Zimbabwe observes a reduction in HIV prevalence, but why?

In 2005, a substantial reduction in national HIV prevalence during the previous year was observed in Zimbabwe. This has been heralded by some as evidence that HIV prevention efforts, particularly strategies based upon Abstinence, Be Faithful and use Condoms (ABC), are having an impact.

However, while this may be true in a general way — there does appear to be some evidence of recent behavioural change, such as a decline in number of partners and casual sex; and greater use of condoms with casual partners —other factors, such as an extremely high rate of mortality, contributed greatly to Zimbabwe’s reduction in prevalence.

Zimbabwe is being held up as an example of the early success of PEPFAR's ABC approach, but do findings from the field really support that claim?

The Ministry of Health and Child Welfare investigates

A report on what happened in Zimbabwe was presented at the PEPFAR Implementers meeting last month in Durban, South Africa, by Dr. Owen Murungi from Zimbabwe’s Ministry of Health and Child Welfare (MoHCW) during a session on ABC.

According to Dr. Murungi, after the dramatic decline in HIV prevalence during 2004 was registered in Zimbabwe, nearly everyone was shocked. “The big question to all of us was, is this real? What’s happened?” he said.

So a review was conducted to determine whether other available data corroborated the finding, and whether the cause for the decline was due to high mortality rates or an actual decrease in incidence. Then, if there was a decline in incidence, could it be explained by natural dynamics of the epidemic or by behaviour changes.

The Ministry pulled together data from 30 different sources including every available survey, records from antenatal clinics, PMTCT and VCT programmes, census data, and clinical trials such as the ZVITAMBO (a large vitamin study) and the Manicaland studies (from rural Zimbabwe). Then working with experts from the UNAIDS reference group, they held a meeting to review the data with all the local stakeholders from government, research and donor groups (including the US Centers for Disease Control, USAID, the UK Department for International Development, UNFPA, UNICEF, London’s Imperial College and the teams from the major local clinical trials).

All the data seem to agree that the fall in HIV prevalence was indeed real. According to antenatal clinic (ANC) data, in the year 2000, the HIV prevalence in Zimbabwe was 32.1% and in the following two years it hovered around 30%. No data were available for 2003 but in 2004 the HIV prevalence had fallen to 23.8% and the test for the trend was statistically significant (p

Rural populations tend to have a lower general HIV prevalence, but the downward trend was similar from the two time periods of 1998-2000 and 2001-2003 in the Manicaland studies. In women between the ages of 15-44 years old, ANC data suggested that the HIV prevalence fell from 21.1% to 19.2% (p value not significant) with declines in most age groups except for women over 30. Another survey from that study looked at women in households, and found that the prevalence was actually much higher. Even so, in the household survey, the prevalence also fell from 25.9% to 22.3% (p=0.015).

The Manicaland study also looked at men between the ages of 17 and 44 years old, where there was also a decrease in prevalence, from 19.5% to 18.2% (p=0.01), with declines in all age groups except men over 35. It is interesting to note that very few of the younger men are infected, but in the years 1998-2000 close to 50% of the men between the ages of 30-34 were HIV-infected, falling to around 40% in the next survey.

Reasons for reduced HIV prevalence in Zimbabwe: decreased incidence or cumulative mortality?

Even though this study is being touted as proof that ABC works, the data that Dr. Murungi presented paint a much more complex picture. A very large part of the reduction in HIV prevalence was actually due to the very high mortality rate for people with HIV in the country.

According to Dr. Murungi, all deaths are recorded in Zimbabwe, and beginning as early as the mid-80’s, there was a steep rise in mortality which Dr. Murungi attributed to AIDS, but which he believes began to level off, although at very high rates, around the year 2000. A closer investigation of the data from Harare suggests that this has indeed been the case in the capital, although the most recent data from Bulawayo shows mortality still rising.

Of course, many people with HIV return to their home villages to die. So looking at the Manicaland study sites (again in rural Zimbabwe), the death rates in men appeared to peak in the year 2000 at around 31-32 deaths per 1000 person years falling to around 26 deaths per 1000 person years in 2002/3 (However, the confidence intervals for these findings are quite similar, so we cannot judge with certainty if this represents a true decrease).

In women, rural death rates peaked in 2001, at just below 25 deaths per 1000 person years, falling to around 23 deaths per 1000 person years, although again, the confidence intervals overlap.

