AIDS, conflict and the media in Africa: risks in reporting bad data badly

"The fact that many Sudanese will return to their homes from countries where the HIV/AIDS prevalence is very high, will doubtless increase the likelihood of a further spread of the epidemic" [10].

Conflict in Sudan has displaced millions of people and forced hundreds of thousands to flee across borders. In the south of the country, the 21-year old war between the Government of Sudan and the Sudan People Liberation Movement/Army has caused the internal displacement of approximately 4 million people, to which the more than 1.5 million displaced as a consequence of the more recent conflict in Darfur must be added. As of the end of 2003, more than 500,000 Sudanese refugees were residing mainly in Uganda, Ethiopia, Kenya and Chad [11].

By the end of 2003, the overall HIV prevalence among adults in Sudan was estimated at 2.3% (95% confidence interval (CI) 0.7 to 7.2) [12]. In 2002, the Sudanese government undertook an HIV survey in 14 of 26 states with the following results: pregnant women at antenatal clinics (1.0%), tuberculosis (TB) patients (1.6%), sexually transmitted infections (STI) patients (1.1%), sex workers (4.4%), tea sellers (2.5%), IDPs (1.0%) and Eritrean refugees (4.0%) [13].

Following progress in the negotiations between the Government of Sudan and the Sudan People Liberation Movement/Army, future peace in South Sudan and the consequent return home of refugees and IDPs was reported widely in the media. Particular attention was expressed regarding high infection rates among Sudanese IDPs and refugees, and the fear that refugees living in host countries with high prevalence rates will return and exacerbate the epidemic in South Sudan [14, 15]. The situation was described as "severe", "on the verge of an HIV/AIDS epidemic" and a "disaster" that was going to "doubtless increase the likelihood of a further spread of the epidemic" [10, 15, 16].

Headline: "HIV/AIDS soars in war-torn northern Uganda" [17].

Northern Uganda is currently devastated by a 19-year old conflict between the Ugandan Government and the Lord's Resistance Army (LRA), which escalated in 1996 causing massive population displacement, further aggravated since the increase of violence in mid-2002. At the end of September 2004, there were 1.4 million IDPs in some 200 camps, living in extremely difficult hygienic and environmental conditions and depending on relief food distribution [18].

Uganda experienced declines in HIV prevalence during the 1990s. National prevalence estimated by antenatal sentinel surveillance peaked in the early 1990s and has fallen to 4.1% (95% CI 2.8 to 6.6) by the end of 2003 [12]. Sentinel surveillance shows that median HIV prevalence in the capital Kampala declined from a high of 29.4% in 1992 down to 8.3% in 2002, while in areas outside major urban centres it decreased from 12.0% to 4.6% [19].

Several reports in the national and international press and information disseminated by international organizations have emphasized that the HIV prevalence in northern Uganda (11.9% in 2002) is nearly twice as high as the national rate (6.2% in 2002) and growing due to conflict [20]. These reports particularly highlighted the high HIV prevalence among IDPs in the camps and the highest prevalence among victims of the LRA [21‐26]. Children, particularly girls, captured by the LRA have been reported to have an HIV prevalence of approximately 50% [17, 23, 27‐29]. In the past, this vulnerable group had already been suggested to be infected with extremely high rates of STIs, including HIV [30]. Furthermore, direct reference to "immorality" among IDPs as explanation for the higher rates of HIV prevalence was attributed to unspecified "NGO (non-governmental organisation) workers" [31].

"It's the refugees' fault. AIDS comes with them!" [32].

Since 1990, the republic of Guinea with a population of 7.5 million persons has hosted hundreds of thousands of refugees from primarily Sierra Leone and Liberia. Until late 2000, the refugees lived in small villages scattered throughout the rural south-eastern Forest region. During September to December 2000, attacks by armed factions in Guinea led to the displacement and consequent transfer of the refugees to safer camps in the north-west of the country [32]. By the end of 2004, there were still approximately 150,000 refugees primarily from Liberia, Sierra Leone and Cote d'Ivoire [2]. Due to recent conflict in Cote d'Ivoire, which has the highest HIV prevalence in the West Africa (7.0%) [12] over 100,000 Guineans have returned home, many to the Forest Region [32].

A national HIV seroprevalence survey, undertaken in Guinea in December 2001, showed an overall HIV prevalence of 2.8% among pregnant women with 4.4% in urban areas and 2.2% in rural areas [33, 34]. An HIV prevalence of 7.0% was reported in two towns in the Forest Region, the area where the refugees were located. By the end of 2003, HIV prevalence was estimated at 3.2% [12].

