Table 1 Summary of studies assessing the impact of community support on ART programme delivery and outcomes in resource-limited settings

Abaasa

Uganda

Retrospective cohort study

Field officers

The AIDS Support Organization (TASO) ART programme displays good adherence and survival

- Retrospective design: not all potential confounders included

(2008)

- Self-reported adherence and the risk of social desirability bias

Assefa

Ethiopia

Descriptive study

Health extension workers

Substantial expansion of HIV/AIDS and ART services in resource-limited context

- Descriptive study design: no causal relationships

(2009)

- No measurement of the extent of the impact of community support

- Secondary and incomplete data

Bedelu

South Africa

Descriptive study

HIV/AIDS counsellors & CHWs

MSF programme using task-shifting and community support achieved near universal coverage without compromising quality of care

- Descriptive study design: no causal relationships

(2007)

- No measurement of the extent of the impact of community support

Benavides

Uganda

Descriptive report

Field officers

Field officers encourage adherence, refill medications and promote family support contributing to TASO’s ART program’s outcomes (adherence rates > 95% , reducing mortality by almost 90% )

- Descriptive study design: no causal relationships

(2006)

- No measurement of the extent of the impact of community support

Celletti

Brazil, Ethiopia, Malawi, Namibia, and Uganda

Desk review, observation & key informant interviews

CHWs

Under certain conditions, the delegation of specific tasks to CHWs can increase access to HIV services and improve quality of care.

- No clear literature search strategy

(2010)

- No clear qualitative methodology (selection of informants, data collection & analysis)

Chang

Uganda

Retrospective cohort study

Peer health workers

Good adherence and survival in community- and faith-based HIV/AIDS care programme

- Retrospective study design

(2009)

- Reliance on clinical and programmatic records

- Underestimation of survival because of lost-to-follow-up rates

- Outcome measurements not measured at exact time intervals

Chang

Uganda

Cluster-randomized trial

Peer health workers

A peer health worker intervention was associated with decreased virologic failure, but did not affect cumulative risk of virologic failure, adherence measures or short-term virologic outcomes

- Limited generalisability: mobile clinic setting

(2010)

- Weakness of design: imbalances between clusters

- Limited statistical power

Cohen

Lesotho

Descriptive study

HIV/AIDS counsellors

Lay counsellor-supported testing and counselling, adherence and case management produced favourable outcomes

- Descriptive study design: no causal relationships

(2009)

- No measurement of the extent of the impact of community support

Etienne

Kenya, Rwanda, Uganda, Tanzania, Zambia, Nigeria, Haiti, and Guyana

Descriptive study

Adherence supporters

Adherence counselling, structured treatment preparation, community home visits, and supportive supervision by community nurse significantly reduced the loss to follow up.

- Descriptive study design: no causal relationships

(2010)

- Potential selection bias at the facility level

Gusdal

Ethiopia & Uganda

Qualitative study

(peer) HIV/AIDS counsellors

Peer counsellors served as facilitators of adherence, role models and bridges to the health system

- Selection bias: no information on patients lost-to-follow-up

(2011)

- Saturation of data not achieved

Hermann

Ethiopia, Malawi, and Uganda

Desk review & descriptive field research

CHWs

Present CHW programmes are essential for ART scale-up and comprehensive care but have insufficient attention to quality supervision, continuous training,and the life experience of PLWHA

- No clear literature search strategy

(2009)

- No clear methodology for the literature analysis

Idoko

Nigeria

Quasi-experiment

DOT ART supporters

Patients accessing treatment support (daily/ twice weekly/weekly observed therapy) demonstrated better treatment outcomes compared to control group

- Limited genralisability: one facility

(2007)

- Small sample size: limited statistical power

- No statistically significant differences

Igumbor

South Africa

Retrospective patient record review

adherence supporters

Patients with community adherence support maintained a suppressed VL and remainedin care for a longer period as opposed to patients lacking this support

- Retrospective study design (no causal relationships)

(2011)

- No measurement of the extent of the impact of community support

Jaffar

Uganda

Cluster-randomised equivalence trial

Field officers

Home-based HIV care was as effective as facility-based care: similar virological failure and mortality rates

- Refusals and withdrawals can create selection bias

(2009)

- Weakness of design: imbalances between clusters

Kabore

Lesotho, South Africa, Namibia, and Botswana

Observational cohort study

CHWs, HBC volunteers & adherence supporters

Community support was associated with more rapid and greater CD4 increase and higher levels of adherence. Home-based care and/or food support was associated with greater improvements in HRQoL.

