Local beliefs and ART programme factors both influence late ARV starts in sub-Saharan Africa

This article is more than 13 years old. Click here for more recent articles on this topic

Late treatment initiation in HIV clinics in sub-Saharan Africa is being influenced by numerous factors, and there is no single `quick fix` that will improve the situation, researchers from the International Center for AIDS Care and Treatment Programs report in the journal AIDS.

They found that factors relating both to the service design and to the population of the district in which a clinic provided care played important parts in determining the risk of late treatment initiation.

An estimated 44% (5 million) people in need of treatment in sub-Saharan Africa are now on ART, up from approximately 100,000 in 2003.

Glossary

adjusted odds ratio (AOR)

Comparing one group with another, expresses differences in the odds of something happening. An odds ratio above 1 means something is more likely to happen in the group of interest; an odds ratio below 1 means it is less likely to happen. Similar to ‘relative risk’. 

multivariate analysis

An extension of multivariable analysis that is used to model two or more outcomes at the same time.

structural interventions

Programmes which attempt to alter the social, economic, political or environmental factors which drive the HIV epidemic. Examples include programmes to support female education or gender-based violence, legal changes to support harm reduction, and policy changes to reduce stigma and discrimination against key populations.  

sensitivity

When using a diagnostic test, the probability that a person who does have a medical condition will receive the correct test result (i.e. positive). 

confounding

Confounding exists if the true association between one factor (Factor A) and an outcome is obscured because there is a second factor (Factor B) which is associated with both Factor A and the outcome. Confounding is often a problem in observational studies when the characteristics of people in one group differ from the characteristics of people in another group. When confounding factors are known they can be measured and controlled for (see ‘multivariable analysis’), but some confounding factors are likely to be unknown or unmeasured. This can lead to biased results. Confounding is not usually a problem in randomised controlled trials. 

One of the greatest challenges remains getting people into care and onto treatment early. Too many are accessing treatment at an advanced stage of their illness (at very low CD4 cell counts).  In resource-poor settings there is a 3-4 fold higher death rate in the first year on ART compared to outcomes in resource-rich settings. Not only does this severely limit the potential of ART, it also adds to already overburdened health care systems, increases costs and misses potential opportunities for HIV secondary prevention.

The authors looked at quarterly aggregate monitoring and evaluation indicator data from ICAP / Columbia University-supported HIV care clinics in eight sub-Saharan African countries: Ethiopia, Kenya, Lesotho, Mozambique, Nigeria, Rwanda, South Africa and Tanzania. Cohort information from the 267 sites was combined with updated programme level data from site surveys in accordance with the date the cohort started ART and with national contextual-level information from Demographic and Health Surveys (DHS).

The data were analysed using three multivariate models: only programme-level factors, only contextual-level factors and a combination of programme and contextual factors significant in the first two models.

For the purposes of measuring late initiation centres – and chronological cohorts within those centres, in order to capture changes in service delivery   – were classified according to the median CD4 cell count at treatment initiation. A low median cell count was classified as one below 111 cells/mm3. Across all 1690 cohorts starting ART the median CD4 cell count at treatment initiation was 136 cells/mm3.

Multivariate analysis showed that the specific factors associated with late initiation were:

  • No PMTCT affiliation (AOR: 3.6; 95% CI: 1.0-12.8) [indicative of a delayed diagnosis after infection];

  • Lower level of AIDS knowledge [indicative of willingness to engage in care after an HIV diagnosis] and lower uptake of HIV testing in the district;

  •  The availability of outreach services for ART patients only, compared to availability for both pre-ART and ART patients (AOR:2.4; 95% CI: .5-3.9); fewer compared to more adherence support services (AOR: 1.6; 95% CI:1.0-2.5), lower provider to patient ratio (AOR: 2.2; 95% CI: 1.3-4.0)[a delayed ART start once in care]

The high rates of late ART start in the region were programme-specific and independent of the site characteristics.

The study also found that a high local prevalence of the belief that “limiting themselves to one HIV-uninfected sexual partner reduces HIV risk” was associated with a high risk of late treatment initiation (AOR 1.09 per unit increase in belief prevalence, 95% CI 1.06-1.11).

The authors suggest this finding could reflect either a false sense of security among the undiagnosed or “reverse causality”, that is HIV prevention efforts take place in areas where HIV death and illness are the highest.

The authors note their findings highlight the need when designing HIV treatment programmes to further understand the reasons why ART is started late (late diagnosis, late engagement into care, and delayed ART start after confirmation of eligibility).

The risk of starting ART at a low CD4 cell count decreased over time. The authors suggest this may be because of increased HIV testing uptake and ART availability. However, they note this is not necessarily the case.

They cite studies from the United States that found high rates of late diagnosis, delayed enrolment into care and so late ART initiation. An estimated 26% of people in the United States (and New York City) are diagnosed with AIDS at the same time as their HIV diagnosis.

The authors caution that in sub-Saharan Africa “without substantial expansion in HIV testing, increased efforts at engagement in care, close clinical and immunological monitoring for ART eligibility, the challenge of late ART start may well persist.”

Strengths of the study include the use of routinely collected aggregate data used for monitoring and evaluation. Data were gathered from a wide variety of sites including rural and small sites without electronic data systems so increasing the generalisability of their findings.

Sensitivity analyses supported the strength of the findings.

The authors note that because their study was observational rather than randomised there is a possibility of confounding.

Data quality in the service delivery context is often limited and may have influenced the results.

The analysis included rural sites, however only 15% of patients were from rural sites so limiting the generalisation of the findings.

The authors conclude “structural interventions targeting points from the start of ART along the continuum from infection to diagnosis to care are needed.”

References

Nash D et al. Program and contextual-level determinants of low median CD4+ cell count in cohorts of persons initiating ART in 8 sub-Saharan African countries. AIDS 25: advance online publication, doi:10.1097/QAD.0b013e32834811b2, 2011.