Effects of three contraceptive methods on depression and sexual function: An ancillary study of the ECHO randomized trial

Abstract Objective To compare the effects of depot medroxyprogesterone acetate (DMPA‐IM), levonorgestrel (LNG) implant, and copper intrauterine device (IUD) on mood and sexual function. Methods At the Effective Care Research Unit in South Africa, women already randomized in the ECHO Trial to the three methods were asked to participate in this study. Participants were interviewed at 3 and 12 months after enrollment using the Beck Depression Inventory and Arizona Sexual Experiences Scale, and at 12 months using the WHO‐5 Wellbeing Index and the Patient Global Impression scale. Results A total of 605 women participated. There was little difference in depression at 3 months across the three study groups. Contrary to our hypothesis, at 12 months, depression was lowest among DMPA‐IM users (16/167, 9.6%) and highest among IUD users (28/158, 17.7%) (p = 0.032). There was little difference in sexual function at any time‐point. More women in the DMPA‐IM group felt “very much better” on the PGI scale than in the IUD and LNG implant groups (p = 0.003). Conclusion Depression may be less likely with DMPA‐IM than with the other methods 1 year after initiation. Major differences in sexual functioning are unlikely. Unhappiness related to not using DMPA‐IM, the most popular method in our setting, may have skewed results. Trial registration number: PACTR201706001651380.


| INTRODUC TI ON
Contraceptive discontinuation is a major issue that impacts unintended pregnancy rates. 1 Therefore, understanding the reasons for discontinuation is necessary to develop effective contraceptive services with appropriate counseling to inform women's contraceptive choices, and to manage adverse effects appropriately should these arise. Discontinuation rates for injectable progestins and the copper intrauterine device (IUD) in our setting are high, reported in a large locally conducted randomized trial as 16.5% and 14.7%, respectively, at a median of 20 months after initiation. 2 Furthermore, in the recently published Evidence for Contraceptive options and HIV Outcomes (ECHO) Trial in which our Effective Care Research Unit participated, contraceptive discontinuation was associated with 71% of 181 pregnancies that occurred during the course of the trial. 3 Despite their common occurrence, differences in adverse effects between contraceptive agents remain poorly elucidated.
Specifically, the effect of contraception on mood has been understudied, particularly in low-and middle-income countries, with published reports from high-income countries suggesting the possibility of both an increase and a decrease in depression risk. [4][5][6][7] A nationwide cohort study from Denmark found that, compared with non-users, use of various types of hormonal contraception was associated with a higher risk of subsequent depression and use of antidepressants 4 ; hormonal contraception has also been linked with suicide attempts and suicide. 5 In contrast, a national study from the USA reported reduced levels of depressive symptoms among sexually active women taking hormonal contraception. 6 In addition, a Swedish cohort study found an association between contraception and the use of psychotropic medicine among adolescents. 7 The effects of contraceptives on mood may be complex, including, for example, biological effects on endogenous steroid hormones and psychological effects, such as relief from anxiety about unintended conception. The only study to date that has evaluated biological and psychological effects related to contraception was a randomized controlled trial (RCT) conducted in South Africa, which found increased postnatal depression among women using progestin-only contraception. 8 Such confusing and contradictory findings suggest that robust evidence from randomized trials is needed, but such studies are rare because of anticipated difficulties in randomizing women to different contraceptive agents.
To our knowledge, no RCTs have evaluated the relationship between contraceptive use and depression, if any, among women outside of the postnatal period. Limited evidence from two small RCTs conducted among postnatal women in South Africa have shown that, when administered soon after pregnancy, the risk of postnatal depression may be increased with injectable progestin contraception. 8,9 However, as a result of the short follow up of women in these trials, it is not known whether this adverse effect of progestin-only contraception resolves or remains with longer-term use. Similarly, the impact of contraceptive agents on sexual function has not been well-studied and limited evidence suggests that they may decrease or increase libido, or have no effect. 10 The multicenter ECHO Trial randomized women to intramuscular depot medroxyprogesterone acetate (DMPA-IM), the IUD, and the levonorgestrel (LNG) implant and evaluated the effect of these agents on HIV incidence. The trial, therefore, provided a unique opportunity for us to also compare effects on depression and sexual function within the context of a stringent RCT protocol. 3 We hypothesized that the progestin-only contraceptives DMPA-IM and the LNG implant may be associated with lower mood and more sexual dysfunction compared with the IUD.

