Protective Behaviors Associated With Gender During the 2018-2020 Ebola Outbreak in Eastern Democratic Republic of the Congo

Key Points

Question  Are gender differences in Ebola virus disease (EVD)–related information, knowledge, and perceptions associated with differences in protective behaviors?

Findings  This survey study of 1395 randomly selected adults in Ebola-affected areas of eastern Democratic Republic of the Congo found that significant gender differences in preventive behavioral outcomes were mediated by gender differences in EVD knowledge and belief in rumors.

Meaning  This study’s findings suggest that addressing gender differences in access to preventive and protective measures of information, communication messaging, and community engagement could enhance the effectiveness of outbreak control interventions.

Importance  In 2018 to 2020, the Democratic Republic of the Congo experienced the world’s second largest Ebola virus disease (EVD) outbreak, killing 2290 individuals; women were disproportionately infected (57% of all cases) despite no evidence of differential biological EVD risk. Understanding how gender norms may influence exposure to EVD, intensity, and prognosis as well as personal protective behaviors against the virus is important to disease risk reduction and control interventions.

Objective  To assess whether men and women differ in personal protective behaviors (vaccine acceptance, health-seeking behaviors, physical distancing) and the mediating role of EVD information and knowledge, perceived disease risk, and social relations.

Design, Setting, and Participants  This cross-sectional, multistage cluster survey study of 1395 randomly selected adults was conducted in the Ebola-affected regions of North Kivu from April 20, 2019, to May 10, 2019. Path analyses were conducted using structural equation modeling to examine associations among study variables. Statistical analysis was conducted from August 2019 to May 2020.

Main Outcomes and Measures  The main behavioral outcomes of interest were (1) vaccine acceptance, (2) formal health care seeking, and (3) self-protective behaviors. The primary factor of interest was self-reported gender identity. We also assessed sociodemographic factors.

Results  Among the study’s 1395 participants, 1286 (93%) had Nande ethnicity and 698 (50%) were women; the mean (SD) age was 34.5 (13.1) years. Compared with female participants, male participants reported significantly higher levels of education, wealth, and mobile phone access. There were associations found between gender and all EVD preventive behavioral outcomes, with evidence for mediation through EVD knowledge and belief in rumors. Men reported greater EVD knowledge accuracy compared with women (mean [SE] score for men: 12.06 [0.13] vs women: 11.08 [0.16]; P < .001), and greater knowledge accuracy was associated with increases in vaccine acceptance (β = 0.37; P < .001), formal care seeking (β = 0.39; P < .001), and self-protective behaviors (β = 0.35; P < .001). Lower belief in rumors was associated with greater vaccine acceptance (β = −0.30; P < .001), and greater EVD information awareness was associated with increased adoption of self-protective behaviors (β = 0.23; P < .001).

Conclusions and Relevance  This survey study found gender differences in adopting preventive protective behaviors against EVD. These findings suggest that it is critical to design gender-sensitive communication and vaccination strategies, while engaging women and their community as a whole in any response to infectious disease outbreaks. Research on the potential link between gender and sociodemographics factors associated with disease risk and outcomes is needed.

From August 2018 to June 25, 2020, the world’s second largest Ebola virus disease (EVD) outbreak, and the tenth outbreak to occur in the Democratic Republic of the Congo (DRC), infected 3481 individuals and claimed 2290 lives. Of the infected individuals, 57% (n = 1970) were women.¹ Because there is no evidence for a biologically driven difference in EVD risk by sex, the disproportionate number of women affected has been attributed to gendered norms and behaviors related to caregiving and decision-making roles within a household.² There is, however, a paucity of gender-disaggregated data from which to examine EVD risk. Gender norms may influence exposure to EVD, intensity, and prognosis as well as personal protective behaviors against the virus. Risk reduction and control interventions need to account for mechanisms by which gender is associated with EVD transmission and outcomes.^2,3

