Botswana’s adult HIV mortality rate falls since the rollout of free antiretroviral therapy

This article is more than 16 years old. Click here for more recent articles on this topic

ART and mortality

It is well established that ART reduces the risk of mortality in people with severe HIV disease, and ART programmes in developing countries have demonstrated evidence of ART’s therapeutic benefits among treated patients.

“However, the impacts of ART programmes in reducing mortality at the population level have yet to be documented in sub-Saharan Africa and are of strategic importance in assessing the ultimate effectiveness of this intervention,” said Dr Rand Stoneburner of the University of California, San Francisco, who presented the late breaker paper.

Between 2003 and 2005, since the launch of Botswana’s national antiretroviral therapy (ART) programme, there was a decline in the country’s adult mortality rate according to a late breaker report presented on Thursday August 17th at the Sixteenth International AIDS Conference in Toronto. The declining mortality rate was most pronounced in the districts of the country where ART first became available and where ART coverage is the most extensive.

Although these results are preliminary, and there is no way to show direct causality, these findings nonetheless provide tantalising evidence that ART is reducing premature mortality in Botswana — and indeed represent the first evidence at the population level in any African country that the introduction of ART could be extending the life expectancy.



In everyday language, a general movement upwards or downwards (e.g. every year there are more HIV infections). When discussing statistics, a trend often describes an apparent difference between results that is not statistically significant. 

mother-to-child transmission (MTCT)

Transmission of HIV from a mother to her unborn child in the womb or during birth, or to infants via breast milk. Also known as vertical transmission.

advanced HIV

A modern term that is often preferred to 'AIDS'. The World Health Organization criteria for advanced HIV disease is a CD4 cell count below 200 or symptoms of stage 3 or 4 in adults and adolescents. All HIV-positive children younger than five years of age are considered to have advanced HIV disease.

However, the same decline in mortality has not yet been observed among the children of Botswana.

A matter of Botswana’s survival

“Every life is precious,” said Botswana’s Minister of Health Professor Sheila Dinotshe Tlou at a symposium earlier in the week. With a population of only 1.7 million people, and one of the highest HIV prevalences in the world, Botswana could ill afford waiting to treat its infected population. So, beginning in 2002, Botswana was one of the first countries on the continent to provide ART to its population with advanced HIV disease. The programme has been one of the world’s most successful — reaching 85% of those estimated to need treatment.

In order to measure the overall impact of HIV and the ART programme on the population both country-wide and within specific health districts, the Ministry of Health (MoH) and researchers from UCSF looked at the trends over time in mortality by age, sex and place from 1991-2005. Botswana keeps very close track of its mortality statistics (an internal audit found that the Botswana’s Central Statistical Office captures between 95-97% of the adult death’s in Botswana). (For 2005, data on deaths through June and reported by November were annualised).

Trends in reported deaths (from the MoH data) were independently corroborated through an ongoing nightly census of hospital admissions, discharges and deaths. The hospital midnight census data are completely independent from the Botswana mortality surveillance system and were routinely available to the researchers for the period from the 1990s through the end of 2003. However, because of a two year lag in the collation and public release of the midnight census data, the researchers themselves had to rapidly review the data from 2004 to June 2005 (which is why this study’s findings must be considered ‘preliminary’).

The researchers explored several questions in these sets of data. Firstly, do mortality trends in Botswana reasonably reflect HIV dynamics over time? Are there declines in suspected HIV deaths over time? If there are declines in the death rate, do they correlate with when or how the ART rollout occurred? And do changes in the death rate differ with the ART coverage (extent of ART uptake) or the timing or initiation of the ART rollout in different health districts?


As HIV prevalence increased in Botswana in the 1990s and early 2000s, mortality rates among 25-54 year-olds increased four-fold. Among these deaths, the proportion of non-institutional deaths (which were reported from outside of hospitals) actually decreased from 26% to 15% — so more deaths were occurring in the hospitals. Among people 25-54 years-old, there was an eight-fold increase in institutional deaths from 1991/2-2003.

AIDS as a cause of death increased from 4% in 1992 to 27% in 2003. AIDS became the leading cause of institutional deaths in 1996 in Botswana.

But by 2003, the country was beginning to put significant numbers of people with HIV on treatment.

Botswana currently has over 62,000 people living with HIV on treatment — and ART works as well in Botswana as anywhere else. According to Dr Stoneburner, at twelve months in the programme, the median CD4 cell count increase is 166 cells/mm3, patient follow-up is 90%, the adherence rate is 85%, 86% of the subjects have viral loads below 400 copies/ml, 8% deaths on treatment, and the survival rate on treatment is 83%.

The programme appears to be decreasing the adult mortality rate at the population level as well. Trends in institutional deaths crested in 2000 and then decreased 8% in 2003-2004 and 20% in 2004-2005, coincident with increasing numbers of patients on ART. The declines were greatest in those aged 25-54 years old and remained fairly stable in younger and older populations “providing further evidence that this reduction in the trend is related to a reduction in HIV-related deaths,” said Dr Stoneburner.

There is also a geographic association between the reduction in mortality and the parts of the country where the rollout first occurred (Gaborone, Francistown, Serowe and Maun) and where coverage is now the greatest. For example, mortality declined 27% and 17%, respectively, in the Gaborone and Francistown districts. Again, district level mortality declines were greatest among 20-54 year-olds and significantly correlated with both the initiation date and coverage rate (p<.05 art="" district="" of="" programs.="">

These declines in mortality (and within this age group) were corroborated by the midnight mortality data from hospitals in Gaborone and Francistown.

Cautions and concerns

“Our preliminary analyses provide evidence of an early association between ART uptake and declines in mortality in Botswana [but] must be interpreted with caution certainly until further verified by updated and validated 2005 and 2006 mortality statistics” said Dr Stoneburner. However, he concluded that “alternative hypotheses to explain these findings including natural dynamics, effects of other interventions, surveillance artefacts or other biases seem to be less plausible.”

However, Dr Stoneburner made one sobering observation — the lack of a change in mortality in young children. Since the HIV epidemic, the infant mortality rate has grown steadily worse. But one would expect the trends to be improving in this population as well, especially since Botswana has also successful introduced prevention of mother-to-child transmission (PMTCT) to over 85% of the population — and the current HIV transmission rate reported for this programme is 6% or less.

“The absence of similar declines in children is disturbing and needs to be explained urgently,” said Dr Stoneburner.


Stoneburner R et al. Declines in adult HIV mortality in Botswana, 2003-2005: evidence for an impact of antiretroviral therapy programs. Sixteenth International AIDS Conference, Toronto, abstract THLB0507, 2006.