Our study found that published data do not suggest that women are at a disadvantage when compared to men when accessing HAART in southern Africa. The predominance of women in HAART programs to an extent reflects the gender proportions of the adult population infected with HIV in southern Africa. For instance, the Zambia Demographic and Health Survey 2000–01 [6
] national HIV prevalence for persons aged 15 yrs to 49 years was for 17.8% for females and 12.9% for males. As a proportion of all adults accessing HAART, women are in the majority outnumbering men.
Several authors have reported that women access HIV care earlier than men in the United States [36
]. Others authors though have reported opposite findings [39
]. The improvements in access of HIV testing among women attending antenatal care (ANC) has been reported as an important entry point for these women into HIV care and the support pipeline [40
]. In southern Africa, there is also a drive to encourage women attending ANC to be tested for HIV in an effort to reduce vertical HIV transmission. Such drives could contribute to the higher access of women to HIV care services. The extent however to which HIV antenatal services have contributed to overall HAART programs need to be evaluated.
In some HIV testing programs outside of antenatal care, there has been a predominance of females over men in many settings [41
]. In 2004 in Malawi, out of 283,467 persons tested for HIV across the country, 15% were women accessing antenatal care. However, in free standing testing services provided by the Malawi AIDS Counseling and testing Organization (MACRO), of the 33,441 people tested, only 31% were female [42
]. This pattern of low female representation was observed at three sites. Whether women were under-represented because they had other access through antenatal care was not described. It has to be recognized also that although there may be more males accessing HIV testing, proportionally more women are likely to test positive than males in southern Africa. Women are the primary care givers in homes in virtually all countries in southern Africa and are likely to present to health facilities with sick family members. They also generally have greater interaction with health care services as they obtain maternity and family planning related care. Women's familiarity with health services may be working in their favour by facilitating their access to health services for HAART.
In many sub-Saharan countries, it is believed that women are mainly infected with HIV by their own spouses [43
]. It is therefore possible that men may be reluctant to acknowledge that they may be infected compared to women who may be forthright in accessing care believing, and little stigmatized as HIV is perceived to have been acquired from a spouse. Men on the other hand may fear that they may be perceived to have acquired infection outside of marriage. Krawczyk et al reported that anticipated stigma was associated with delay in seeking HIV care in the United States [10
Bachmann has reported that late initiation of HIV treatment was less cost-effective than early initiation of therapy in southern Africa [45
]. It is therefore pertinent that an assessment is made to determine at which points in the care continuum delays occur in accessing care. Delays can occur due to patient factors and intrinsic health system factors [46
Badri et al have suggested that for resource-limited settings, prioritizing symptomatically very sick patients for HAART is likely to be the most rational way of scaling-up HAART programs [47
]. Palombi et al however argued that "this minimalistic approach" is likely to put many people in harms way unnecessarily by delaying therapy to a point where it may not be as effective anymore due to unredeemable immune compromise [48
]. The harm that may result from delay in initiating therapy has probably been shown in Malawi by Zachariah et al [49
] where advanced HIV disease is shown to significantly contribute to mortality among patients on HAART. Hosseinipour et al [50
] have also demonstrated a high mortality of patients within the first few weeks into entry to HAART programs. If males are presenting late into treatment programs they are likely to suffer significant mortality and therefore limit the benefits of HAART programs. There is need to assess whether males have worse mortality while on HAART compared to females. A study by Mshana et al reported that barriers in accessing HAART included difficulties in identifying a treatment buddy, transport costs to treatment centers, fear of disclosure of HIV status and perception that hospitals were unfriendly and confusing [51
]. The treatment "buddy approach" is aimed to enhance adherence to therapy as patients have an identified treatment support person. However, as the requirement to identify a treatment buddy seems widespread in southern Africa [51
], there is need to assess its pros and cons. For instance, could it be that men have trouble identifying treatment buddies and so may be under-represented in treatment programs? Any suggestion to change the policy should also consider the potential loss in patient treatment adherence that may occur in the absence of a treatment buddy, who is currently perceived as key in supporting patients.
HIV prevalence is highest in the 30–34 years group at 19.2% and 29.4% respectively. Of particular note also is the fact that the mean or median ages for patients on HAART treatment ranged from 33 and 38 years. This suggests that the groups with highest likelihood of symptoms i.e. approximately 10 years following infection are also being fairly represented in treatment programs. It is also important to note that most recently infected person do not become clinically eligible for HAART under the current guidelines in operation in most of the countries. People who are starting treatment may have acquired infection 5 to 10 years before and only becoming eligible for HAART due to waning immunity. As this study was restricted to adult treatment programs, the situation in children may be different.
With reference to Table , it is interesting to note that in general the gender distribution seem to be maintained across programs despite of differences as to whether patients were required to pay for medications or not.
It is interesting to note that the study by Hosseinipour et al [18
] reported on a period when a fee-paying HAART treatment program was running in Lilongwe i.e. before free treatment was started. During this period, there were slightly more males (51.5%) than females (48.5%) and patient drop out was at 35.5%. In Zambia high drop-out rate were also observed prior to 2004 when most of the patients were obtaining HAART on fee-paying basis [28
]. At this time, there were 52.1% females versus 47.9% males on HAART. Subsequent to the introduction of national free HAART services, there are more women than males on treatment. However, in general there was still female gender predominance even in paying treatment programs.
There are several limitations of this review. Firstly data were mostly obtained from peer reviewed journals indexed in the databases that we reviewed. Treatment programs that did not publish reports were therefore missed. We however complemented the lack of peer-reviewed publications with abstracts from conferences. Again this source of data may be selective. If HAART treatment that did not publish their experiences have different gender distribution from those that published, then our conclusions may have to be revised. We however do not believe that the non-published experiences were overall systematically different from the reports reviewed in our study. The strength of our study also comes from the fact that we have presented data obtained at different time periods within the same country but still consistently showed that the gender distribution was in favour of females. This probably offers strongest evidence that the gender pattern is unlikely to be spurious or fluctuating but rather it is a stable phenomenon. The report from Tanzania [32
] was less informative as the actual numbers were not reported. The percentages of patients on HAART were however provided.
As our search was limited to publications in English, it was possible that we missed some relevant articles published in French (Democratic Republic of Congo) and Portuguese (Angola and Mozambique). We were however able to find articles in English from Mozambique, a Portuguese-speaking country, and these were included in the review. There was also likely to be overlapping of patients cohorts among the reports. For instance, in any country where both national treatment figures and localized patient numbers where obtained, many of the patients at the local setting would also be counted within the national figures. We do believe that such double counting in fact emphasizes the point that even when both larger and smaller patient cohorts were assessed, the gender distributions were maintained.