Individual self-protective and care-seeking behaviors can reduce the risk and transmission of EVD, including (1) preventive behaviors such as avoiding physical contact with those who are sick or dead, (2) agreeing to vaccination, and (3) seeking care when experiencing EVD-compatible symptoms. Risk-reductive behaviors can be influenced by a supportive environment that ensures access to EVD vaccination and high quality and safe medical care, EVD treatment centers, and accurate and timely information about the outbreak and containment measures.^4-6

The mechanisms by which gender is associated with attitudes, perceptions, and preventive behaviors during epidemic outbreaks globally are under-researched. The present study aimed to characterize the role of self-reported gender identity in perceptions, attitudes, and behaviors regarding EVD during the 2018 to 2020 EVD outbreak in the eastern DRC, using detailed survey data collected from a high transmission region during the outbreak. We hypothesized that men and women differ in personal protective behaviors, including vaccine acceptance, health-seeking behaviors, and physical or social distancing, and that these differences are mediated by gender differences in levels of EVD information and knowledge, perceived disease risk, and social relations.

This survey study was approved by the Massachusetts General Brigham Human Research Committee and ad hoc ethic committees at the Research Center on Democracy and Development in Africa, Free University of the Great Lakes Countries in the DRC. Oral rather than written consent was used to facilitate comprehension by participants, build trust, and reduce the risk of collecting identifiable information.

Study design and data collection have been described previously⁷ and are summarized here, following the American Association for Public Opinion Research (AAPOR) reporting guideline. Cross-sectional survey data were collected in EVD-affected areas in North Kivu province of eastern DRC from April 20, 2019, to May 10, 2019, a period of high EVD transmission. We used a multistage cluster random-sampling approach to identify participants. First, using comprehensive lists, we randomly selected 30 avenues or cells (clusters) per city, the lowest administrative unit. Then, 16 households with at base 1 adult participant (aged at least 18 years) per household were randomly selected per cluster. The sample size was calculated to estimate proportions on a few key perception and behavior indicators in the given population for a 95% CI and 10% precision. We used a 0.5 proportion estimate, and the sample size was multiplied by 2 for comparison by sex, as well as by the number of cities, and to account for an estimated design effect of 2. The sample size was increased by 20% to account for nonresponse and logistical constraints. Surveys were administered at the time of recruitment by a team of experienced university researchers trained in the conduct of interviews. If selected respondents were absent, 3 attempts were made to contact them over the course of the survey. To facilitate interviews, female and male interviewers were assigned to same sex respondents, providing equal representation in the sample. Participation was anonymous, voluntary, and without compensation. Interviewers were trained on preventive measures to reduce risk of exposure, and all households were provided printed information on transmission prevention and avoidance at the end of the questionnaire.

The standardized, structured questionnaire was developed in French, and translated into Swahili. Independent experts reviewed and validated the translation, and local experts established face validity. Pilot interviews were conducted to test and validate the questionnaire. Interviewers were experienced in community survey methods and participated in a 1-week training course on the questionnaire content and sampling protocol. Survey data were collected electronically using KoboToolbox (Harvard Humanitarian Initiative).

The 3 main behavioral outcomes of interest were (1) EVD vaccine acceptance, (2) formal EVD health care seeking, and (3) EVD self-protective behaviors. A vaccine acceptance score was created by summing responses on 2 dichotomous survey items: belief that the vaccine was safe and acceptance of the vaccine, if offered. EVD health care–seeking was measured with an open-ended question that asked where, if at all, respondents would seek care if they suspected they had EVD. The 4 response categories were as follows: (1) formal EVD-equipped health services, including hospitals, health centers, and EVD treatment units; (2) formal but non-EVD-equipped services, including pharmacy or community health workers; (3) informal services, including traditional healers, religious leaders, friends, and family; and (4) no health care–seeking. EVD self-protection was computed as a sum score of key self-protective behaviors that participants reported engaging in since the outbreak began. These self-protective behaviors were represented by 16 dichotomous items relating to direct avoidance of EVD cases, targeted social distancing, general physical and social distancing, and hand hygiene. A complete listing of all behaviors is shown in Table 1.

