key: cord-1018281-gx28rm0v
authors: Brandt, Amelia J.; Katalenich, Bonnie; Seal, David W.
title: Qualitative Review of Organizational Responses to Rumors in the 2014–2016 Ebola Virus Disease Outbreak in Liberia and Sierra Leone
date: 2021-09-30
journal: Glob Health Sci Pract
DOI: 10.9745/ghsp-d-21-00203
sha: d27a2bb83ed87a93e2c0e0d40f49164d98fd3347
doc_id: 1018281
cord_uid: gx28rm0v

INTRODUCTION: Rumors and misperceptions were a persistent challenge in the response to the 2014–2016 Ebola Virus Disease (EVD) outbreak in West Africa. This study aimed to document organizational approaches to identifying and addressing rumors and provide practical recommendations for future outbreaks. METHODS: We conducted semistructured interviews with 34 individuals who participated in the EVD response in Liberia and/or Sierra Leone. Interviews focused on the general organizational approach and organizational response to specific rumors. Interviews were recorded and transcribed verbatim. RESULTS: Most respondents reported that rumors were considered an organizational priority and their importance increased over time. Formal rumor identification systems using community-level reporters were described in Liberia and Sierra Leone as well as varied informal systems. A wide range of approaches was used to address rumors including Community Led Ebola Action, Community Led Total Sanitation, drama performances, Ebola Treatment Center/Unit-based approaches, radio, leveraging community leaders as an information source, and organization change. Interpersonal and community-led approaches were described most often. Staff whose professional roles did not involve rumor management reported informally addressing rumors with colleagues and beneficiaries. Rumors reflecting valid concerns with the EVD response, such as potential infection in health care facilities, were addressed through organizational change and improvement. DISCUSSION: Interpersonal and community-led approaches were considered effective by participants and hold promise for future outbreaks. Informal systems developed at Ebola Treatment Centers/Units highlighted how these facilities may be utilized as an information hub. Professionals who interact with beneficiaries, especially local staff, are likely to address rumors informally and organizations may benefit from considering local staff an asset in rumor management. Rumors alerted responders to issues in the EVD response, but this may not be the most efficient mechanism to receive and address concerns.

T he 2014-2016 Ebola Virus Disease (EVD) outbreak in West Africa was devastating. In Guinea, Liberia, and Sierra Leone there were 28,616 confirmed, probable, and suspected EVD cases and 11,310 deaths. 1 Although numerous risk and emergency communication manuals, guidance, and trainings existed at the start of the outbreak, persistent communication challenges highlighted gaps in these approaches. 2 A coordinated response to the outbreak did not begin until approximately 6 months after the first human contact with the Ebola virus 3 and inadequate and inappropriate communication early in the outbreak proved counterproductive. [4] [5] [6] Early messaging, for example, overemphasized the importance of bushmeat in transmission while underemphasizing the comparatively higher risk of human-to-human transmission. 7, 8 Rumors, misperceptions, and community resistance presented significant obstacles in controlling EVD spread. 4, 5 Rumors, defined as "unverified and instrumentally relevant information statements in circulation that arise in contexts of ambiguity, danger, and potential threat," 9 can be problematic, but are an expected and adaptive response to frightening and ambiguous situations. 10 Information is necessary for planned action and when it is unavailable or not trusted, people may come together to pool information in an attempt to develop a reasonable understanding of a situation. [9] [10] [11] Rumors are not traditionally viewed as an information source but do reflect fears, hopes, and concerns of the populations in which they circulate. As such, they can be an invaluable resource for informing communication approaches and outbreak response.

This study aimed to capture the range of approaches used to identify and address rumors during the 2014-2016 EVD outbreak in Liberia and Sierra Leone. Guinea was excluded because the interview guide drew on rumors previously identified in Liberia and Sierra Leone, and rumor data were not available for Guinea.

Although program descriptions describing formal organizational responses to rumors exist, 4, 5, 12, 13 this study sought to describe both informal and formal approaches used by EVD responders in a variety of roles, locations, and organizations to identify and address rumors to provide practical recommendations for rumor identification and management in future outbreaks. Rumors continue to be of interest in public health as illustrated by their impact on the ongoing COVID-19 pandemic. 14, 15 To better understand how to identify and manage rumors, it is important to identify lessons learned from previous outbreaks and encourage innovation. In particular, studying informal rumor identification and management techniques practiced by EVD responders with direct community contact can provide unique insights into bottom-up approaches to identifying and addressing rumors.