| Design and participants
The protocol for this study was published in BMJ Open in 2018. 11 The study is ancillary to the ECHO trial, which was a large multicenter RCT with three parallel arms (1:1:1). In the ECHO trial, participants were randomized to three contraceptive methods (IUD, DMPA-IM, and LNG implant) and the effects on HIV incidence and pregnancy rates were compared (ECHO trial registration: NCT02550067. 3 The trial was conducted at sites in Kenya, South Africa, Swaziland, and Zambia. Participants were sexually active, HIV seronegative women between the ages of 16 and 35 years who were seeking effective, long-acting contraception and who gave informed consent.
Exclusion criteria included using any of the study contraceptive agents in the preceding 6 months and planning a pregnancy in the next 18 months.
In the ECHO trial, 7829 women in total were randomized and followed up for 18 months. The findings, which were published in the Lancet in 2019, showed no clear differences in HIV incidence and pregnancy rates between the three contraceptive methods evaluated. 3 No data on mental health were collected in the main trial.
The objectives of this ancillary study were to compare the effects of DMPA-IM, the IUD, and the LNG implant on mood and sexual function. Participants were drawn from one of the trial centers, the Effective Care Research Unit, in East London, South Africa, and additional informed consent was obtained. Out of 7829 women enrolled in the ECHO trial, 615 came from our center; of these, 605 consented to participate in the ancillary study.

| Procedure
Following online registration, the ECHO trial centrally randomized participating women to DMPA-IM (150 mg/ml, Depot-Provera ® ; Pfizer, Puurs, Belgium) given intramuscularly every 12 weeks, insertion of the Copper IUD (Optima TCu380A ® ; Injeflex, Sao Paolo, Brazil) at enrollment, or insertion of the subdermal LNG implant (150 mg, Jadelle ® ; Bayer, Turku, Finland) at enrollment. Participant and personnel blinding was not possible because of the nature of the interventions, and participant follow up was conducted at 1, 3, 6, 9, 12, 15, and 18 months. ECHO participants at our site who additionally consented to participate in this ancillary study were enrolled at the 1-month ECHO follow-up visit, and were interviewed at the 3-and 12-month visits for this purpose by trained research assistants. The 12-month follow up was an extension of the protocol that had originally planned a 3-month follow up only. Some participants (n = 53) only contributed data at the 12-month visit.

| Outcomes and instruments
Primary outcomes of this ancillary study were depression and sexual dysfunction. Depression was evaluated using the Beck Depression Inventory (BDI-II) and sexual function was evaluated using the Arizona Sexual Experiences Scale (ASEX). 12,13 The BDI-II has previously been translated into the local language (IsiXhosa), validated, and used in our study population. 14 The ASEX scale is a five-item scale that quantifies sex drive, arousal, vaginal lubrication, ability to reach orgasm, and satisfaction from orgasm. 14 We have used the ASEX scale in a previous study in our population and found it to be a useful tool although, to our knowledge, it has not been formally validated in this context. 9 Following enrollment into the ancillary study at the 1-month ECHO trial visit, these scales were administered by trained research assistants at the 3-month and 12-month assessments.
As mood and an individual's perception of well-being are related, we also asked participants to complete the World Health Organization (WHO) Well-being Index, which consists of five statements on well-being that can be rated from 0 to 5. 15 The well-being score, which can range from 0 to 25 with 0 representing worst possible and 25 representing best possible quality of life, was calculated by adding the figures of the five answers. In addition to the WHO Well-being Index, we used the Patient Global Impression Scale (PGI).
This consists of one question on an individual's perception of the effect of a treatment on their well-being. 16 The WHO Well-being Index and PGI scale were administered at the 12-month follow-up visit. Each of these tools can be self-administered, but to optimize the quality of the data, trained research assistants asked and explained the instrument questions to participants in their local language, and recorded participants' responses.

| Statistical analysis
The ECHO trial was designed to detect a 50% difference in HIV incidence between any of the three contraceptive methods with 80% power. 3 For the ancillary study, we assumed a reduction in the incidence of depression with the alternative contraceptive methods compared with DMPA-IM of 50% based on a rate of depression with DMPA-IM of 26.5% from a previous study conducted in our setting. 9 We calculated that we would need at least 158 participants in each arm to detect a difference with 80% power between any of the three arms. 11 Stata Statistical Software V.15 (StataCorp., College Station, TX, USA) 17 was used for analysis, which was performed by YB. We used standard BDI-II thresholds to categorize any depression (≥14), moderate depression (≥20) and severe depression (≥29) 12 ; and the standard ASEX threshold to categorize sexual dysfunction (≥19) among other suggested criteria. 13 As per the preplanned analysis, we performed separate pairwise χ 2 tests between each of the three methods. The χ 2 test was used to compare categorical outcomes, unless events were equal to or less than five; in these instances, the Fisher's exact test was used. As per the pre-planned analysis, we performed separate pairwise χ 2 (or Fisher's exact) tests between each of the three methods. Mean scores were compared using the Kruskal-Wallis test and for non-normally distributed data we used the Wilcoxon test to compare medians and interquartile ranges.