Our primary risk factor of interest was self-reported gender identity (men, women). Gender is used rather than biological sex to highlight the contextual factors related to gender norms and values that are associated with health behaviors. We also assessed the following sociodemographic factors: age group (18 to 30 years, 31 to 45 years, and at least 46 years), education (none, primary completed, secondary completed), wealth index (none, poorest, poor, and rich), ownership of mobile phone (yes, no), and ethnicity (Nande, other).

We constructed the following 5 potential mediators of the association between gender and the outcomes by combining the scores of individual items with similar themes: (1) EVD information awareness score, (2) EVD knowledge accuracy score, (3) EVD risk-perception score, (4) belief in rumors score, and (5) social relations score. The EVD information awareness score was quantified by the extent participants reported receiving information on EVD-related topics by summing responses on 6 binary (yes, no) items related to EVD prevention, symptoms, knowledge of where to seek health care, knowledge of actions to take if a relative or neighbor has EVD, updates on EVD in the province, and overall EVD response. The EVD knowledge accuracy score assessed knowledge of EVD symptoms, transmission modes, and causes. The score was computed as the total number of correct answers. The EVD risk-perception score measured perceived risk of contracting EVD within 30 days following survey administration by combining participant responses into 4 risk categories (improbable, undefined, probable to others, probable to self and relatives). The belief-in-rumors sum score combined binary items assessing whether participants believed that EVD was real, if it was invented by authorities for financial or other gains, or if it was invented to destabilize the region, which were commonly cited rumors in the DRC. The social relations score was a summary score comprising items on perceived quality of social relationships with family members, neighbors, and members of the same or other ethnic groups on a 5-point Likert scale from very bad to very good.

All analyses were conducted using the complex sample module in SPSS version 25 (IBM Corp) and survey data analyses in Stata version 16 (StataCorp) from August 2019 to May 2020. Data were weighted to reflect the unequal probability of sampling between cities, using the best available population estimates for the 3 cities. Frequencies, percentages, and statistically significant χ² tests for gender differences are reported for categorical variables; weighted means, standard errors, and 95% CIs are reported for continuous variables (Table 1 and Table 2). We evaluated associations between mediator and outcome variables to assess collinearity (Table 3). Testing was 2-sided, and the threshold for statistical significance was P < .05.

We conducted path analyses using a structural equation modeling framework to examine associations among variables. We used the full information maximum likelihood (FIML) estimator to account for item-specific nonreponse (refusal) found in 70 (5.0%) of the questionnaires; survey weights to account for sampling probability; and robust standard errors to deal with non-normality of the data after the null hypothesis for multivariate normality was rejected using the Doornik-Hansen test. We calculated goodness-of-fit indices to assess model fit: root mean square error of approximation (RMSEA) (best if less than 0.06), comparative fit index (CFI) (best if greater than 0.90), and the coefficient of determination (CD) (best if greater than 0.08).⁸ Variables of interest were selected based on theoretical considerations and previous studies.^2,3,7,9,10 All pathways between gender, mediators, and outcomes were included in the final model, and pathways between demographic covariates and mediators were finalized based on exploratory stepwise regression and modification indices.

We examined associations (eFigure in the Supplement) between the key risk factor (gender), mediators (EVD information awareness, EVD knowledge accuracy, EVD risk-perception, belief in rumors, and social relations), and EVD outcomes (vaccine acceptance, formal care-seeking, and self-protective behaviors). We assessed the association of gender with each variable, along with associations of all risk factors and mediators with each outcome. We tested for mediation associations by examining indirect pathways between gender and each outcome through the mediator variables, including location, education, and wealth. Covariances were freely estimated among all risk factors, as well as between error terms.

