Sex practices and awareness of Ebola virus disease among male survivors and their partners in Guinea

Study design

EVD male survivors were eligible if aged >15 years and able to show the certificate that every survivor was given at discharge from the ETU, which confirmed their Ebola-free status in the blood. They were recruited using databases from the CNLEB, WHO and USA Centres for Disease Control (CDC), which in total had records of 491 EVD male survivors aged >15 years. These lists were compared with those of the EVD survivors' associations in each local community. We aimed to recruit as many eligible survivors as possible. Because many were lost to follow-up (ie, moved to another community, had fake number or no contact), we only invited 240 (49%) of the eligible survivors (Supplementary figure). Invited EVD survivors were asked to bring along their sexual partner(s). For every three recruited EVD survivors, we aimed to recruit one individual from the same neighbourhood for the comparison group. For this group, we managed to recruit only 83% (65/78) of the original sample size because of the stigma around the disease (many EVD survivors did not talk about the disease to their relatives or did not want them to know about their disease). The comparison group was recruited with the help of the Ebola survivors’ associations using random digit dialling from a list of up to 20 relatives or acquaintances given by each survivor at enrolment. Individuals in the comparison group were from the same communities as survivors and partners. They were not matched to them, but each regression model we adjusted for region and zone of residence. Participants were recruited in the prefectures of Coyah and Forécariah (lower Guinea), Macenta and Guéckédou (forested Guinea), and in the capital, Conakry. Recruitment of study participants started on 21 April 2016 in Forécariah and ended on 23 June 2016 in Guéckédou. All interviews were face-to-face and were done at the prefectural centres of the CNLEB in urban areas, and in the closest healthcare centres in rural areas. They were always done in private rooms, in the presence of only the interviewer and the interviewee, and the latter could choose the preferred language for the interview between French and one of the local languages (Susu, Malinké, Pular and Konianké). All interviewers were either medical doctors or socioanthropologists from Guinea, able to speak one or more local languages in addition to French, and trained in Good Clinical Practice. Every participant signed a written informed consent. The following data collection tools were used: three questionnaires, one for each group (ie, survivors, partners and the comparison group), a concept paper for HIV/syphilis counselling and a document to report the HIV or syphilis-positive cases. In addition, recording devices were used for in-depth interviews.

Sexual behaviour, EVD awareness and other indicators were assessed through a questionnaire. After the interview, a counselling session was done to explain the use of a HIV/syphilis dual Rapid Diagnostic Test (RDT) and the consequences of a potential positive diagnosis, following the guidelines of the National Committee for the fight against AIDS (CNLS). The RDT was administered to all consenting EVD survivors and a sexual health promotion intervention strictly recalling WHO interim advice on EVD sexual transmission was given to everybody.14 Long-acting benzathine penicillin G for single intramuscular injection was prescribed to syphilis-positive EVD survivors. HIV- and syphilis-positive patients were referred to the closest hospital to access treatment.

Questionnaire

The questionnaire was interviewer-assisted. The part of the questionnaire on sex practices was completely based on UNAIDS questions to assess sexual behaviour.19 Safe sexual behaviour was defined as reporting regular condom use (using condom at each sexual intercourse) or sexual abstinence (>12 months). Only for EVD survivors, the following UNAIDS indicators were collected: (i) higher risk sex in last year (‘higher risk sex’ refers to sex with a casual partner compared with sex with primary sexual partner); (ii) condom use at last higher risk sex; and (iii) commercial sex in last year.

Concerning risk awareness of EVD sexual transmission, virologists, clinicians, epidemiologists and socioanthropologists participating in the study, with experience in HIV/AIDS and Ebola research in sub-Saharan Africa, agreed on six questions to assess awareness of risk of EVD sexual transmission with a score from 0 to 6 (Supplementary file). Young individuals who reported not having had first-time sex and old ones who reported having stopped sexual activity were excluded from the awareness analysis. This short questionnaire was validated using a pilot study: the six questions about Ebola virus persistence and the risk of sexual transmission were put to a small random sample of people (n=15) with diverse educational backgrounds, recruited through the Ebola associations on a voluntary basis, to validate the representativeness of the awareness score. These people were not included in the final study. The validation was done comparing the scores with the true knowledge of participants on the subject, in its turn assessed through in-depth interviews with all participants of the pilot study. Based on the result of the pilot study, the awareness score was then transformed into a binary variable (<4, not aware; >4, aware). Zone of residence (urban versus rural), region of residence, age, years in education and employment (this last divided into three categories based on the expected income: (i) lower income: housewife/farmer/labourer; (ii) higher income: State employee/business owner; and (iii) unpredictable income: other) were treated as potential confounders. Zone of residence was assessed by interviewers based on the address of residence declared by the study participants.

Statistical analysis

We used STATA 13.0 to perform statistical analyses. For binary and categorical variables, we reported the number of observations (n) and the percentage (%), while for continuous variables we reported the median and IQR. For comparison between groups, Pearson χ² test (or Fisher exact test if few observations) was used. For comparison between categories of the same variable, the Wald test was used. To obtain an estimation of the p value of the association of a categorical variable with a binary outcome, the Likelihood ratio test (LRT) was used, which compared the full model with a model without the exposure of interest. We used logistic regression to look at the association of sexual behaviour with study group and examine factors associated with awareness of the risk of sexual transmission of Ebola virus among survivors and their partners. To adjust for the study design, region of residence and zone of residence (urban versus rural) were forced variables of each model. Age, years in education and employment were treated as potential confounders. For logistic regression data reporting, we showed the number of individuals with outcome among the individuals with the exposure of interest (n/N). Interactions of region of residence with zone of residence (urban versus rural) as well as years in education with employment, region of residence and zone of residence were explored by LRT using final logistic regression models with the following outcomes: (i) sexual behaviour and (ii) risk awareness among EVD survivors. Since missing data were few (<2% for all variables used in the main analysis), they were excluded from the analyses.

Qualitative data collection and analysis

In-depth individual interviews were mainly used to clarify, confirm and follow up on key issues examined in the survey. In total, 25 EVD survivors, 25 partners and 10 individuals of the comparison group were interviewed in the five different prefectures of the study. Participants were chosen using the following criteria: time passed from ETU discharge for survivors (for half <8 months, for the other half >15 months); (i) younger age (18–45, because they were expected to be sexually active and more open to questions about sexual behaviour); and (ii) openness to answer questions about sex during the first questionnaire for all study participants. Thematic analysis was used to define the main clusters of concepts related to the knowledge of the disease, the awareness of risk of transmission of Ebola virus and the barriers to condom use for EVD survivors and partners. Interviews were done in French, Susu, Malinké, Pular and Konianké, depending on the preference of each participant.

In addition, 11 stakeholder interviews with health professionals (n=5) and NGO workers (n=6) were carried out to explore challenges faced when informing about the EVD transmission risk after recovery. Volunteer stakeholders were recruited in the local hospital of the prefectures visited during the study. Interviews were registered on recording devices and transcribed by two independent interviewers before analysis. Direct observations (eg, interactions between interviewer and interviewee) were used to contextualise the individual interviews.

Characteristics of the study participants

We invited 240 EVD male survivors to participate in the study. From 21 April to 23 June 2016, 234 of them were recruited in five prefectures (response rate 98%, figure 1). For all EVD survivors, the median duration between discharge from an ETU and study enrolment was 557 days (IQR 440–733). We asked the participants to bring their sexual partner(s) to the interview. Of them, 42 came alone, 136 (58%) came with one partner, 51 (22%) with two partners and 5 (2%) with three partners. All partners were females. Of the 256 recruited partners, 187 were married to their respective EVD survivor (73%). We also recruited 65 individuals, of which six were females (9%), for the comparison group. Nearly a third (30%) of the study participants were from Conakry, 32% from lower Guinea (Coyah and Forécariah) and 38% from forested Guinea (Guéckédou and Macenta), where the epidemic originated in December 2013.20 The overall median age was 28 (IQR 22–35). Age was similar between the comparison group (31, IQR 25–38) and survivors (32, IQR 25–41), but partners were younger (25, IQR 20–30, p<0.001, Pearson χ² test). Survivors were less educated than the comparison group (p=0.003, χ² test), while there was no difference in employment between the two groups (table 1).

Request permission

Sexual behaviour

In total, 145 (26%) participants reported safe sexual behaviour (38% of EVD survivors, 11% of partners and 14% of the comparison group). Of those, 81 reported regular condom use (22% of survivors, 9% of partners and 14% of the comparison group), while 42 reported sexual abstinence (15% of survivors, 2% of partners and none of the comparison group) (table 2). The inconsistency between sexual behaviour in survivors and partners is, at least to some extent, due to second and third partners (22% of all recruited partners) of the same EVD survivor.