Overall, the yearly mortality rate for people with HIV in Zimbabwe was somewhere over 4% per year. However, by itself, the mortality rate could not effect a reduction in prevalence unless there had also been a reduction in incidence of HIV infection (from the peak incidence rate). In other words, at some point in the last several years, people with HIV began dying at a higher rate than new people were becoming infected.

Over the years, a number of studies have looked at HIV incidence in Zimbabwe. In the first one, Mbizvo et al., in 1993, the incidence was around 5% in antenatal women. Around the year 2000, the ZVITAMBO study observed an incidence that was around 3.6%. Among men, a survey in male factory workers, that the Zimbabwe AIDS prevention survey (ZAPS) conducted in 1994, found the incidence to be about 3.5%. Seven years later (2001), a similar survey in male factory workers reported an incidence of less than 2%.

Although these are cross-study comparisons, the data do suggest a falling incidence — at least between 1993/4 and 2001. If the current incidence is roughly around 2%, at the current mortality rate, the prevalence would decrease substantially each year.

Are changes in incidence due to the natural dynamics of the disease or behaviour change?

Reductions in HIV incidence could be the result of natural dynamics of the HIV epidemic or due to behaviour change (people having less risky sex). For example, over time, any epidemic is somewhat self limiting. Mortality plays more than one part in this, because it doesn’t only decrease prevalence directly, it can decrease incidence as well, by decreasing the pool of infectious individuals who can spread the infection. This is particularly the case when most of those who have died are men (because infected men are much more likely than women to spread the infection to more than one partner).

However, although he didn’t exclude natural dynamics during his presentation, in a conversation afterwards, Dr. Murungi said that colleagues at Imperial College in London had run simulations suggesting that other factors besides natural dynamics were needed to explain the changes in incidence observed in Zimbabwe.

So the MoHCW investigated what the available data sources could tell them about the contribution of behaviour change to the decrease in incidence and prevalence. Several surveys have explored age of first sex among 15-24 year olds over the last 20 years. Dr. Murungi believes they show that the age of sexual debut has increased over the last several years, although the data he presented were difficult to scrutinise closely (a dozen or so studies’ findings scattered both up and down across a graph, which made it difficult to evaluate the trends or determine how comparable the data were).

However, over the last several years, there did appear to be a clear and substantial fall in the percentage of young men who reported having had sex during the last 12 months with non-regular partners. Reported condom use with non-regular partners had also increased in the last five years. Dr. Murungi noted that there has also been a steady increase in the number of condoms in circulation, particularly socially marketed condoms (rather than public sector condoms).

In the Manicaland study, statistically significant changes in reported sexual behaviour were observed for both males and females in 1) the age of sexual debut, 2) new partners in the last year/month and 3) the number of current partners. The data supporting positive behaviour changes in these rural settings are convincing but it should be kept in mind that most new infections occur when people leave the rural settings to find work away from their families in urban or industrial areas. Whether they take these behaviour changes with them to other settings is another matter.

Problems with attribution and intervening variables

What was missing from Dr. Murungi’s presentation was the time to discuss the sort of prevention messages that were being spread most in Zimbabwe in the late 90’s and early 2000’s — and who was doing it. This was not a prospective study so the association with programmes in operation there today is tenuous at best.

What is also interesting is that most of the data suggesting that there had been a change in behaviour come from the last five or six years, while the most recent data suggesting that there has been a decrease in incidence came from the years 2000 and 2001 suggesting that the incidence had in fact been falling over the course of the late 1990’s (while mortality was rising).

It would be useful for someone to compare and contrast what has happened in Zimbabwe with what is going on in Botswana, where despite massive efforts and funding spent on ABC-based prevention messages, the HIV prevalence in Botswana remains extremely high (38.5%) according to the UN report.

Again the effects of such a high mortality rate in Zimbabwe need to be considered. Over the course of the PEPFAR meeting, there was much talk about “creating enabling environments” that support and encourage people to abstain or be faithful or use condoms consistently. Well, history has shown that observing large numbers of people sick and dying of HIV can be a powerful motivator for changing behaviour. It is not a for-mula for enabling HIV prevention that any sane person would promote however.

There could also be a host of other negative “enabling” factors that played a part in the reported behaviour change. Its important to remember that this is, after all, Zimbabwe. Since the year 2000, Zimbabwe’s economy has ground to a halt; the country suffered from floods, followed by severe drought and endemic food insecurity. In this context it is odd, to say the least, that the official mortality rate reported peaked before all that trouble began. But even if famine and inflation didn’t increase the numbers of people dying, the calamity would have increased the costs of caring for a person with HIV tremendously.