This was a cross-sectional qualitative study.

The study was reviewed and approved by the Tulane University Social-Behavioral Institutional Review Board. Participants provided oral informed consent.

Participants were recruited via Facebook and Twitter posts on the lead author's personal pages, the lead author's professional networks, and the CORE Group, Global Alliance for Nursing and Midwifery, and Healthcare Information for All listservs. Snowball sampling was used to reach the target sample size of 30 to 50 or until saturation was reached.

Thirty-four individuals who participated in the EVD response in a professional capacity in Liberia or Sierra Leone participated in the study. Of these, 16 participants worked in Sierra Leone, 14 worked in Liberia, and 4 worked in both countries. Twenty participants were international staff and 14 were local staff. Individuals aged younger than 18 years and/or unable to communicate in English were excluded.

First, participants completed a short survey about the organizations they worked with during the EVD response. The survey was available on the Health Insurance Portability and Accountability Act (HIPAA)-compliant Johns Hopkins University instance of Qualtrics.

Then, participants completed semistructured interviews on organizational response to common rumors about EVD. Interviews were conducted in person, by phone, or via their preferred digital platform (e.g., Skype, WhatsApp, Zoom). Interviews conducted via Skype and Zoom were audio-recorded using platforms' integrated recording capacity. All other interviews were recorded via QuickTime Player version 10.5 on a laptop computer. Audio recordings were labeled with a participant code and transcribed verbatim.

Data collection took place between September 24, 2019, and June 24, 2020.

Survey data were downloaded from Qualtrics and analyzed in Microsoft Excel. Interview transcripts were analyzed in Dedoose version 8.3.21 using individual-level thematic content analysis based upon an a priori analysis plan (Table 1) coders reviewed the first 20 interview transcripts to develop a codebook based on emerging themes. Each transcript was coded by 1 coder and reviewed by a second coder who identified code discordance. Coders discussed discordant codes until reaching consensus. Participants were not provided with a rumor definition but were rather asked how they defined rumors. This was an intentional approach to allow participants to describe their experience based on their own understanding of rumors and how they were addressed.

Of participants who completed the organizational characteristics survey (n=30), 17 worked for 1 organization during the outbreak, 11 worked for 2 organizations, and 2 worked for 3 or more organizations. In total, participants reported working at 40 organizations ( Table 2) . The professional and educational backgrounds of participants varied and included international development, emergency response, public health, and clinical care, with international development and public health being the most common. Many participants had experience in multiple areas.

At an organizational level, the importance of identifying and addressing rumors varied. Some participants reported a high-priority level but also described how the priority level varied between departments. Several participants indicated that rumors were a low priority in their organizations. Increasing priority of rumors over time emerged as a common theme. Table 3 presents illustrative quotes regarding rumor priority.

Of the 34 participants, 13 described a formal rumor identification system. Formal systems described in Liberia and Sierra Leone consisted of a national network of community-level individuals reporting rumors via mobile phone. Additionally, a few participants also mentioned formal systems reviewing mass media to identify rumors. However, although the formal systems were national, several participants were unaware of their existence.

Many participants discussed how organizations identified rumors with informal or ad hoc methods, most often through in-person communication between staff and community members. In some cases, staff members who were told about rumors worked in social mobilization or communication roles, although in some cases local staff were asked about rumors even if their professional role did not include any formal communication responsibilities. For example, a local staff working with an international nongovernmental organization focused on creating business opportunities described how he was frequently asked about rumors. Some participants also described other systems used to identify rumors of EVD cases but not other types of rumors. Table 4 presents illustrative quotes regarding the rumor identification processes.

Participants described a wide range of rumor management approaches using interpersonal communication and mass media channels. Participants emphasized that using multiple communication channels was vital to ensure that people in remote or isolated areas were reached, as well as the importance of different complementary approaches (e.g., radio, community meetings, drama) that allowed for wide distribution of information as well as gave people opportunities to ask questions and raise concerns.