Among EVD survivors, some UNAIDS sexual behaviour indicators were also estimated21: 37% of them had had higher risk sex in the last year (higher compared with sex with primary partner—that is, sex with a casual partner) and, amongst them, 27% reported having used a condom at last higher risk sex. Of those who last had unsafe higher risk sex, 9 (10%) had been discharged for less than a year, 47 (51%) for 1–1.5 years, 28 (30%) for 1.5–2 years and 8 (9%) for >2 years. 5% (12/233) of EVD survivors reported having had sex with a commercial sex worker (CSW) in the last year (table 2), of which one reported no condom use. Of those, three had been discharged from the ETU for more than 2 years, three for 1.5–2 years, five for 1–1.5 years and one for less than a year.

The prevalence of other sexually transmitted infections (STIs) among EVD survivors was low (1% for HIV type 1 and 1% for syphilis, none was HIV type 2-positive). Of the four STI-positive subjects (two for HIV type 1, two for syphilis), none reported sex with a CSW in the last year. All four reported having had unsafe sexual behaviour and no use of condom the last time they had sex with a casual partner.

Unmarried participants were more than twice as likely to have safe sexual behaviour than married ones (p=0.001, Wald test). EVD survivors were five times as likely to have safe sexual behaviour (aOR 5.59, 95% CI 2.36 to 13.2) than the comparison group (table 3). However, partners did not have safer sexual behaviour than the comparison group (p=0.73, Wald test). The difference in the behaviour of survivors and that of the comparison group was confirmed when regular condom use (instead of safe sexual behaviour, therefore excluding sexual abstinence) was used as the outcome (aOR 2.48, 95% CI 1.02 to 5.99, table 3). Participants from Conakry were more than three times as likely to use a condom regularly than those from forested Guinea (p=0.007, Wald test).

Awareness of risk of sexual transmission of Ebola virus

In total, 255 participants (46%) including 131 survivors (57%), 94 partners (37%) and 30 individuals in the comparison group (47%) were aware of the risk of EVD sexual transmission (table 2). Among those, 187 (75%) were aware that the virus can persist for 9 months or more in the semen, while 19% (n=48, of which 27 were survivors) thought the virus could not persist for more than a month. Those survivors were not those discharged from the ETU earlier in the outbreak (p=0.58), when the persistence of Ebola virus was not known potentially to last so long.

EVD survivors were almost three times as likely to be aware of the risk of EVD transmission after recovery than the comparison group (aOR 2.85, 95% CI 1.45 to 5.59, table 4). Among partners, 115 (51%) reported having been informed by their husband/boyfriend about the risks of EVD sexual transmission (Supplementary table 1). This was strongly associated with their awareness of EVD transmission potential after recovery (aOR 20.5, 95% CI 8.92 to 47.4). Age was associated with risk awareness among partners, middle-aged women (aged 25–39 years) being almost three times as likely to be aware of the risk compared with younger women (aged 15–24 years, p=0.027, LRT), but not among EVD survivors (p=0.23, LRT). EVD survivors who had >13 years of education were 10 times as likely to be aware of the disease transmission potential after recovery than those with no education (p<0.001, LRT). Similarly, those who were state employees or business owners were almost three times as likely to be aware of the risk compared with the lower-income work category consisting of labourers, farmers and housekeepers (p<0.001, LRT). EVD survivors from outside Conakry (from both lower or forested Guinea) were roughly 3–4 times as likely to be aware of the risk compared with those from Conakry (p=0.009, LRT, Supplementary tables 1 and 2).

Qualitative findings

We report here key findings that emerged from the qualitative analysis to help contextualise survey results: in-depth interviews confirmed that EVD survivors are reluctant to talk about viral persistence in their semen. Some EVD survivors had lost a wife/girlfriend from EVD and did not want to inform their new partner as they feared stigma and/or losing her. EVD survivors already experienced physical and emotional pain as well as deep stigmatisation. They perceived informing the partner about their potential infectiousness as being sick. Some were even forced to move to another community:

 "Within our community of origin, no head of the family would accept to give us his daughter in marriage, because they all associate us to Ebola".

Moreover, increasing scientific knowledge on the potential length of viral persistence led to contrasting recommendations: the length of the suggested period of sexual abstinence increased throughout the outbreak at release from ETU and during follow-up programmes. This created confusion among EVD survivors and some lost their faith in health authorities.

In-depth interviews also showed that the vast majority of the study participants associated condom use with infidelity. Other main barriers were lack of sexual pleasure, conception desire and lack of knowledge. The latter includes popular beliefs such as condom use leading to infection or condom possibly ‘exploding’ during sexual intercourse. Few participants reported refusing condom use because it was unacceptable for their faith.

Stakeholder interviews with health workers from public and NGO structures highlighted the challenge for health workers to communicate to a cured person the potentially long-lasting persistence of the virus.