ranked variables), and phi (binary variables). Correlations with an absolute value greater than 0.2 were included in multivariate models. Covariates with an absolute correlation value greater than 0.8 were considered multicolinear and only the more strongly correlated variable was added to the model [42].

We used longitudinal mixed effects regression models with fixed effects estimators to control for potential confounders and quantify the association between treatment and psychosocial health outcomes after 6 and 12 months of ART. We evaluated and compared the results of two sets of models; one set that adjusted for baseline differences in psychosocial scores, and one set that did not (profile analysis). We controlled for potential confounders in each model by including all variables with a bivariate correlation coefficient greater than 0.2. In each treatment group, we estimated mean psychosocial outcomes at 6 and 12 months, and mean change from baseline after 6 and 12 months. Change from baseline was modeled for each individual, and averaged within each treatment group. We found that the magnitude, direction, and significance of results were similar across the two modeling approaches. Mathematical models for these analyses are shown in Appendix 1, and results from the profile analysis are reported in Appendix 2.

We multiply imputed missing baseline covariate data based on other covariate and outcome information using Markov Chain Monte Carlo methods (SAS MI Procedure). Correlations among repeated measures were modeled using a compound symmetry matrix in the adjust-for-baseline analysis, and an unstructured matrix in the profile analysis. We used the OpenMRS medical record system to double-enter and store data [43] and SAS software, version 9.2 (SAS Institute Inc., Cary, NC, USA) to conduct data analyses.

Results

Table 1 summarizes the characteristics of the 610 study participants. The two study populations, though similar in age, differed across a number of baseline characteristics. Individuals who received clinic-based care only (N = 306) demonstrated more advanced WHO HIV stages (p = 0.0009), greater immune suppression (p = 0.0002), and greater food insecurity (p < 0.0001) compared to individuals who received community-based accompaniment (N = 304). On the other hand, individuals who received community-based accompaniment generally appeared worse off in terms of access to care, and socioeconomic variables. Specifically, individuals in the community-based accompaniment treatment arm were less literate (p < 0.0001), had lower body mass indices (BMIs) (p = 0.0026), and lived farther from the clinic (p < 0.0001). Although uncommon, more participants in the clinic-based care group died (n = 22, 7.2 %) and were lost to follow up (n = 10, 3.3 %) than the accompaniment group (died: n = 13, 4.3 %; lost to follow up: n = 3, 1.0 %) [29].
Table 1

Baseline characteristics of study population

Variable (N)

Clinic-based care

Community-based accompaniment

Test statistic

p value

Continuous variables (median and range)

  

 Age in years (N = 610)

37 (21, 63)

37 (21, 80)

−0.63 (t test)

0.5259

 Baseline CD4 cell count (N = 610)

218.0 (0.0, 349.0)

241.5 (18.0, 350.0)

−3.70 (t test)

0.0002

 BMI (N = 606)

21.4 (13.4, 31.2)

20.3 (13.3, 31.6)

3.02 (t test)

0.0026

Categorical variables (frequency and percent)

  

 Female (N = 610)

201 (66 %)

175 (58 %)

4.25 (χ2)

0.0392

 Civil status (N = 610)

  

4.58 (χ2)

0.2052

  Never married

12 (4 %)

16 (5 %)

  

  Married or living with a partner

175 (57 %)

176 (58 %)

  

  Divorced or separated

25 (8 %)

35 (12 %)

  

  Widowed

94 (31 %)

74 (25 %)

  

 WHO HIV stage (N = 609)

  

16.42 (χ2)

0.0009

  1

44 (14 %)

73 (24 %)

  

  2

90 (30 %)

106 (35 %)

  

  3

164 (54 %)

117 (38 %)

  

  4

7 (2 %)

8 (3 %)

  

 TB treatment at ART start (N = 608)

8 (3 %)

15 (5 %)

2.21 (χ2)

0.1367

 ART regimen (N = 607)

  

18.12 (χ2)

0.0004

  D4T, 3TC, NVP

211 (69 %)

187 (62 %)

  

  ZDV, 3TC, NVP

57 (19 %)

96 (32 %)

  

  D4T, 3TC, EFV

9 (3 %)

9 (3 %)

  

  ZDV, 3TC, EFV

27 (9 %)

11 (4 %)

  

 Bactrim prophylaxis (N = 603)

302 (100 %)

301 (100 %)

(Fisher’s exact)

0.2757

 Able to read (N = 610)

234 (76 %)

173 (57 %)

26.28 (χ2)

<0.0001

 Time to clinic (N = 604)

  

189.97 (χ2)

<0.0001

  <30 min

106 (35 %)

23 (8 %)

  

  30–60 min

109 (36 %)

39 (13 %)

  

  1–2 h

80 (26 %)

106 (36 %)

  

  2–3 h

9 (3 %)

81 (27 %)

  

  >3 h

2 (1 %)

49 (16 %)

  

 Food insecurity category (N = 601)

  

59.60 (χ2)

<0.0001

  None

29 (10 %)

23 (8 %)

  

  Mild

15 (5 %)

9 (3 %)

  

  Moderate

32 (10 %)

110 (37 %)

  

  Severe

229 (75 %)

154 (52 %)

  
Figure 1 shows simple unadjusted trajectories of the mean social support, perceived mental and physical health quality of life, and depression scores plotted by treatment strategy. Table 2 compares simple unadjusted psychosocial outcomes at baseline. Statistically significant differences in psychosocial outcomes between the two groups were observed at baseline. Individuals in the accompaniment arm were more than twice as likely to be depressed, and perceive themselves to have lower mental and physical health quality of life at baseline relative to the individuals who received clinic-based care alone (p < 0.0001 for all comparisons). In contrast, participants in the accompaniment arm had a higher mean social support score (p < 0.0001), possibly reflecting their knowledge of enrollment in a program with community health worker support.
https://static-content.springer.com/image/art%3A10.1007%2Fs10461-013-0431-2/MediaObjects/10461_2013_431_Fig1_HTML.gif
Fig. 1

Trends in psychosocial health outcomes, by ART delivery strategy. The red solid line references patients receiving the national model plus community-based accompaniment, while the yellow dotted line references patients receiving the national model alone (Color figure online)

Table 2

Baseline psychosocial outcome measurements of study population, unadjusted

Variable (N)

Clinic-based care

Community-based accompaniment

Test statistic

p value

Continuous outcomes (median and range)

 Perceived social support (N = 603)

13 (8, 40)

18 (8, 40)

−5.97 (t test)

<0.0001

 Perceived mental health QOL (N = 597)

60.0 (5.0, 100.0)

44.0 (0.0, 100.0)

8.44 (t test)

<0.0001

 Perceived physical health QOL (N = 598)

100.0 (0.0, 100.0)

58.3 (0.0, 100.0)

14.47 (t test)

<0.0001

Categorical outcome (frequency and percent)

 Depression (N = 601)

83 (27 %)

200 (68 %)

98.18 (χ2)

<0.0001

Results of the multivariable analyses are shown in Tables 3 and 4, including adjusted estimates at 6 and 12 months, estimated change since baseline within each group, and comparison of change between the two groups (difference-in-differences). After adjusting for important covariates and baseline differences in psychosocial health outcomes, we noted substantial improvements in both groups during the first year of treatment in terms of depression, and perceived mental and physical health quality of life, though improvement was greater for individuals who received community-based accompaniment. After 12 months of clinic-based ART, the average prevalence of depression fell by an additional 44.3 %, and perceived mental and physical health quality of life improved by more than twice as much among patients in the accompaniment arm versus clinic-based care arm (p < 0.0001 for all estimates). Social support followed a different pattern. In the group receiving clinic-based care alone, perceived social support started low, decreased in the first 6 months of treatment (p < 0.0001), and returned to baseline levels by month 12 (p = 0.2845). In contrast, among patients who received accompaniment, perceived social support started higher, and was substantially higher after month 6 (5.0 point gain, p < 0.0001) and month 12 (9.7 point gain, p < 0.0001) relative to baseline resulting in moderate-to-high perceived social support throughout the study. At the end of 1 year of treatment, the accompaniment group demonstrated higher perceived social support (p < 0.0001) and mental health quality of life (p < 0.0001), a lower prevalence of depression (p = 0.0257), and a similar perception of physical health quality of life relative to those who received clinic-based care alone (p = 0.0530).
Table 3

Multivariable models of community-based accompaniment and psychosocial health outcomes at 6 months, adjusting for baseline psychosocial outcome values

 

Score at 6 months

Change 0 to 6 months

Difference-in-differences*

Estimate (95 % CI)

t test (mean)/Chi square test (%)

p value (compare groups)

