Hormonal contraception and higher risk of non-AIDS-defining events in Nashville cohort
Polly Clayden, HIV i-Base
Studies evaluating the effect of hormonal contraceptives (HC) on HIV disease progression have shown conflicting results. Previous findings have been from resource limited settings (RLS) and have not looked at the effect of HC on non-AIDS defining events (non-ADE).
Mainly observational data from Africa and Asia has shown both higher and lower rates of HIV disease progression in women receiving HC. Observational data in HIV negative women has shown an association between HC and metabolic complications.
Vlada Melekhin presented findings from a retrospective cohort study of HIV-positive women attending the Comprehensive Care Center (Nashville, TN) between 1998-2008. The study investigated the association between HC (oral and injectable methods used >28 days) and AIDS-defining events (ADE), non-ADE (ie cardiovascular, renal, liver, and metabolic diseases and non-AIDS associated malignancies) and death.
Eligible women were <55 years old with no history of pulmonary or deep venous thromboembolism, breast cancer, hysterectomy, or bilateral tubal ligation and not pregnant at first clinic visit. Women with no HC were evaluated from their first clinic visit and those using HC at HC start.
Logistic regression analysis included age, race, baseline CD4 count, viral load, and haemoglobin, CD4 nadir, history of ADE, non-ADE, HCV, antiretroviral (ART and non-ART) use, smoking status, IV and non-IV drug use, year of study start, and year of HC start.
Of 467 HC-eligible women, 112 (24%) were on HC at any time during the follow up. At baseline women on HC were younger, median 28.6 vs 35.6 years. They had higher CD4 count 523 vs 364 cells/mm3 and nadir, 340 vs 280 cells/mm3, and lower median viral load, 3.1 vs 4.1 log10 copies/mL. They were less likely to be coinfected with HCV, 5% vs 15% or inject drugs, 16% vs 27%, both p<0.03.
There was no statistical difference in ART use between the HC and no HC groups, 30.4% vs 26.8%, respectively, nor in prior ADE or non-ADE.
Of the 112 women using HC, 51 used oral and the remaining 61 used injectable for a median duration of 7.6 and 13 months respectively, p=0.26.
HC users had longer follow-up compared to non HC users, median 2.8 vs 1.5 years for ADE, 2.8 vs 1.6 years for non-ADE and 3.8 vs 2.1 years for death.
The investigators reported a lower proportion of deaths in the HC group, 6% vs. 15%, p=0.01. But these women had more new cardiovascular non-ADEs, 12% vs. 5%, p=0.02.
In the adjusted analyses, HC use was associated with a statistically significantly higher risk of non-ADE HR 2.0, (95% CI 1.28, 3.1), p=0.02 and non-ADE/death HR 1.89, (95% CI 1.25, 2.87), p=0.03). Risks of ADE and ADE/death were also higher among HC users but did not reach statistical significance: HR 1.51 (95% CI 0.59, 3.85), p=0.39 and 1.49 (95% CI 0.72, 3.11), p=0.29, respectively. Women using injectable HC were at a higher risk of non-ADE and non-ADE/death, HR 2.0 and 1.9 respectively, both p=0.03 and those using oral HC only non-ADE, HR 1.9, p=0.02.
The investigators plan further analyses from this cohort including looking at the effect of ART and suggested as the number of women with HIV who are of child-bearing age increases, it is important to better understand any negative effect of HC on their health.
Investigations into the use of hormonal contraceptive methods and its effect on disease progression in HIV positive women have led to conflicting results,
One randomised controlled trial conducted in Zambia showed risk of CD4 decline or death with hormonal contraception, compared to use of the copper IUD.  But the study was designed to look at the incidence of pregnancy and pelvic inflammatory disease in the IUD users and there was considerable discontinuation and switching between methods. Data from several observational studies do not confirm this effect.
The association with non-AIDS events found by Melekin are interesting but should be interpreted cautiously given that two large trials have reached very different conclusions. Observational data is vulnerable to unmeasured confounding and (as they are in the general population) lifestyles are very different between those that use hormonal contraception and those that do not. These differences could (feasibly) explain differences in incidence of some of these serious events. For example, even smoking and alcohol use may be different in the groups. Whille the researchers may have used propensity scores, these do nothing to tackle unmeasured confounding (and are arguably little better than standard multivariable logistic regression models).
The WHO recently held a stakeholders meeting to review the evidence on hormonal contraception and HIV, not only to consider the effect on disease progression but also female to male HIV transmission and HIV acquisition by negative women. The organisation and partners are producing three systematic reviews and there will be a statement from the consultation.
Currently the WHO medical eligibility criteria for contraceptive use defines hormonal contraceptives as category 1 – ie no restriction on the use of the methods for women with HIV (including AIDS).
- Melekin V et al. Hormonal contraceptive use is associated with a higher risk of non-AIDS-defining events in HIV-1-infected women. 2nd International Workshop on HIV and Women. 9-10 January 2012, Bethesda, MD. Oral abstract O_13.
- Stringer EM et al. A randomised trial of the intrauterine contraceptive device vs hormonal contraception in women who are infected with the human immunodeficiency virus. Am J Obstet Gynecol 2007 August; 197 (2):144-148. Free full text: