Beyond the Headlines: What’s new – and what’s the same – with the latest HC-HIV data

January 12, 2015

A newly published analysis by Lauren Ralph et al and an accompanying commentary in the journal Lancet Infectious Diseases is stirring up questions about the relationship between DMPA (brand name Depo Provera) and other progestogen-only injectable contraceptives and the risk of HIV acquisition among HIV-negative women. Based on a meta-analysis of previously published studies, the report’s authors state that DMPA use is associated with a “moderate risk” of HIV infection.

The study triggered a wave of headlines and tweets that boiled down the complexities and caveats of this analysis into an over-simplified statement that DMPA increased women’s risk of acquiring HIV by about 40 perent. This isn’t precise or true. Because DMPA is an important contraceptive choice – it is discrete and provides protection against unwanted pregnancy for three months after a single shot – and because in many parts of the world, DMPA is the only long-acting, discrete option available to women, it is really important to add nuance to these headlines, while also taking the issue of a link between HIV and hormonal contraception quite seriously.

The first and arguably most important thing to understand about this new paper is that it is not based on new data. It is a new analysis of a set of observational studies of rates of HIV in women using different contraceptive methods. All but one of the studies included in the analysis here has been included in previous systematic reviews. (There is a difference between previous systematic reviews and the statistical meta-analysis used in this paper… more on that follows!)

The bottom line is that this conclusion is not based on new information; it just crunched the numbers in a different way.

Here are some key points to help advocates read, analyze and act on the latest information.

About the data analyzed and the methods

  • The study by Ralph et al is different from previously published papers for two main reasons. First: It is a statistical meta-analysis, rather than a systematic review. Second: It includes one recent paper that was published after the most recent systematic review. What do these terms mean?
    • Systematic reviews, as the name implies, typically involve a detailed and comprehensive plan and search strategy for identifying and synthesizing all relevant studies on a particular topic.
    • A meta-analysis involves using statistical techniques to combine data from multiple studies (such as those identified in a systematic review) into a single quantitative estimate or summary effect size.
    • Conducting a meta-analysis of observational data can be a controversial technique to use. There is often disagreement about whether it is or is not appropriate to combine studies into a single estimate. Some experts have argued that this approach can result in “spurious precision”, since the summary estimate will only be as good as the studies combined to produce it.
  • What does these terms mean in the context of the new paper?
    • Previously published systematic reviews have identified all of the studies that exist at a particular time on this issue. The most recent systematic review published in October 2014 by Polis et al. is available online. That systematic review concluded that the relationship between injectable hormonal contraception and HIV risk is inconclusive. It recommended that women choosing progestin-only injectable contraceptives such as DMPA or NET-EN should be informed of the current uncertainty regarding whether use of these methods increases risk of HIV acquisition. It also emphasized that users of DMPA and other injectables should be empowered to access and use male and female condoms and other HIV prevention tools. Messages and programs that support “dual protection” (against unwanted pregnancy and infection with HIV and other STIs) should be provided to all women at risk of HIV.
    • Prior systematic reviews of the HC-HIV acquisition question have carefully described the methods, results and interpretation of the body of evidence, but they deliberately chose not to have crunched the numbers from all of the studies together, due to methodological concerns with that approach. A meta-analysis is when the number-crunching happens. That’s what makes the Ralph meta-analysis new.
  • The Ralph meta-analysis includes one new study (Crook et al 2014) that was published subsequent to the most recent systematic review.
  • All of the studies considered in the systematic reviews and in this meta-analysis are observational. None of these studies randomly assigned women to different methods, they could have biased results (women who choose a specific contraceptive method might have other factors that affect their HIV risk that the study doesn’t pick up on).
  • The studies that were considered include:
    • Data gathered from studies that were designed to answer other questions (e.g. a study looking at herpes treatment in serodiscordant couples that also documented women’s contraceptive use).
    • Prospective studies that followed women who had chosen different contraceptive methods and gathered information about HIV rates among women using different methods.