Estimate

t test

p value (compare time)

Estimate

t test

p value (compare groups and time)

Mean social support score (N = 603)a

 Clinic-based care

11.5 (10.1, 12.9)

13.28

<0.0001

−3.5

−8.94

<0.0001

8.5

−15.26

<0.0001

 Community-based accompaniment

23.4 (22.1, 24.7)

5.0

12.47

<0.0001

Mean mental health QOL score (N = 588)b

 Clinic-based care

77.1 (74.0, 80.3)

7.55

<0.0001

18.1

17.98

<0.0001

23.4

−13.54

<0.0001

 Community-based accompaniment

90.0 (87.0, 93.1)

41.5

29.57

<0.0001

Mean physical health QOL score (N = 586)c

 Clinic-based care

94.3 (92.1, 96.6)

−0.24

0.8086

9.9

7.35

<0.0001

25.9

−13.03

<0.0001

 Community-based accompaniment

94.0 (91.9, 96.1)

35.8

24.62

<0.0001

Depression prevalence (N = 610)d

 Clinic-based care

9.1 % (4.7 %, 17.8 %)

1.20

0.2727

−13.7 %

−6.33

<0.0001

40.3 %

12.26

<0.0001

 Community-based accompaniment

6.9 % (3.8 %, 12.5 %)

−54.0 %

−21.88

<0.0001

* Difference-in-difference compares 0 to 6 month change between the treatment groups

aCovariates travel time to clinic, food insecurity, baseline mental health score, baseline physical health score, baseline depression status

bCovariates gender, travel time to clinic, food insecurity, antiretroviral regime, baseline social support score, baseline physical health score, baseline depression status

cCovariates age, baseline body mass index, travel time to clinic, WHO HIV stage, baseline mental health score, baseline depression status

dCovariates gender, food insecurity, baseline social support score, baseline mental health score, baseline physical health score

Table 4

Multivariable models of community-based accompaniment and psychosocial health outcomes at 12 months, adjusting for baseline psychosocial outcome values

 

Score at 12 months

Change 0 to 12 months

Difference-in-differences

Estimate (95 % CI)

t test (mean)/Chi square test (%)

p value (compare groups)

Estimate

t test

p value (compare time)

Estimate

t test

p value (compare groups and time)

Mean social support score (N = 603)a

 Clinic-based care

14.6 (13.2, 16.0)

14.74

<0.0001

−0.4

−1.07

0.2845

10.1

−18.13

<0.0001

 Community-based accompaniment

28.1 (26.8, 29.4)

9.7

24.20

<0.0001

Mean mental health QOL score (N = 588)b

 Clinic-based care

79.0 (75.8, 82.1)

11.28

<0.0001

19.9

19.78

<0.0001

30.2

−17.47

<0.0001

 Community-based accompaniment

98.6 (95.5, 101.7)

50.1

35.68

<0.0001

Mean physical health QOL score (N = 586)c

 Clinic-based care

96.1 (93.9, 98.4)

1.94

0.0530

11.7

8.66

<0.0001

28.9

−14.53

<0.0001

 Community-based accompaniment

98.8 (96.6, 100.9)

40.6

27.88

<0.0001

Depression prevalence (N = 610)d

 Clinic-based care

5.1 % (2.4 %, 10.8 %)

4.97

0.0257

−19.8 %

−8.49

<0.0001

44.3 %

12.49

<0.0001

 Community-based accompaniment

1.9 % (0.8 %, 4.3 %)

−64.1 %

−24.03

<0.0001

aCovariates travel time to clinic, food insecurity, baseline mental health score, baseline physical health score, baseline depression status

bCovariates gender, travel time to clinic, food insecurity, antiretroviral regime, baseline social support score, baseline physical health score, baseline depression status

cCovariates age, baseline body mass index, travel time to clinic, WHO HIV stage, baseline mental health score, baseline depression status

dCovariates gender, food insecurity, baseline social support score, baseline mental health score, baseline physical health score

Discussion

These results suggest that community-based accompaniment may be an effective intervention for improving psychosocial outcomes. Although the Rwandan national model of HIV care reduced prevalence of depression and improved patients’ perceived mental and physical health quality of life, the addition of community-based accompaniment had much greater impacts on all psychosocial health outcomes than clinical care alone. The link between mental and physical health appear complex; improved mental health outcomes may be a mediator and/or result of the improved clinical outcomes among adults receiving community-based accompaniment in this cohort [29].