People with AIDS have a tendency to return for care in their home villages but caring for a person with AIDS is difficult enough for a rural family in a time of plenty —during a time of hardship, the family can be strained to the breaking point. This usually increases stigma from which people with HIV suffer. In another study at the Implementers meeting, a team from Population Services International reported some of the challenges trying to work with and empower people living with HIV and AIDS (PLWHA) in Zimbabwe, who they reported experienced high levels of shame, blame and enacted stigma during this period. According to the presentation, in 2005, terms used to describe people with HIV in Zimbabwe included phrases such as “in the departure lounge,” “crossed the red robot,” and “bewitched by goblins.”

No wonder young people in Manicaland don’t want to go that route.

It can also be challenging to organise prevention work in Zimbabwe in the current political climate. For example, Catholic Relief Services has been working with orphan girls in Zimbabwe, who because of the disruption of normal patterns for their domestic and sexual education (loss of aunties to illness, etc.) are poorly informed about reproductive health and general protection issues. However, “local authorities can be politically sensitive to gatherings of youth,” a poster at the Implementer’s meeting reported, and any educational efforts they put together had to “work under the radar.” In fact, many foreign non-governmental organisations have complained about the difficulty of working in Zimbabwe over the last several years — which lends some credence to Dr.Mark Dybul’s (Acting US Global AIDS Coordinator) assertion (in a press conference at the meeting) that the church is often the only organisation with “reach” into some countries.

But the collapsing economy could have additional effects that could decrease mobility (and therefore risk of HIV infection). Unemployment has sky-rocketed in Zimbabwe, and there have been major petrol shortages. The commercial mining sector has collapsed, and factory work has evaporated. Cities no longer offer much work. (It would be interesting to know what has happened to HIV prevalence in the general population in the cities during this period.) So as a result of Zimbabwe’s economic contraction, many of the old hotspots for HIV trans-mission — near the factories and mines, at truck stops along the highway — could be dwindling or people no longer have a reason or the means to go there.

Finally, many of those with the means to get out and look for work have poured into neighbouring countries, including Botswana, and South Africa. Hundreds of thousands of adults in their prime working years (who may represent a substantial proportion of the sexually active and possibly HIV-infected population) have simply left the country. And yet, so far, no one has addressed what impact emigration might have had on Zimbabwe’s HIV prevalence and incidence — and what might happen should they all return home for treatment (see below) which is increasingly available.

These variables need to be explored fully in a multivariate analysis before building any case for prevention strategies upon the basis of what is going on in Zimbabwe. Dr. Murungi says that they still intend to do further analyses on the available data, and that results from a demographic health survey performed last year, should help fill in some of the missing data. Chances are there is still a lot to learn about what has happened in Zimbabwe — but again, it may not yield a formula that anyone would want to mimic.

This is success?

This is not to say that the ABC strategies aren’t working, but teams building effective prevention programmes need better information about best practices and how to create enabling environments that can be replicated in their home countries. While a balanced prevention campaign featuring the ABC approach may indeed reduce the number of new HIV infections, the defenders of PEPFAR’s prevention package may be in too much of a haste to vindicate the approach before their critics.

Adopting Zimbabwe as a poster child for ABC also seems unwarranted especially as the teams that PEPFAR is working with now are carefully gathering data and putting it together in such a way that other groups will be able to learn from their successes. It may take a little longer but it should be far more effective.

Finally, Dr. Murungi stressed that Zimbabwe still has a long way to go “We acknowledge the fact that the prevalence rates are still very high in Zimbabwe. We still have a lot to do. 20% is still very, very high.”

And it might not even be as low as that. Until recently only a very low percentage of Zimbabweans were willing to be tested for HIV. According to another presentation at the Implementer’ meeting, national testing and counselling centres were set up across the country, but as of the first quarter of 2003, HIV prevalence was only 18.3%, which suggested to the authors that they were drawing primarily the “worried well.” Then in April 2004, the Zimbabwean government, with support from the US government began rolling out antiretroviral therapy, which by February 2006 was available at 65 sites nationwide. The rollout has been associated with an increase in the percentage of symptomatic people and people at higher risk of infection who come in for test-ing so that by the fourth quarter of 2005, the HIV prevalence at the national testing sites had increased to 28.9% (p=0.05) — which is much closer to the HIV prevalence reported in earlier this decade.

Finally, more Zimbabweans on ART should mean improved survival — which could increase HIV prevalence. In fact, ART could be one of the major reasons why neighbouring Botswana continues to see a stable rather than a falling HIV prevalence.