- Observational study design: patients select to receive support

(2010)

- Potential selection bias

- High rate of patients who were lost-to-follow-up

Koenig

Haiti

Descriptive study

CHWs

DOT-HAART using CHWs resulted in good viral suppression and high survival rates

- Descriptive study design: no causal relationships

(2004)

- No measurement of the extent of the impact of community support

Kunutsor

Uganda

Randomized controlled trial

Adherence supporter

Patients with an adherence supporter had over 4 times the odds of achieving optimal adherence and were more l ikely to be on time for their clinical appointments

- Limited genralisability: one facility

(2011)

- Potential selection bias (of highly motivated patients)

- Limited statistical power (adherence measurement method)

Morris

Zambia

Descriptive study

Peer health workers

Improved clinical care quality despite growing patient volumes

- Descriptive study design: no causal relationships

(2009)

- No measurement of the extent of the impact of community support

- Programme’s intensive use of resources

Mukherjee

Haiti

Descriptive study

CHWs

Home-based adherence support from a network of CHWs produces low rates of treatment failure

- Descriptive study design: no causal relationships

(2006)

- No measurement of the extent of the impact of community support

Mukherjee

Haiti

Descriptive study

CHWs

CHWs facilitate the uptake of PHC services, including by the most vulnerable households

- Descriptive study design: no causal relationships

(2007)

- No measurement of the extent of the impact of community support

- Limited representativeness of qualitative results

Muñoz

Peru

Quasi-experiment

CHWs & DOT ART supporters

CASA (community-based accompaniment with supervised antiretroviral) participants reported better clinical and psychosocial outcomes compared to control group

- Small sample size

(2010)

- Potential selection bias (confounding differences)

- Limited generalisability

Nachega

South Africa

Randomized controlled trial

DOT ART supporter

DOT-ART was associated with greater median CD4-cell count and better survival rates, but not with improved virological outcomes

- Limited genralisability

(2010)

- Limited time frame of the intervention

- Relatively low incidence of AIDS-defining illness and death

Pearson

Mozambique

Randomized controlled trial

Peer DOT ART supporters

Intervention participants demonstrated significantly higher ART adherence

- Initial phase of ART programme (highly motivated patients)

(2007)

- No blinding of the participants and the study personnel

- Self-reported adherence measure

- Limited generalisability: one facility

Rich et al.

Rwanda

Retrospective medical record review

CHWs

Community based ART produced very high levels of retention and large increases in CD4 cell count. However, the relative impact of the different components of the program could not be determined.

- Descriptive study design: no causal relationships

(2012)

- No measurement of the extent of the impact of community support

- Low data completeness for key variables

- Potential selection bias

Selke

Kenya

Randomized controlled trial

CCCs

Community-based care by PLWAs resulted in similar clinical outcomes as standard care but with half the number of clinical visits

- Limited generalisability

(2010)

- Selection bias: different sublocations & only adherent patients included

- Small sample size (limited statistical power)

Weidle

Uganda

Nested randomised trial

HIV/AIDS counsellors & Field officers

Group education, personal adherence plans, a medicine companion and home-delivery of ARVs by lay counsellors achieved good ART adherence and reponse

- Selection bias: participants selected from a community AIDS organisation

(2006)

- Limited statistical power

- No measurement of the extent of the impact of community support

Wools-Kaloustian

Kenya

Quasi experiment

CCCs

An ART delivery model that shifts patient monitoring and ARV dispensing to CCCs is both acceptable and feasible

- Limited generalisability: one facility

- Programme’s intensive use of resources (PDAs & training)

(2009)

Wouters

South Africa

Retrospective cohort study

CHWs & adherence supporters

Community support predicted better viral suppression and immunological restoration

- Study design: patients select to receive support

(2009)

- Potential selection bias (confounding differences)

- Underestimation of survival because of lost-to-follow-up rates

Wouters

South Africa

Retrospective cohort study

CHWs & adherence supporters

Community support initiatives (CHWs and support groups) positively impacted disclosure to family members

- Study design: patients select to receive support

(2009)

- Potential selection bias (confounding differences)

Zachariah

Malawi

Quasi-experiment

HBC volunteers.

Community support was associated with significantly lower death rate and better ART outcomes

- Study design: not possible to know the exact reasons for the observed differences

(2007)