Hormonal contraception and risk of HIV acquisition: a difficult policy position in spite of incomplete evidence

Injectable hormonal contraceptives are the most widely used modern contraceptive method in many countries, and are especially popular in sub-Saharan Africa. There are two primary types of injectable hormonal contraceptives: depot medroxyprogesterone acetate (DMPA) and norethisterone enanthate (NET-EN).¹

Some studies have suggested that women using injectable contraception are at higher risk of acquiring HIV infection than non-users, although other studies have not shown any significant increase in risk. In settings where the risk of HIV infection is high, these conflicting findings present a difficult choice. Modern contraceptive methods, including injectable hormonal contraceptives, are critically important for preventing unintended and mistimed pregnancies, reducing maternal mortality and avoiding the consequences of unsafe abortion.² Women and contraceptive providers who advise women in settings where there is a risk of HIV infection are faced with a complex balance of risks regarding contraceptive choice, complicated by the fact that the evidence of whether or not there is an increased risk is far from certain. Furthermore, although additional research has been promised, it may not resolve the question in the near future.

The evidence

In early 2012, the World Health Organization (WHO) convened a meeting of experts to discuss recent research and other available evidence on use of hormonal contraception by women at high risk of HIV and those currently living with HIV.³ Nine key studies provide the most relevant and useful information on any link between hormonal contraception and the risk of acquiring HIV infection although all have limitations. The following overview presents relevant evidence.

A prospective study initiated in 1993 of sex workers in Mombasa, Kenya followed study participants monthly for incident HIV infection. Users of combined oral contraceptives and DMPA were 1.5-fold and 1.8-fold more likely, respectively, to become infected than a control group of non-users of hormonal contraception (which included women not currently using any contraceptive method and women who were sterilized) after adjustment for demographic factors, sexual behavior, condom use and a recent history of sexually transmitted or other genital tract infections.⁴ The implications of this study for women at risk of HIV infection, but not as frequently exposed as sex workers, were not clear, and additional studies among women in the general population were conducted. A South African study that followed women at three-month intervals for one year between 1999 and 2002 showed no increase in HIV risk for DMPA users and a non-significantly elevated risk for NET-EN users.⁵ In a prospective study of 4,435 women assessed at three-month intervals in Uganda and Zimbabwe for up to two years between 1999 and 2006, researchers found no overall increase in HIV risk among combined oral contraceptive users compared with non-hormonal contraceptive users after adjustment for demographic factors, sexual behavior and condom use. However, there was a borderline significantly elevated risk among DMPA users.^6,7 Subgroup analyses revealed an increased risk of HIV infection in two important subgroups: women under age 25 (2.8-fold higher) and those without herpes simplex virus type-2 (HSV-2) infection at enrollment to the cohort (4.5-fold higher). This observation prompted a further analysis of the Mombasa study, but no effect modification by age or HSV-2 status was found in that study population.⁸

Other evidence on any relationship between hormonal contraceptive use and HIV infection has come from studies originally designed to investigate the impact of other exposures on HIV incidence. These studies include a trial of different cervical cancer screening strategies,⁹ randomized controlled trials of HSV-2 suppressive therapy,^10–12 and randomized controlled trials of candidate microbicide products.^13–16 Most of these studies are reassuring, showing no significant increased risk of HIV acquisition in hormonal contraceptive users. However, two studies showed a significant two-fold higher risk of HIV acquisition among injectable hormonal contraceptive users. One was a study of HIV-serodiscordant couples from seven countries in southern and eastern Africa followed at three-month intervals for up to two years between 2004 and 2010¹² and the other was a study in Durban, South Africa of HIV-negative women screened for eligibility and followed in a randomized controlled trial of the candidate vaginal microbicide PRO 2000.¹³

The classification of exposure to hormonal contraception was not uniform across studies, with some studies unable to distinguish between different classes of hormonal contraceptives or between different injectables. (Some studies included all hormonal contraception as a single class while others distinguished between oral and injectable contraception.) While most studies followed up with participants every three months, follow-up intervals ranged from one to six months. Longer intervals may have made recall of contraceptive methods and sexual practices less reliable. Also, it is difficult to date onset of HIV infection accurately with longer intervals between HIV tests. This introduces greater uncertainty about whether a particular contraceptive method was used around the time of infection. But the main methodological problem concerns condom use, as condoms are used less by women using hormonal or other reliable methods of contraception. (For example, in the prospective study in Uganda and Zimbabwe, 4% and 7% of combined oral contraceptive and DMPA users, respectively, reported consistent condom use at baseline compared with 63% of study participants not using hormonal contraception.)⁶ In addition, the accuracy of condom recall and reporting may differ between hormonal contraceptive users and non-users, which means that adjustment in the analysis for differences in reported condom use is at best imperfect.