Before the national survey, it was assumed that the overall HIV prevalence in Guinea was lower than 2.8%. The media reported that the higher HIV prevalence in the Forest Region was likely due to the influx of refugees, blaming their "loose morals" that resulted in "flourishing promiscuity and prostitution" [32].

"The fact that many Sudanese will return to their homes from countries where the HIV/AIDS prevalence is very high will doubtless increase the likelihood of a further spread of the epidemic" [10].

"Infection rates are particularly high among vulnerable groups, such as internally displaced persons (IDPs) and refugees" [10].

Concern regarding stigmatizing statements against refugees living in Sudan and Sudanese refugees living in surrounding host countries began after the Integrated Regional Information Network (IRIN) [10] reported on a UN newsletter [16]. Although refugees often come from countries of origin with lower HIV prevalence and move to live among host country populations with higher HIV prevalence, each case is context specific and must be examined independently [36]. The degree of interaction among refugees and host communities as well as their HIV knowledge and behaviour all contribute to the HIV prevalence in both populations. Sudanese refugees living in Kenya and Uganda still have much lower HIV prevalence than their surrounding host communities despite living among them for over a decade [36]. Furthermore, Sudanese refugees living in Kenya have shown better knowledge and less risky behaviour than some Sudanese living in South Sudan [37]. Blanket statements such as the first one in the text box above are unhelpful and can be damaging. These and other statements in the UN report and IRIN article were published without checking with other sources, such as the United Nations High Commissioner for Refugees, NGOs working with the refugees, or the refugees themselves. After discussion with IRIN, the news organization acted in a rapid and professional manner. They removed the original article from their website, interviewed different experts in the field to provide a more balanced view, and changed the first statement listed above to "The fact that many Sudanese will return to their homes from countries where the HIV/AIDS rate are high might increase the likelihood of a further spread of the epidemic" [15].

Headline: "GUINEA: Refugee influx adds fuel to AIDS crisis in southeast Guinea" [32].

Another IRIN report stated that refugees are being blamed for spreading HIV among the local Guinean population. Statements from local Guineans and even refugees provide credence to the assertion that refugees spread HIV. However, the HIV prevalence has never been measured among refugees in Guinea and there is insufficient evidence to know the effect that the mixing of refugees and locals have had on the HIV prevalence in the region. Only near the end of the article does it state that condom use and knowledge of HIV/AIDS were much lower among Guinean youth than Liberian refugees.

"The HIV-infection rate in Sudan, according to UNFPA, is already considered an epidemic, making Sudan the country with the highest infection rate in North Africa and the Middle East region" [10].

Although Sudan has the highest HIV prevalence in North Africa and the Middle East, it has the lowest prevalence compared with the countries surrounding it to the East, West and South [12]. The former statement was made to buttress the point that Sudan has a serious HIV problem. Indeed, HIV is a very serious issue in Sudan; however, the fact remains that HIV prevalence is much higher in surrounding countries than Sudan itself.

"The rate of HIV/AIDS infection in northern Uganda is nearly double that in the rest of the country..." [29].

HIV prevalence in the Northern Uganda, represented by the surveillance site located in Lacor hospital in Gulu, is often correctly reported as nearly double the national average. This point estimate is then used to 'prove' that the conflict in Northern Uganda has caused the HIV epidemic to flourish. The estimated HIV prevalence for Gulu is however not the only one above the national average and is in fact similar to other urban sentinel sites in areas not affected by conflict (Table 1). Moreover, trends over time actually show that the HIV prevalence in Northern Uganda has decreased over time, similar to many other sentinel surveillance sites in the rest of the country. The decrease in the HIV prevalence in Gulu has been significant from 27% in 1993 to 12.8% in 1998 to 11.9% in 2002 (p < 0.001) [19, 38]. The trends in Table 1 clearly show that, from 1993 to 2002, the reduction of HIV prevalence in the sentinel site for Northern Uganda has been larger than that of many other sentinel sites. Despite an increase in the intensity of the conflict from 1996 onwards, trends from 1998 to 2002 show a larger decrease in Gulu than in peaceful southern and eastern Uganda, represented by the sentinel site of Mbarara and Mbale regional hospitals, respectively.

Therefore, basic epidemiological principles must be used when interpreting HIV prevalence data. Trends over time must be examined as opposed to comparing point prevalence estimates as occurred above.