Worldwide, exceedingly high rates of depression and other mental health morbidity has been observed in HIV-infected populations [44, 45]. Community-based accompaniment appears to have closed gaps in perceived physical health and mental health, including depression, in this population. For these indicators, patients in community-based accompaniment started with lower scores at baseline and experienced greater improvements in the outcomes after the first year of treatment, though both treatment groups reported good psychosocial health after the first year of treatment. Community-based accompaniment may have mitigated some of the complex and compounding interactions of poor mental health and HIV including the emotional toll of an HIV diagnosis, ART side effects, neuropsychiatric complications caused by the virus itself, or disease-related stigma [46], as well as environmental circumstances such as physical violence, instability, or deprivation and extreme poverty that put one at higher-risk for HIV [47]. Regardless of the cause, the challenge of HIV and mental health co-morbidity is increasingly recognized in Rwanda and beyond, and experts are calling for the integration of effective, comprehensive mental health and social support interventions within the context of HIV services [44].

Scale up of integrated mental health and HIV services may be easier in Rwanda than in other countries, though a stepped approach may still be necessary. Mental health services in Rwanda are a national health priority and available in all district hospitals, largely as a consequence of the Genocide. Also, the country has a national system of government-supported community health agents who conduct health promotion activities in villages [48]. Even in a country with such capacity, the recruitment, training, and on-going support of community health workers to provide community-based accompaniment to HIV patients is a large undertaking, and a scale-up strategy might initially target those patients with the greatest needs.

Unmet subsistence needs and lack of a confidant or instrumental support have also been shown to exert important effects on mental and physical health, and are key areas in which community-based accompaniment aims to intervene [49]. Despite starting with a higher level of social support (and theoretically less room for improvement), patients in the community-based accompaniment arm experienced gains in social support throughout the first year of treatment, while patients in the clinic-based care group reported a drop in social support at 6 months and a return to baseline levels of social support after 12 months. A study of community-based accompaniment in Peru showed a similar rise in social support in the intervention group, and drop in social support score in the control group as family members’ emotional and material resources became exhausted over time, and supportive family members withdrew when the patient was no longer critically ill [32].

This study contributes to the limited quantitative literature about the impacts of community-based accompaniment for ART on psychosocial health outcomes, with a larger sample size than most other studies to date [32, 5052]. A key pilot study in Peru found that community-based accompaniment was associated with greater improvements in stigma, social support, self-efficacy, access to health services, retention, adherence, and viral load suppression than clinic-care alone [32], which corroborates the psychosocial results of this cohort presented here and clinic results reported elsewhere [29]. Qualitative studies of comprehensive community health worker-based models in Haiti, Peru, and the US provide critical insights about the mechanisms through which community-based accompaniment impacts adherence and perceptions of wellbeing through emotional support and direct, pragmatic assistance with food, transportation, and other socioeconomic needs [30, 31]. Although the literature is limited, comprehensive community health worker-based models like the one tested here, consistently show improved health outcomes in HIV-infected patients across multiple countries and settings, now including Rwanda.

Due to the observational nature of this study the groups that did and did not receive community-based accompaniment differed in some ways. Numerous data on clinical and socio-demographic factors allowed us to adjust for important confounders, although we cannot completely discard the possibility of confounding due to an important unmeasured difference between these populations.

In conclusion, we found a high rate of depression and low perceived social support and quality of life in this HIV-infected adult population in rural Rwanda. Although these outcomes generally improved with ART in both cohorts, they improved to a greater extent among individuals who received community-based accompaniment. In rural, resource-limited settings with difficult access to health services, community-based accompaniment is an example of an effective intervention that yields improved clinical status as well as mental health and well-being.

Acknowledgments

Financial support for this work was provided by the Doris Duke Charitable Foundation and the Department of Global Health and Social Medicine at Harvard Medical Schools. We thank the dedicated teams of research and clinical staff that supported this project and the participants who participated in this study. We also thank Sidney Atwood for assistance with statistical programming; Bethany Hedt-Gauthier for helpful discussion regarding the data analysis plan; and Daniela Colaci, Carmen Contreras, and volunteers at Translators Without Borders for Spanish language translation.

Copyright information

© Springer Science+Business Media New York 2013