A total of 1419 households were approached, among which 8 (0.6%) had no available participants and 16 (1.1%) refused participation. Individual participants were approached in 1395 households (98.3%). Among the selected households, a total of 1420 individuals were approached, among who 6 were not available after multiple attempts, and 19 refused to participate.

The final sample included 1395 participants, all with completed interviews, residing in the cities of Beni, Butembo, and Katwa in eastern DRC. Participants’ mean (SD) age was 34.5 (13.1) years. For the sociodemographic risk factors, men reported higher levels of educational attainment and mobile phone ownership (Table 1). We found significant differences in topics of EVD-related information received by men and women (Table 1). More women than men reported receiving EVD information about cases in the province (612 women [91%] vs 526 men [77%]; χ² = 49.56; P < .001), response efforts (610 women [90%] vs 529 men [76%]; χ² = 44.74; P < .001), where to seek care (612 women [90%] vs 587 men [86%]; χ² = 9.10; P = .003), and dealing with EVD cases (593 women [87%] vs 567 men [82%]; χ² = 7.88; P = .005). Women were more likely than men to believe EVD rumors (eg, 226 women [34%] believed that EVD does not exist vs 131 men [20%]; χ² = 36.73; P < .001).

More men than women thought that vaccines could prevent and cure EVD (422 men [59%] vs 335 women [48%]; χ² = 18.88; P < .001) and were more willing to accept the vaccine (312 women [43%] vs 258 men [37%]; χ² = 7.11; P = .008) (Table 1). Men were more likely to seek formal health care for EVD (616 men [89%] vs 493 women [71%]; χ² = 106.52; P < .001). Men reported engaging in more direct avoidance behaviors for EVD self-protection (eg, avoid contact with people suspected of recent contact with someone infected by EVD: 521 men [75%] vs 375 women [56%]; χ² = 61.38; P < .001), whereas women reported engaging in more general physical and social distancing such as physical interaction avoidance (eg, reduce physical interactions with relatives: 318 women [46%] vs 188 men [24%]; χ² = 57.09; P < .001), public space avoidance (151 women [22%] vs 16 men [2%]; χ² = 127.30; P < .001), and funeral avoidance (206 women [30%] vs 176 men [25%]; χ² = 4.11; P = .04). Women reported more handwashing compared with men (544 women [81%] vs 482 men [71%]; χ² = 20.22; P < .001).

We found women, compared with men, had higher mean (SE) scores for reported EVD information awareness (women: 5.42 [0.06] vs men: 5.02 [0.07]; P < .001), belief in rumors (women: 1.27 [0.05] vs men: 0.97 [0.05]; P < .001), quality of social relations (women: 11.19 [0.09] vs men: 10.04 [0.08]; P < .001), and self-protective behaviors (women: 5.78 [0.14] vs men: 5.33 [0.09]; P = .006) in the overall sample (Table 2). Men, compared with women, displayed higher EVD knowledge accuracy (men: 12.06 [0.13] vs women: 11.08 [0.16]; P < .001), risk perception (men: 1.23 [0.04] vs women: 1.15 [0.13]; P = .04), EVD vaccine acceptance (men: 1.02 [0.03] vs women: 0.85 [0.04]; P = .001), and formal health care seeking (men: 2.74 [0.03] vs women: 2.22 [0.05]; P < .001).

The estimated direct and indirect associations from the path analyses are presented in Table 4, and pathways are depicted in the Figure. Model fit statistics indicated adequate fit for the path analysis model (RMSEA = 0.08; CFI = 0.92; CD = 0.26) and additional fit indices are reported in the eTable in the Supplement.

Higher EVD knowledge accuracy was associated with increases in all 3 EVD outcomes: vaccine acceptance (β = 0.37; P < .001), formal care seeking (β = 0.39; P < .001), and self-protective behaviors (β = 0.35; P < .001). Lower believe in EVD rumors was associated with higher EVD vaccine acceptance (β = −0.30; P < .001) and greater formal care seeking for EVD (β = −0.09; P = .007); and better social relations was also associated with higher EVD vaccine acceptance (β = 0.11; P < .001) and greater formal care seeking for EVD (β = 0.09; P < .001). Engagement in EVD self-protective behaviors was also associated with higher EVD information awareness (β = 0.23; P < .001), and vaccine acceptance was also associated with higher EVD risk perception (β = 0.05; P = .04).