Interpersonal communication approaches were discussed most often and were generally implemented at the community level. In Sierra Leone, one such approach was a participatory methodology called Community-Led Ebola Action. One participant described how this approach was used to address fears and rumors about personal protective equipment and ambulances. Participants described a similar approach that was used country-wide and coordinated through a consortium. Another international staff member in Liberia discussed how Community-Led Total Sanitation, the approach on which Community-Led Ebola Action was based, was adapted for community engagement in Lofa County, Liberia. Community-Led Ebola Action 16 and Community-Led Total Sanitation 17,18 have previously been described in detail in the literature. These approaches emphasize the capacity of communities to develop their own solutions to challenges that meet scientific standards for outbreak control as well as the importance of demystifying situations and objects that cause fear. Another community engagement approach used musical and drama performances. Participants also discussed the importance of listening to questions and concerns from the community, gathering accurate information, and sharing that information with the community to close the communication loop.

Informal interpersonal communication efforts were also frequently discussed, especially those that took place in the Ebola treatment center/unit (ETC/ETU) context. Another international staff member working at an ETC/ETU described a more formal approach to using downtime to provide education and reduce misinformation. Although this approach used interpersonal communication to start, it later expanded to mass media. These approaches illustrated the potential for the ETC/ETU to be a setting for information sharing and addressing rumors.

Radio was the most common medium for mass media approaches, but billboards, posters, and social media were also mentioned. Interactive radio programs were generally seen to be the most effective mass media efforts. One international staff member working in Liberia and Sierra Leone and described other radio programming and efforts to work with journalists to reduce and address rumors.

Consistent messaging was discussed as a key attribute to both mass media and interpersonal communication approaches. In Sierra Leone, participants described a standard messaging guide that was produced by a Social Mobilization Action Consortium, while in Liberia participants reported that official messaging was centrally approved. Although consistent messaging was considered important to effective communication, several participants also discussed how initial messaging about the high likelihood of death from EVD was actually counterproductive and caused people to hide EVD cases. This illustrated the importance of considering the potential unintended effects of messaging. A common theme in participants' discussions of both interpersonal and mass media rumor management approaches was the importance of delivering information through trusted sources. The first step described for identifying trusted sources in communities was entering a community in a culturally appropriate way. One international staff member working in Sierra Leone discussed the need to be cautious in entering communities; approaches that worked in some communities may not work in others. After entering a community, participants discussed approaches to identifying trusted community members and then using those people to share information as community members may be more likely to accept information from a trusted source.

While the majority of rumor management approaches discussed were focused on communication, several participants also provided examples of organizational changes. For example, several participants discussed how family access to the ETC/ETU was increased over time and decreased rumors. Some organizational changes were made in direct response to a rumor. For example, bags used for safe burials were modified to allow family and friends to see the face of the deceased in response to rumors that these bags were filled with rocks, which illustrated the importance of transparency in reducing rumors.

Several participants discussed how traditional beliefs could affect communication and rumor management. One international staff member working in Sierra Leone described how this appeared to contribute to the persistence of the rumor that EVD was not real. In contrast, a local staff member also working in Sierra Leone described how they took traditional beliefs into account to lead to desired health behaviors. This contrasting Another theme that emerged in several interviews was how to address rumors that the participants felt were based on truth, at least in part. Participants mentioned rumors of people profiting from EVD, chlorine being dangerous, nepotism in ETC/ETU hiring, and health workers transmitting EVD as having some basis in truth. Participants generally indicated that when addressing these rumors the best approach was to acknowledge the element of truth. Table 5 presents interview quotations for emergent rumor management themes and summarizes the key insights for rumor management approaches.

A common rumor in Sierra Leone was that health workers were spreading EVD. Participants mentioned identifying the rumor in several different ways.

Five of the 9 participants who discussed this rumor believed that there was some truth to this rumor, but several believed the fears may have been overstated. Despite participants believing that there was some element of truth to the rumor, they also recognized that it was very damaging and led to the stigmatization of health workers and decreased health-seeking behavior.

The primary approach to addressing this rumor was organizational change and improvement of infection prevention and control procedures. As 1 participant put it:

You can't message your way out of something like that. -International staff, Sierra Leone

Interpersonal communication approaches and mass media campaigns were also used to complement organization change efforts. Table 6 presents illustrative quotes describing the organizational response to this rumor.

Rumor identification systems using mobile phones and large networks of social mobilizers appeared to be effective for gathering rumors, but interviews seemed to indicate that there was limited awareness of these systems. This indicates a need for improved coordination and communication about rumor identification and management efforts. It is also worth considering how information from rumor identification systems can be used to improve outbreak response. While rumors and misinformation are often considered within the purview of communications, rumors are also an important information source regarding how operations can be improved. This was illustrated by the operational changes that several organizations made in response to rumors in the community.

It also was a consistent theme that EVD responders heard and responded to rumors even if this was not part of their professional role. This was especially common among local staff. This presents a potential risk as staff members may respond in a way that exacerbates a negative rumor but also presents an opportunity. Organizations can use their internal communication structures to encourage staff members to report these rumors and provide the necessary information and skills to respond to them. For example, it could be beneficial to allocate a portion of time during staff meetings to discuss questions, concerns, and rumors that staff are hearing from the communities they work with. Alternatively, an organization could appoint a focal person to which that community feedback could be directed. Social and behavior change training (effective communication, facilitation, negotiated behavior change, etc.) could be beneficial to prepare staff members to respond to rumors. These efforts should take into account existing workloads and be cognizant of situations where this would cause an undue burden on staff members.

The use of mass media messaging and interpersonal communication at a community level to address rumors is not a new idea. Many EVD responders discussed how the 2014-2016 EVD outbreak was a learning experience and that both mass media and interpersonal communication approaches steadily improved as the outbreak went on. However, this type of approach may underutilize rumors as an information source. The results of this study illustrate how rumors highlighted issues within the response that were later addressed, but this process was ad hoc as there was not a structure in place for community members to share rumors or feedback directly with those who could address their concerns.

Despite this challenge, numerous examples of how community feedback was used to improve the outbreak response were described. One of the most striking examples was the rumor that health workers were spreading EVD. This rumor was perceived by several EVD responders as having some basis in truth and is supported by evidence of nosocomial transmission during the outbreak in Sierra Leone. 19, 20 As such, the primary response to this rumor was no communication or messaging, but rather improving infection prevention and control A common rumor in Sierra Leone was that health workers were spreading EVD.

Organizations can use their internal communication structures to encourage staff members to report rumors and provide the necessary information and skills to respond to them. Use of multiple approaches and channels I think what is important is having a combination of all these channels so that you're sure you're reaching people even the remotest areas of the country. So although you know committee meetings are better because you can have conversations, people can ask questions, and you respond but in terms of creating awareness and making sure you reach as many people as possible, I think a combination of all these channels was quite important. Because there's also posters that are put out there, so for those that way are able to read in order to get that information and based on what they've seen on posters, for example, they, during committee meetings, they could go ask more questions or based on what they would have heard on the radio, when they come face-to-face with a health worker, they have an opportunity to ask questions. -International staff, Sierra Leone Importance of using multiple complementary channels to disseminate information

You have the community champions. You have those to follow up so they were also used because the mobilizers during those triggering were able to listen to the communities to get the community perspective about how they could also protect themselves and keep themselves safe because honestly speaking I think most times we think the communities do not have . . . a scientific explanation to how they are able to prevent or protect themselves against diseases like Ebola and that. But if you engage the communities, really and you sit with them you discuss then you realize that they also have some explanations and some ideas on how they would be able to do it and that meets scientific standards. 

[It] was a like a way of bringing community members together in groups and addressing like their fear and their perceptions and their experiences and stories they'd heard and all those kind of like listening to all of that and then giving them the actual information and kind of addressing directly those, you know, maybe misconceptions that they had or clarifying if they were correct ones and then helping them to figure out what they need to do.

It is important to listen to community questions and feedback, gather appropriate information, and communicate back with the community to close the loop.

We didn't just go and perform but we created a stage that will reduce the number of rumor because people ask questions that they feel somebody told them, and we wrote those questions down and we communicate[d] their back with our sponsor, like UNICEF, right . . . and they will send the real information. Radio I think in terms of addressing rumors, the best way was usually through interactive programs. So whether you have an interactive program on radio, TV, or at a community level, programs where people can ask questions and answers are provided. I think those were the best in terms of addressing rumors, because sometimes the rumor is spread about a particular issue and then you respond with a messaging without getting to hear from the people who are spreading the rumors, but I think the best ones always platforms that give people an opportunity to ask questions and get responses. -International staff, Sierra Leone

Interactive radio programs were perceived to be the most effective.