"Data from the mid-1990s onwards suggested the infection rates had risen rapidly in the conflict-affected areas of southern and western Sudan. According to a national prevalence and behaviour survey conducted in 2002, HIV prevalence was already four percent among pregnant women attending clinics in the refugee camps" [10].

There are no data at present suggesting that the HIV prevalence in South and West Sudan is rising rapidly. No source is provided for the statement in the above text box. In South Sudan, IRIN reported that the Sudan Evangelical Mission found HIV prevalence to be 4–10% in Rumbeck and 17–21% in Yambio in their voluntary counselling and testing (VCT) centres [10]. However, VCT data present biases and cannot be used as proxy for HIV prevalence in the general population due to the biases involved in self selection of clients and referrals from medical personnel. As a comparison, HIV prevalence rates from VCT services in both Kitgum and Pader districts in neighbouring north Uganda, are estimated at 20% (AVSI-supported PMTCT sites program data. unpublished), whereas PMTCT results show the HIV prevalence to be 7.9%–9.9% in Kitgum and 4.6% in Pader [39].

The second sentence in the above quotation discusses HIV prevalence among Eritrean refugees living in the Eastern part of the country from the Sudanese National AIDS Control Programme's 2002 report. The comparison, however, should not be among Sudanese pregnant women attending antenatal care (ANC) from all over the country (HIV prevalence of 1.0%) with refugee pregnant women in East Sudan (4.1%) but rather between Sudanese pregnant women from the sentinel site nearest to the refugee camps (El Gedarif). The data provided by the 2002 report does not allow disaggregation by state. However, in 1998, Sudan reported the HIV prevalence in the ANC sentinel surveillance in El Gedarif as 4.0% [37, 40], almost identical to the 4.1% reported among the refugees. There is limited HIV data from Eritrea; one sentinel site outside an urban area in 1994 showed 3% prevalence while one major urban area site in 1999 showed 4%; both are similar to the sites found in El Gedarif in Sudan and among the refugees [41]. Finally, the same IRIN article also claims that HIV infection rates are particularly high among vulnerable groups, such as IDPs and refugees. However, this is also incorrect; the HIV prevalence among IDPs was estimated to be 1.0% [13, 36].

"About half the girls who escape from the rebels are found to be HIV positive, doctors say" [25].

Claims of extremely high HIV prevalence among children abducted by the LRA are not supported by data; no systematic HIV testing has ever been undertaken among this group. The statement reported above, originally made by a Government official [27] and then reported by numerous NGOs and news agencies, was subsequently dismissed by organizations responsible for the major reception centres for these children in the region (personal communications and email exchanges between members of the Psychosocial Working Group and Child Care Centres in North Uganda, August 2004).

[AIDS is the] "leading cause for death, constituting 69% of deaths in Gulu area, or three times higher than direct killings during military confrontation" [26].

Misleading or false information regarding HIV and AIDS in Northern Uganda has been parroted repeatedly by news agencies, apparently without verification. Figure 1 attempts to trace the origin and the subsequent reporting between July and November 2004 of three main statements: 1) HIV prevalence in Northern Uganda is double that of the whole country; 2) HIV prevalence is 50% among those children abducted and subsequently released by the LRA; and 3) AIDS is the cause of 69% of the total deaths in Northern Uganda.

The first two issues were discussed previously and have been shown to be incorrect. The third issue relates to a recent report on the humanitarian situation of Northern Uganda which claimed AIDS to be the leading cause of death (69%) in the Gulu area, three times higher than direct killings during military confrontation [26]. The source for this statement was not provided, and coverage of HIV testing is limited to few areas within Gulu District. Conversely, Médecins Sans Frontières undertook at the end of 2004 a population-based survey and reported alarmingly high rates of mortality in the insurgent-affected districts of Lira and Pader in Northern Uganda (crude mortality rate of 2.8/10,000/day and under five mortality rate of 5.4/10,000/day) [42]. As in the Democratic Republic of Congo [43], the main causes of death were indigenous communicable diseases that have been exacerbated due to the conflict, such as malaria (47% of all deaths), respiratory disease (28%) and diarrhoeal disease (21%); however, AIDS could be an underlying factor that was not reported.

To further support this point, several months after the initial and incorrect statements were reported, all three re-emerged in the media [44] and, more recently, IDPs in North Uganda have been reported as having an average HIV/AIDS prevalence six times higher than the national average [45].