| Ethical approval and study registration
Baseline characteristics of participants in the three study groups were similar, suggesting that the ECHO randomization process from which our study groups were derived was successful. There was no statistically significant difference between the study groups with respect to BDI or ASEX findings at 3 months (Table 2). [20.3%] for the IUD and LNG implant groups, respectively), but this did not reach statistical significance (Table 3). With the PGI scale, the majority of participants in each group felt "very much better"; however, significantly more participants using DMPA-IM group felt "very much better" compared with those using the LNG implant and IUD (p = 0.003). Only one participant, in the LNG implant group, indicated that she felt "a little worse" (Table 4).
At the end of the study, 466/487 women (95.7%) were still using contraception; however, 38/487 (7.8%) had changed from their study-allocated method. Fewer women in the DMPA-IM group changed their allocated method compared with the LNG implant and IUD groups (p = 0.002); most changes made were to DMPA-IM from the IUD and LNG implant, 15/158 (9.5%) and 13/162 (8.0%) of women allocated to these methods, respectively (Table 5).

| DISCUSS ION
This prospective study conducted alongside the ECHO Trial at one of the ECHO Trial sites (the Effective Care Research Unit) in South Africa at 3 and 12 months after randomization found that at 12 months women using DMPA-IM were at lower risk of depressive symptoms than women using the IUD or the LNG implant and were more likely to feel "very much better" on the PGI scale. No significant differences in the risk of sexual dysfunction or sexual function scores were found.
Before conducting the study, we hypothesized that depressive symptoms would be higher with the progestin-only contraceptives based on previous studies that have found depression to be increased among users of postnatal injectable hormonal contraceptives. 5  Two out of 552 participants did not contribute ASEX data. c Defined as an ASEX score of ≥19 or any one ASEX scale item with a score of >5 or any three items with a score of >4. 14 This is the first prospective study associated with a randomized trial where follow up with respect to these outcomes has been extended to 12 months after contraception initiation, and in a non-postnatal cohort. Rates of any depression in this ancillary study were lower than that seen in a previous study in our setting 9 and may be due

LNG implant vs IUD
A little worse 0 (0.0%) 0 (0.0%) 1 (0.6%) 0. The ECHO trial findings show that women using the IUD were more likely to experience menorrhagia, pelvic pain, dysfunctional uterine bleeding, and dysmenorrhea than those using DMPA-IM. 3 Dysfunctional uterine bleeding was also more common with the LNG implant than with DMPA-IM in the ECHO trial. Therefore, it is plausible that the higher rates of depressive symptoms with the IUD and the LNG implant at 12 months may be related to these troublesome menstrual disturbances.
Although this ancillary study recruited women after randomization, it included 90% of the originally randomized participants, and the proportion of ECHO-randomized participants in each intervention group was similar at both time-points evaluated. In addition, because of the alignment of this ancillary study with the stringent follow up of the associated rigorous RCT, 12-month follow up was high. These study strengths mean that meaningful bias due to selective participation is unlikely. Another strength is that we performed intention-to-treat analysis; therefore, the effect of more women switching to DMPA-IM than to the other methods would have had the effect of under-estimating effects of DMPA-IM rather than overestimating them.
A limitation of the study is that we did not collect data on mood and sexual functioning at baseline; neither did we record the use of anti-depressant medication or counseling; however, such health interventions are rare among women in our setting. Another limitation is that we did not evaluate previous contraceptive use among women in our study, which may have influenced the potential for women to switch methods or to discontinue a given method. As sug-  3 and confirms the importance of counseling prospective users about these adverse effects and managing them effectively when they arise.
In conclusion, major differences in sexual functioning between women receiving DMPA-IM, the IUD, and the LNG implant are unlikely. DMPA-IM use in South Africa is probably associated with a lower risk of depression at 12 months after contraceptive initiation; however, evidence from other settings is needed.

DATA AVA I L A B I L I T Y S TAT E M E N T
We support data sharing and such requests can be emailed to Mandisa Singata-Madliki (mandisa.singata@gmail.com).

ACK N OWLED G M ENTS
The The contents of this paper are solely the responsibility of the authors and do not necessarily reflect the views, decisions or policies of the institutions with which they are affiliated, the ECHO study funders, or the supporting governments.

CO N FLI C T S O F I NTE R E S T
The authors have no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
MS-M conceived the study and contributed to its conduct and to writing of the manuscript. JH contributed to study design, interpretation of data and writing of the manuscript. TL contributed to study design and wrote the first draft of the manuscript. YB contributed to the study design, data analysis and writing of the manuscript. FC-Ld'H contributed to study design, and reviewed and advised on the draft manuscript. All authors approved of and contributed to the final manuscript.