Women reported engaging in more self-protective behaviors. Women also reported higher EVD information awareness, greater belief in EVD rumors, and better social relationships. Men reported more formal care-seeking and registered higher EVD knowledge scores.

We identified an indirect association for men having higher EVD vaccine acceptance and seeking more formal modes of EVD care than women, with mediation for both outcomes through higher EVD knowledge accuracy, lower belief in EVD rumors, and despite lower quality social relations.

The tenth and largest EVD epidemic in the DRC had a devastating impact on a region experiencing protracted armed conflict and political crises for more than a decade. This study examines the pathways by which gender differences in EVD-related information, knowledge, and perceptions are associated with risk-related attitudes and behaviors. We analyzed data collected during the EVD outbreak with the aim of better understanding social and behavioral factors associated with EVD-related gender disparities.

Women reported higher EVD information awareness but also greater belief in EVD rumors than men. This is consistent with previous findings from Ebola-affected areas conducted during the beginning of the outbreak.⁷ Conversely, men reported more accurate knowledge of EVD symptoms, causes, and transmission compared with women. Unpublished qualitative interviews found that most of the first waves of EVD-related training and communication implemented by the government and responding organizations were targeted at men due to their greater access to formal communication channels.

Greater EVD knowledge accuracy was, in turn, associated with increases in all 3 EVD preventive behavioral outcomes: vaccine acceptance, formal care seeking, and self-protective behaviors. Lower belief in EVD rumors was associated with greater vaccine acceptance, and greater EVD information awareness was associated with increased adoption of self-protective behaviors. These findings reinforced the important role of accurate and targeted information in outbreak control and prevention.^7,11-13

We found associations between gender and the EVD preventive behavioral outcomes, as well as evidence for mediation of the gender associations through the pathways of EVD knowledge and belief in rumors. Men reported overall higher EVD vaccine acceptance and formal EVD health care seeking, which was mediated by their higher reported EVD knowledge accuracy and lower belief in rumors (the strongest factors associated with vaccine acceptance) compared with women. Nevertheless, women reported engaging in a greater number of self-protective behaviors compared with men.

These gender differences reinforce previous findings on the role of greater trust and reduced misinformation in vaccine acceptance.^14,15 The findings suggest that increasing vaccine acceptance requires engaging in transparent and targeted communication to support individuals in making informed decisions rather than pressuring people to receive vaccines. Women may have been less informed about vaccines because communication materials about vaccines were not available in the appropriate format. Early decisions to exclude pregnant and breast-feeding women from accessing vaccines, with limited communication, and ultimately a policy reversal created misunderstanding and suspicion among women.^16,17

There are several potential barriers to care-seeking in this population. Free health care appears to have encouraged care-seeking.¹⁸ Women, however, may have been more likely than men to avoid formal health care facilities.¹⁹ We found that greater perceived EVD risk and higher EVD knowledge accuracy were the most robust factors associated with formal health care seeking for EVD symptoms. This supports qualitative research from the West Africa EVD outbreaks that found fear of the disease and misinformation as the biggest barriers to treatment seeking.^6,20 Higher EVD knowledge accuracy and greater acceptance of formal health care services among men, compared with women, is consistent with research describing access to accurate EVD information for women.²¹ Mobility for treatment-seeking among women may further be reduced in areas of armed conflict and heightened insecurity, due to gender-based violence and attacks on health facilities and health care workers.²²

Greater EVD information and knowledge was found to be associated with the adoption of more self-protective behaviors. Cultural and social norms indicate that women are more likely to be caregivers, prepare food, and have greater contact with dead bodies during traditional burial practices.^2,3,23 This may explain why women engaged more frequently than men in self-protective behaviors.