About the findings

  • The Ralph meta-analysis concludes that there was evidence of a “moderate increase” (40%) in HIV risk in the ten studies of DMPA use that were combined. There was no evidence of increased HIV risk associated with oral hormonal contraceptive pills or in other progestin-only injectable contraceptives (e.g. NET-EN). The meta-analysis of all the studies yields an overall increase in risk of about 40 percent or a hazard ratio (HR) of 1.40.
  • The increased risk is greatest in women at “high risk” of HIV infection, which the authors define as women who engage in commercial sex work and/or those who are in serodiscordant couples. A sub-analysis that excluded data from high risk women found a 30% increase, or HR of 1.30 among women in the “general population.”
  • The authors note that “Meta-analyses of observational studies, like observational studies themselves, are inherently prone to bias and cannot be used to address whether the association between hormonal contraception and HIV is causal”. This caution is consistent with previously published systematic reviews of observational data, in which authors have deliberately chosen not to quantitatively combine estimates.

About the authors’ interpretation and conclusion

  • The authors argue that the estimate of increased risk (40% or an “odds ratio” of 1.40) should be used to guide more precise models of what it would mean if DMPA does, in fact, increase HIV risk. These models calculate the relative contribution to new HIV infections given different theoretical estimates of the magnitude of risk associated with DMPA use. All of the modeling studies conclude that the question of whether DMPA increases HIV risk is of greatest relevance in Eastern and Southern Africa, where injectable contraceptive use and HIV rates are both high.
  • The authors also argue that that the moderate risk associated with DMPA use should be weighed against the risks of maternal morbidity and mortality if DMPA is “banned”. This is a misleading statement. In the many discussions at WHO, country and community levels that have taken place in the past few years on this issue there is no scenario or proposal in which DMPA would be banned or even removed from programs without provision of a comparable alternative. The relevant proposals and programs—as exemplified by South Africa’s new contraceptive policy—seek to expand “method mix” (the range of options women can choose from). Specifically, the proposals and programs identified by advocates, funders and many other stakeholders focus on expanding the use of other long-acting, discrete methods, such as implants and the intrauterine device (IUD), that could be used instead of DMPA by women making informed choices based on what is known and unknown about all the options available.
  • The authors argue that the findings “emphasize an immediate need” to look more closely at how DMPA use might impact risk among women in serodiscordant relationships and/or women engaged in sex work. They suggest additional analyses of existing data in these populations, but don’t suggest a change in policy or messaging.

About what should happen next

  • These findings are not news in and of themselves. They identify the same trend seen in some individual studies and utilize the same information as in previously-published systematic reviews. However, each time a study or analysis is published on this issue, particularly when the results suggest a significant effect, in a zone of such uncertainty—it triggers fresh discussion and debate. There have already been a range of media reports focusing on the “40 percent” figure and suggesting that this is now the definitive estimate of HIV-related risk for injectable HC users. In light of this publication, it would be appropriate for the WHO to re-convene an expert stakeholder group to review both current recommendations and communications strategies regarding DMPA and similar products with a particular focus on the countries in East and Southern Africa where rates of HIV and DMPA use are high.
  • The study and the accompanying commentary reference, without naming, the proposed ECHO trial that would use a randomized design to directly measure rates of HIV in women using three different methods: DMPA, the Jadelle implant and the copper IUD. AVAC has worked in coalition with ICW East Africa, the ATHENA Network and many other women’s organizations to articulate the urgent need for clarity on the relationship between HIV risk and DMPA and other hormonal contraceptive methods. We have articulated the need for a trial that provides clarity and is able to influence policy. It is critical that the ECHO team engage with civil society stakeholders to explore the understanding and implication of this paper as part of a broader discussion about the planned launch of the trial later in 2015.
  • During the recent “summit” of the FP2020 initiative (the global family planning initiative that aims to increase women’s access to contraception worldwide), FP2020 leadership indicated that it would await and follow the results of the ECHO trial as well as country and WHO guidance. It is invaluable to the field for FP2020 to convene a meeting of family planning policy makers and implementers in potentially-affected countries to discuss existing plans, proposed expansion of method mix and processes for interpreting and acting on these results.
  • There is a robust civil society constituency following the issues around HC-HIV. Members of this dialogue have diverse views on whether a randomized trial such as ECHO is on the critical path—but are united in the need for family planning and HIV programming to:
    • Address the uncertainty with clear messages on knowns and unknowns, risks and benefits of all methods;
    • Invest in increased method mix today; and
    • Sustain investment in developing new contraceptive, HIV prevention and, especially, multipurpose prevention options that could, in the future, reduce HIV risk and prevent unwanted pregnancies.

AVAC will continue to work with partners to distill new findings, convene dialogues with scientists and other partners and ensure that an informed advocacy voice helps guide decisions in this key area. To get involved and hear what’s next, watch this space or contact us.