Radio dramas that were that were done at that time trying to remember the name of it. Mr. Plan-Plan or something like that, but there were you know some radio dramas and radio programming that were that were broadcast that would touch on addressing some of those some of those rumors. There were the journalist trainings. So to make sure that, you know, journalists were reporting more accurately. So there are a couple other things that that were being done at that time as well that, you know, would help to provide more accurate information, which would hopefully then reduce the number of rumors . . . And you know, maybe one thing to add is that you know, I think there was a real sense of wanting to be very careful about not repeating the rumor. So, you know by addressing the rumor you're really kind of just accelerating that accurate information rather than repeating the rumor in a way that's you're telling people that it's not true. So that's something that you know where people were really careful not to do. Because I didn't want to exacerbate that that rumor. -International staff, Sierra Leone and Liberia

Journalist training can help prevent rumors by increasing accuracy of media reports.

It was the kind of feeding up from the ground and then feeding back down. Kind of changing up the message guides like every couple of weeks or every month to ensure that they had. That they were kind of addressing the most current rumors. It is important to consider potential unintended effects of messaging.

If you just show up with you know and have, you know, PPEs you know with chlorine sprayers and start giving messages, if you're lucky, the best thing . . . is they won't listen to you. The other thing is . . . run you out of the village.

-International staff, Liberia

Community entry is a sensitive process that needs to be done carefully and in a culturally appropriate way.

Organizational Response to Rumors in the Ebola Virus Disease Outbreak www.ghspjournal.org If you go to a village that you know is infested, you call the Elder, like the Big Chief. Have him call his people because there were no mobile phones in order to make an appointment . . . Having called the elders, called the people, and tell them look we're coming back the next day because we're doing it from town to town. Like we spend a whole week out of town because you actually make appointments. Like, look we coming back tomorrow, Saturday at 4 o'clock. Let other people be here. We got a bundle of good news for you. -Local staff, Liberia

In some of the chiefdoms that we worked in, the paramount chiefs were very well respected, and they were sort of the purveyors of a lot of trust, and a lot of trying to you know, if you wanted access to the community you had to go through them. I know that there were some communities that we worked in or that we worked with that was the opposite, is that the Paramount Chief was not seen as being legitimate. It was somebody who was put in place because they were, you know, somebody's father or brother or connected somebody who is already powerful, and so the community actually didn't trust them. -International staff, Sierra Leone

Entry strategies vary, so it is important to gain an understanding of the community leadership structure before entering.

You have to go you have to drill down deeper and find out who in each community is already an accepted source of information and whether they are knowledgeable about the issue, whether it's Ebola or an earthquake or whatever . . . All communities have some natural leaders. Some of them are either formally appointed like the village Chief or Council of some kind or you have just individuals who are recognized by other members of the community as having skills of organizing people of putting together trainings or those kind of things. -International staff, Liberia

Communities may be more likely to accept information from a trusted source in the community rather than from an outsider.

And if there is any new information you share with people . . . so that it comes from the trusted source first. Otherwise, if you try to counter it, it's difficult. With misinformation, you just continue to provide accurate information. Probably through trusted channels in the community. So like religious leaders, traditional leaders, and mobilizers who are residing in the communities . . . You may be able to use those people to provide information to the populace regarding rumors. -Local staff, Sierra Leone Organizational change One of the things that was changed in the short time I was there, was the opening the possibility for the family to see the people inside the treatment center, with quite a distance but being able to communicate with them. And the other thing that was, I think, quite important was also . . . opening the possibility in the treatment center, to see the dead body from the distance.

-International staff, Liberia and Sierra Leone

Increasing family's ability to communicate and see their loved ones in the Ebola Treatment Center / Ebola Treatment Unit increased trust and reduced rumors.

I do think actually the changes that were made at the treatment centers to increase a patient's ability, family's ability to see the patients I think that that has a huge impact, on trust and rumors, I think in a lot of ways probably more so than communication. procedures in ETC/ETUs and health centers. Rumors that indicated mistrust of ETC/ETUs were seen by most EVD responders as being understandable given that in the early part of the outbreak family members or friends often went to the ETC/ETU and were never seen again. Increasing access to ETC/ETUs and modifying burial practices to allow safe family participation was perceived by participants to successfully reduce rumors, but more importantly, led to a more compassionate response that respected the humanity of those affected by the outbreak.