Calls to integrate women and gender-related concerns into the response to disease outbreaks have emerged,^9,24,25 echoing broader issues on the provision of care, accessibility, and reporting of gender-disaggregated data^9,24 and highlighting the need to address male bias in policy and planning.⁹ The response in DRC, apart from community dialogues for women, did not sufficiently target nor engage women, nor did it adapt the medium of its messaging to facilitate improved communication with women. In future responses to infectious disease outbreaks such as COVID-19, it is critical to engage women in the response and communication strategy.

This study has some limitations. First, the study is limited to urban areas in eastern DRC and may not be generalizable to rural or culturally distinct areas. Although our analyses adjusted for key factors that may be associated with gender, social, and behavioral EVD outcomes, the possibility of unmeasured confounders cannot be ruled out. These may include women’s caregiving roles, gender norms, mobility and access to care, and involvement in funeral practices, among others. Additionally, the data were self-reported and may be subject to social desirability, information, and selection biases, which should be taken into consideration when interpreting the results. Interviews were conducted in Swahili and/or French, which are the commonly spoken languages in the urban area under study, but not necessarily the preferred language of respondents. The instrument and training of interviewers were designed to reduce risks of bias. Since the variables under study relate to perceptions and attitudes, self-reporting was considered an appropriate data collection method.

The study data were collected at the peak of the EVD outbreak in DRC using a rigorous multistage cluster random-sampling process. The quantitative analysis used a structural equation modeling framework because of its ability to simultaneously assess multiple direct and indirect associations for a range of variables, to compare statistical fit across models, and to establish and test a comprehensive risk pathway framework for health behaviors.

The results provide insights into the sociocultural dimensions of gender vulnerabilities and EVD risk, while also examining the mediating role of EVD-related information, knowledge, and perceptions in the association between gender and EVD preventive behaviors. Self-protective behaviors including vaccine acceptance are the cornerstone of EVD control, especially given limited therapeutic options. Understanding gender-related factors associated with protective behaviors can help to shape outbreak control interventions. Our findings contribute to a better understanding of the role of gender in EVD control interventions.

Accepted for Publication: December 13, 2021.

Published: February 16, 2022. doi:10.1001/jamanetworkopen.2021.47462

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Pham PN et al. JAMA Network Open.

Corresponding Author: Phuong N. Pham, PhD, 14 Story St, 2nd Flr, Cambridge, MA 02139 (ppham@hsph.harvard.edu).

Author Contributions: Pham, Sharma, and Vinck had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Pham, Sharma, Zikomangane, Vinck.

Acquisition, analysis, or interpretation of data: Pham, Sharma, Bindu, Nethery, Nilles, Vinck.

Drafting of the manuscript: Pham, Sharma, Vinck.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Pham, Sharma, Nethery, Vinck.

Obtained funding: Pham, Vinck.

Administrative, technical, or material support: Pham, Bindu, Zikomangane, Vinck.

Supervision: Pham, Bindu, Vinck.

Conflict of Interest Disclosures: Dr Nethery reported receiving grants from the National Institutes of Health outside the submitted work. No other disclosures were reported.

Funding/Support: The Enhanced Learning and Research for Humanitarian Assistance (ELRHA) and United Nations Development Programme (UNDP) provided funding for the study.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We are grateful for the team of data collectors, supervisors and all whose who participated in the study. We dedicate the paper in memory of our dear colleagues and friends, Professor Emile Okitolonda Wemakoy, Professor Patrick Kayembe, and Jean-Paul Zibika, who all contributed to this study. Additional non-author contributors include Kanuni Buroko, George Nyembo Bilauri, and Salomon Bagbuh.

Additional Information: The instrument and individual participant data that underlie the results reported in this study can be made available for replication purposes upon reasonable request to the corresponding author and after consultation with the corresponding author’s internal institutional review board.