The international group of experts convened by WHO weighed the evidence and considered the implications for women at high risk of HIV infection. The evidence was summarized and recommendations developed using the GRADE process (www.gradeworkinggroup.org). The WHO recommendations on the use of hormonal contraception among women at high risk for HIV infection were based on a systematic review of the epidemiological data, as well as a thorough discussion of biological plausibility, implications for country programs, the importance of HIV prevention and risks associated with unintended pregnancy including maternal mortality and pregnancy-related morbidity. A decision analysis evaluating the competing risks of HIV acquisition and maternal mortality in Chad, Kenya, South Africa and Uganda was also considered.¹⁷ The group concluded that the evidence was not sufficiently strong to change current recommendations in the WHO Medical Eligibility Criteria for Contraceptive Use.¹⁸ Based on current evidence the group recommended that women at high risk of, and those living with, HIV infection can continue to use all methods of hormonal contraception without restriction as well as male and female condoms.³

The implications

The available evidence leads to some clear implications for individual women at risk of HIV infection who wish to use a contraceptive method:

•	It is critically important that all women have access to and use condoms whenever there is a risk of HIV transmission. Consistent use of condoms greatly lowers the risk of HIV transmission and means women need be far less concerned about a potential increased risk of HIV infection when choosing their hormonal contraceptive method.
•	If there is a higher risk of HIV infection with use of hormonal contraceptives, DMPA appears to be more implicated than other hormonal methods (NET-EN or combined oral contraceptives). Information on the safety of NET-EN is very limited, and there is no information on the safety of implants or progestogen-only pills in relation to the risk of HIV acquisition.
•	Since the risk of HIV infection appears to be greater during pregnancy,19,20 this means that stopping DMPA without switching to another reliable contraceptive method could still leave women at higher risk for HIV as a result of unintended pregnancy.21 Furthermore, pregnancy itself is associated with considerable risk in settings with high maternal morbidity and mortality.
•	Switching from DMPA to other reliable contraceptive methods, such as implants, intrauterine devices, combined oral contraceptives or tubal ligation, may be an option where women have access to these methods. But many women living in areas with high HIV incidence have a limited range of methods available to them.

Potential policy options

Several policy options could be considered given the evidence, but all have serious limitations and shortcomings.

•	Ban DMPA in countries with high HIV incidence? The evidence of harm with DMPA is not sufficiently strong for such drastic action, particularly since it is a popular contraceptive method for women living in areas of high HIV prevalence where few reliable alternatives are available. The majority of countries with large HIV epidemics also have high maternal mortality rates, which means that replacing DMPA with less reliable options may reduce deaths from HIV but result in more deaths in women of reproductive age.21 Avoidance of unintended pregnancy is the most effective way of reducing maternal deaths. Singh and colleagues estimated that over 40% of global maternal deaths were averted in 2008 by contraception alone.2 Rodriguez and colleagues concluded that an additional nine maternal deaths would occur per HIV infection averted if DMPA were removed from use and not replaced by at least 70% of women switching to an intrauterine device or combined oral contraceptives.17
•	Advise women and their partners to reduce their individual risk of HIV infection by abstaining from sex or insisting on condom use? Such an approach reinforces the message that hormonal contraception does not reduce risk of HIV infection and all women (and all men!) must use condoms whenever there is a risk of HIV infection. But ensuring accessibility, availability and use of condoms, male and/or female, continues to be a major challenge in many settings with high HIV incidence. In addition, abstinence is seldom either a preferred choice or an option for most women, and experience shows that such a policy has little success in avoiding unwanted pregnancy or HIV infection.
•	Expand the contraceptive method mix to include methods that are safer with regard to HIV risk and more reliable with regard to preventing unintended pregnancy, such as sterilization, intrauterine devices and implantable hormonal contraceptives? As these methods are much less widely used in the African region, introducing and sustaining them will take a large investment of resources in provider training, public advocacy, individual counseling and the redesign of components of family planning programs.
•	Conduct more research to understand the risks associated with different contraceptive methods? Investments in research must continue in order to resolve the current uncertainty as quickly as possible. Completed HIV prevention trials of microbicides and oral pre-exposure prophylaxis are being analyzed, but the quality of information on contraceptive exposure and condom use may limit their value. There is discussion about launching a new study designed specifically to address the HIV risk associated with different contraceptive methods, but such a study would not yield results for at least three to five years.

Women and their partners must be provided with information and contraceptive options so that they can make their own decisions. But not all women will find this a satisfactory solution, and look to experts to provide guidance that they can then apply to their own situation. And what should policy makers do, while waiting for the experts to agree on whether there really is an excess risk, and if there is, what yardstick to use to measure the competing risks and benefits? Experts have a duty to provide easily understood information that will help program managers and users apply evidence to their own context and priorities. In the meantime, we are all caught in a difficult policy situation.

Acknowledgements

Many thanks to Mary Lyn Gaffield (Department of Reproductive Health and Research, WHO) for providing valuable comments.