This study had several limitations. EVD responders working with large international nongovernmental organizations were overrepresented in the sample due to the use of snowball sampling and the author's own professional network. The lead author had worked in Sierra Leone during the 2014-2016 EVD outbreak, which resulted in higher levels of interconnection between participants who had worked in Sierra Leone and may have contributed to more similar opinions among problem. So a big part of it is you've actually got to just make the [infection prevention and control] better at the level of the health facility and then you can start to bring people in and say let's go for a tour of this facility. Why don't you participate in an evaluation of the infection prevention and control of the facility, use the rumor phone call line to let us know if people are not wearing gloves or if they're reusing needles, you know, ask people to sort of be agents like have some kind of they have to be agents of making it better I think a lot of the time. It's about working with health care workers to understand that [infection prevention and control] is not an optional but it's also a way of keeping them safe from sickness. I think you've really got to attack it on both supply and demand side if you want to get some measure of trust back into health system or in a Health service. -International staff, Sierra Leone I think we just got to make a better . . . response. Right? I think one way to beat those types of rumors . . . is to put up results that show that essentially protect people and doesn't exacerbate the issues that we're all talking about. So I feel like that's really the only way to do that, the only way to really truly build trust after trust has been broken through a number of different things where it's rumors that have truth but are damaging nonetheless would be to build trust through proving that we can do this right. Which is can we reduce health care worker infections and nosocomial infections? When persons that are believed to be sick interact with health care workers and response workers as well. -International staff, Sierra Leone those participants. No incentive was provided for participating in the study so it is likely that participants were motivated primarily in their own interest in the topic, which may have led to some sampling bias. To encourage openness and honesty in participant interviews, participants were not asked to provide the name of the organization(s) they worked with. However, as organization names are not included in the data, it is not possible to take into account duplication in the organizations represented. The data were also limited due to the long amount of time that had passed between the EVD outbreak and the time of interviews.

This study provides an overview of how rumors affected the 2014-2016 EVD response at an organizational level. While participants believed rumors had a clearly negative impact on health-seeking behavior, they were also instrumental in improving the EVD response. This study illustrates that rumors should be a key consideration in outbreak responses from the start of an outbreak response and should be considered in all aspects of the response, not only as an issue to be addressed via communication. Rumor management: mass media They tried to hold up people who were survivors and people who were Ebola champions to show what their contribution had been in terms of fighting Ebola and the fact that basically that they were to be trusted and they highlighted a lot of the Ebola response workers and the health care workers in that, which was really cool. -International staff, Sierra Leone

Ebola outbreak in West Africa epidemic curves

Communicating Risk in Public Health Emergencies: A WHO Guideline for Emergency Risk Communication (ERC) Policy and Practice. WHO

A theory-based socioecological model of communication and behavior for the containment of the Ebola epidemic in Liberia

Health Communication Capacity Collaborative

Social mobilization and community engagement central to the Ebola response in West Africa: lessons for future public health emergencies

Communications Challenges and Good Practices

Briefing: Ebola-myths, realities, and structural violence

Revisiting ebola communication

Rumor Psychology: Social and Organizational Approaches

Improvised News: A Sociological Study of Rumor. Bobbs-Merrill Company

Rumor as group problem solving: development patterns in informal computer-mediated groups

Combatting rumors about Ebola: SMS done right

Tracking rumors to contain disease: the case of DeySay in Liberia's Ebola outbreak

The COVID-19 pandemic: making sense of rumor and fear

Impact of rumors and misinformation on COVID-19 in social media

Community engagement during outbreak response: standards, approaches, and lessons from the 2014-2016 Ebola outbreak in Sierra Leone

Chambers R. Going to Scale With Community-Led Total Sanitation: Reflections on Experience, Issues and Ways Forward. Institute of Development Studies

Nosocomial transmission of Ebola virus disease on pediatric and maternity wards

Cite this article as: Brandt AJ, Katalenich B, Seal DW. Qualitative review of organizational responses to rumors in the 2014-2016 Ebola virus disease outbreak in Liberia and Sierra Leone

Acknowledgments: We would like to thank the Ebola Virus Disease responders for their service during the 2014-2016 outbreak and for volunteering their time to participate in this research. We would also like to thank Lina Moses, Dominique Meekers, and Janet Ruscher for their feedback and guidance on this research and Alisha Thompson and Ryan Blyth for their work transcribing interviews.

Author contributions: AJB conceived of the research, recruited study participants, collected interview data, led the data analysis, and drafted the manuscript. BK analyzed interview data and reviewed and revised the manuscript. DWS provided supervision and feedback on study design, data collection, and analysis and reviewed and revised the manuscript.Competing interests: None declared.